This is a reproduction of a library book that was digitized

by Google as part of an ongoing effort to preserve the

information in books and make it universally accessible.

https://books.google.com
 NYPL RESEARCH LIBRARIES

3 3433 07023788 2
4
De Morgan
‫وب‬
 FORMAL LOGIC :

OR,

The Calculus of Inference,

Neceſſary and Probable.

BY

AUGUSTUS DE MORGAN
Of Trinity College Cambridge,
Fellow of the Cambridge Philofophical Society, Secretary of the Royal
Aſtronomical Society, Profeſſor of Mathematics in
Univerſity College London.

Καλὸς ὁ νόμος ἐάν τις αὐτῷ νομίμως χρῆτας .

LIBRARY
LONDON :
TAYLOR AND WALTON,

Bookſellers and Publiſhers to Univerſity College,

28, Upper Gower Street.
M DCCCXLVII
S

3112
 PREFАСЕ .

THE ſyſtem given in this workextends beyond that

commonly received, in ſeveral directions. A brief
ſtatement of what is now ſubmitted for adoption into
the theory of inference will be the matter of this preface.
In the form of the propoſition, the copula is made as
abſtract as the terms : or is confidered as obeying only
thoſe conditions which are neceſſary to inference.
Every name is treated in connection with its contrary
or contradictory name ; the diſtinction between theſe words
not being made, and others ſupplied in conſequence.
Eight really ſeparable forms of predication are thus ob-
tained, between any two names : the eight of the common
ſyſtem amounting only to fix, when, as throughout my
work, the two forms of a convertible propofition are
confidered as identical.

The complex propoſition is introduced, conſiſting in

the coexistence of two fimple ones. The theory of the
fyllogifm of complex propoſitions is made to precede
that of the ſimple or ordinary fyllogifm ; which laſt is
deduced from it. I have only uſed the word complex,
becauſeſimple was already appropriated (fee page 85).
iv Preface.
By the introduction of contraries, the number of valid
ſyllogiſtic forms is increaſed to thirty- two, connected to-
gether by many rules of relation, but all ſhewn to contain,
each with reference to its own diſpoſition of names and
contraries, only one form of inference.
The distinction of figure is avoided from the beginning
by introducing into every propoſition an order of refer-
ence to its terms .
A fimple notation, which includes the common one,
gives the means of repreſenting every fyllogifm by three
letters, each accented above or below. By inſpection of
one of theſe ſymbols it is ſeen immediately, 1. What
fyllogifm is repreſented, 2. Whether it be valid or in-
valid, 3. How it is at once to be written down, 4. What
axiom the inference contains, or what is the act of the
mind when it makes that inference (chapter XIV).
A fubordinate notation is uſed (page 60) in abbrevia-
tion of the propoſition at length.
Compound names are confidered, both when the com-
poſition is conjunctive, and when it is disjunctive. Diſtinct
notation and rules of transformation are given, and the
compound fyllogifms are treated as reducible to ordinary
ones, by invention of compound names.
The theory of the numerical ſyllogiſm is inveſtigated,
in which, upon the hypothefis of numerical quantity in
both terms of every propoſition, a numerical inference
is made.
But, when the numerical relations of the ſeveral terms
are fully known, all that is unusual in the quantity of the
predicate is ſhown to be either ſuperfluous, or elſe, as I
have called it, ſpurious.
 V
Preface.
The old doctrine of modals is made to give place to
the numerical theory of probability. Many will object
to this theory as extralogical. But I cannot fee on what
definition, founded on real diſtinction, the exclufion of it
can be maintained. When I am told that logic confiders
the validity of the inference, independently of the truth
or falſehood of the matter, or ſupplies the conditions
under which the hypothetical truth of the matter of the
premiſes gives hypothetical truth to the matter of the
conclufion, I ſee a real definition, which propounds for
confideration the forms and laws of inferential thought.
But when it is further added that the only hypothetical
truth ſhall be abſolute truth, certain knowledge, I begin
to fee arbitrary distinction, wanting the reality of that
which preceded. Without pretending that logic can take
cognizance of the probability of any given matter, I
cannot underſtand why the ſtudy of the effect which
partial belief of the premiſes produces with reſpect to the
conclufion, ſhould be ſeparated from that of the conſe-
quences of ſuppoſing the former to be abſolutely true.
Not however to diſpute upon names, I mean that I
ſhould maintain, against those who would exclude the
theory of probability from logic, that, call it by what
name they like, it ſhould accompany logic as a ſtudy.
I have, of courſe, been obliged to expreſs, in my own
manner, my own convictions on points of mental philo-
ſophy. But any one will fee that, in all which I have
propoſed for adoption, it matters nothing whether my
views of the phenomena of thought, or others, be made
the basis of the explanation. So far therefore, as I am
vi Preface.
confidered as propofing forms of fyllogifm, &c. to the
logician, and not giving inſtruction to the ſtudent of the
ſcience, the reader has nothing to do with my choice of
the terms in which mental operations are ſpoken of.
In the appendix will be found ſome remarks on the
perſonal controverſy between Sir W. Hamilton of Edin-
burgh and myſelf, ofwhich I ſuppoſe the celebrity of my
opponent, and the appearance of part of it in a journal
ſo widely circulated as the Atheneum, has cauſed many
ſtudents of logic to hear or read ſomething.
At the end of the contents of ſome chapters in the
following table, are a few additions and corrections, to
which I requeſt the reader's attention.

A. DE MORGAN .

University College, London,

October 14, 1847 .
 TABLE OF CONTENTS .

* The articles entered in Italic, are thoſe, the contents of which belong
to the peculiar ſyſtem preſented in this work.

CHAPTER I. First Notions (pages 1-25).

Firſt notion of Logic, I ; Reduction of propoſitions to fimple
affirmation and negation, 2, 3 ; Distinction between negation and
affirmation requiring a negative, 3 ; how two negatives make an
affirmative, 4 ; propoſitions, 4; their relations, contraries and contra-
dictories, 5 ; Quantity of fubject and predicate, 6 ; Converſes, 7 ;
fundamental notion of inference, 8 ; Material repreſentation, 8, 9 ;
fyllogifm, 9 ; its elements, 9 ; fyllogifms ofdifferent kinds of conclu-
fion, 10, 11, 12 ; collection of reſults, 12 ; rules of fyllogifm, 13 , 14 ;
weakened concluſions andstrengthenedpremises, 15 ; the figures, 16,
17, 18 ; collection ofeffentially differentfyllogifms, 18, 19 ; examples,
19, 20 ; à fortiori ſyllogiſm, 20, 21 , 22 ; hypothetical fyllogifm, 22,
23 ; demonſtration, direct and indirect, 23, 24 ; conversion of a di-
lemma, 25 .

* This chapter may be omitted by those who have fome know-

ledge of the ordinary definitions and phraseology of logic. It is strictly
confined to the Ariftotelian forms and fyllogifms, and is the reprint of
a tract publiſhed in 1839, under the title of ' First Notions of Logic
(preparatory to the ſtudy of Geometry)' : the only alterations are ;-
the change of phraseology, as altering ' fome X is Y' into ' fome Xs
are Ys, ' &c.; the correction of a faulty demonſtration ; and a few
omiffions, particularly of ſome inſufficient remarks on the probability
ofarguments.

CHAPTER II .- On Objects, Ideas, and Names (pages 26-46).

Definition of Logic, 26 ; our poſition with reſpect to mind, 26,
27 ; Doubt on the uniformity ofprocess in all minds, 27 ; exiſtence of
things external to the perceiving mind, 28, 29 ; fubject and object,
ideal and objective, 29, 30 ; idea the fole knowledge, 30 ; object,
why then introduced, 30 ; extent ofits meaning, 30, 31 ; abſtraction,
qualities, relations, 31, 32 ; innate ideas, 32 ; distinction ofneceſſary,
T
 D

De Morgan
‫ور‬
 FORMAL LOGIC :

OR,

The Calculus of Inference,

Neceſſary and Probable.

BY

AUGUSTUS DE MORGAN
OfTrinity College Cambridge,
Fellow of the Cambridge Philoſophical Society, Secretary of the Royal
Astronomical Society, Profeſſor of Mathematics in
Univerſity College London.

Καλὸς ὁ νόμος ἐάν τις αὐτῷ νομίμως χρῆτας

3

LONDON :
TAYLOR AND WALTON,
Bookſellers and Publiſhers to Univerſity College,
28, Upper Gower Street.
MDCCXLVII
3112
 PREFACE .

HE ſyſtem given in this work extends beyond that

THE ſyſtem given
commonly received, in ſeveral directions. A brief
ſtatement of what is now ſubmitted for adoption into
the theory of inference will be the matter of this preface.
In the form of the propoſition, the copula is made as
abſtract as the terms : or is confidered as obeying only
thoſe conditions which are neceſſary to inference.
Every name is treated in connection with its contrary
or contradictory name ; the diſtinction between theſe words
not being made, and others ſupplied in conſequence.
Eight really ſeparable forms of predication are thus ob-
tained, between any two names : the eight of the common
ſyſtem amounting only to fix, when, as throughout my
work, the two forms of a convertible propofition are
confidered as identical.
The complex propoſition is introduced, confifting in
the coexistence of two fimple ones. The theory of the
fyllogifm of complex propoſitions is made to precede
that of the ſimple or ordinary fyllogifm ; which laſt is
deduced from it. I have only uſed the word complex,
becauſeſimple was already appropriated (ſee page 85).
iv Preface.
By the introduction of contraries, the number of valid
ſyllogiſtic forms is increaſed to thirty-two, connected to-
gether by many rules of relation, but all ſhewn to contain,
each with reference to its own diſpoſition of names and
contraries, only one form of inference.
The distinction of figure is avoided from the beginning
by introducing into every propoſition an order of refer-
ence to its terms.
A fimple notation, which includes the common one,
gives the means of repreſenting every ſyllogiſm by three
letters, each accented above or below. By inſpection of
one of theſe ſymbols it is ſeen immediately, 1. What
ſyllogiſm is repreſented, 2. Whether it be valid or in-
valid, 3. How it is at once to be written down, 4. What
axiom the inference contains, or what is the act of the
mind when it makes that inference (chapter XIV).
A fubordinate notation is uſed (page 60) in abbrevia-
tion of the propofition at length.
Compound names are confidered, both when the com-
poſition is conjunctive, and when it is disjunctive. Diſtinct
notation and rules of transformation are given, and the
compound fyllogifms are treated as reducible to ordinary
ones, by invention of compound names.
The theory of the numerical ſyllogiſm is inveſtigated,
in which, upon the hypothefis of numerical quantity in
both terms of every propoſition, a numerical inference
is made.
But, when the numerical relations of the ſeveral terms
are fully known, all that is unusual in the quantity of the
predicate is ſhown to be either ſuperfluous, or elſe, as I
have called it, ſpurious.
 Preface. V

The old doctrine of modals is made to give place to

the numerical theory of probability. Many will object
to this theory as extralogical. But I cannot fee on what
definition, founded on real diſtinction, the exclufion of it
can be maintained. When I am told that logic confiders
the validity of the inference, independently of the truth
or falſehood of the matter, or ſupplies the conditions
under which the hypothetical truth of the matter of the
premiſes gives hypothetical truth to the matter of the
conclufion, I ſee a real definition, which propounds for
confideration the forms and laws of inferential thought.
But when it is further added that the only hypothetical
truth ſhall be abſolute truth, certain knowledge, I begin
to fee arbitrary distinction, wanting the reality of that
which preceded. Without pretending that logic can take
cognizance of the probability of any given matter, I
cannot underſtand why the ſtudy of the effect which
partial belief of the premiſes produces with reſpect to the
conclufion, ſhould be ſeparated from that of the confe-
quences of ſuppoſing the former to be abſolutely true.
Not however to diſpute upon names, I mean that I
ſhould maintain, against those who would exclude the
theory of probability from logic, that, call it by what
name they like, it ſhould accompany logic as a ſtudy.
I have, of courſe, been obliged to expreſs, in my own
manner, my own convictions on points of mental philo-
ſophy. But any one will fee that, in all which I have
propoſed for adoption, it matters nothing whether my
views of the phenomena of thought, or others, be made
the basis of the explanation. So far therefore, as I am
vi Preface.
confidered as propoſing forms of fyllogifm, &c. to the
logician, and not giving inſtruction to the ſtudent of the
ſcience, the reader has nothing to do with my choice of
the terms in which mental operations are ſpoken of.
In the appendix will be found ſome remarks on the
perſonal controverſy between Sir W. Hamilton of Edin-
burgh and myself, ofwhich I ſuppoſe the celebrity of my
opponent, and the appearance of part of it in a journal
ſo widely circulated as the Atheneum, has cauſed many
ſtudents of logic to hear or read ſomething.
At the end of the contents of ſome chapters in the
following table, are a few additions and corrections, to
which I requeſt the reader's attention.

A. DE MORGAN .

Univerſity College, London,

October 14, 1847 .
 TABLE OF CONTENTS .

* The articles entered in Italic, are thoſe, the contents of which belong
to the peculiar ſyſtem preſented in this work.

CHAPTER I. First Notions (pages 1-25).

Firſt notion of Logic, I ; Reduction of propofitions to fimple
affirmation and negation, 2, 3 ; Distinction between negation and
affirmation requiring a negative, 3 ; how two negatives make an
affirmative, 4 ; propoſitions, 4 ; their relations, contraries and contra-
dictories, 5 ; Quantity of fubject and predicate, 6 ; Converſes, 7 ;
fundamental notion of inference, 8 ; Material repreſentation, 8, 9 ;
fyllogifm, 9 ; its elements, 9 ; fyllogifms of different kinds of conclu-
fion, 10, 11, 12 ; collection of reſults, 12 ; rules of fyllogifm, 13 , 14 ;
weakened concluſions andstrengthenedpremises, 15 ; the figures, 16,
17, 18 ; collection ofeffentially differentfyllogifms, 18, 19 ; examples ,
19, 20 ; à fortiori fyllogifm, 20, 21 , 22 ; hypothetical fyllogifm, 22 ,
23 ; demonſtration, direct and indirect, 23, 24 ; converfion of a di-
lemma, 25.
* This chapter may be omitted by those who have fome know-
ledge of the ordinary definitions and phraseology oflogic. It is strictly
confined to the Ariftotelian forms and fyllogifms, and is the reprint of
a tract publiſhed in 1839, under the title of ' First Notions of Logic
(preparatory to the ſtudy of Geometry)' : the only alterations are ;-
the change of phraseology, as altering ' fome X is Y' into ' fome Xs
are Ys, &c.; the correction of a faulty demonſtration ; and a few
omiffions, particularly of ſome infufficient remarks on the probability
of arguments .

CHAPTER II .- On Objects, Ideas, and Names (pages 26-46).

Definition of Logic, 26 ; our poſition with reſpect to mind, 26,
27 ; Doubt on the uniformity ofprocess in all minds, 27 ; exiſtence of
things external to the perceiving mind, 28, 29 ; ſubject and object,
ideal and objective, 29, 30 ; idea the fole knowledge, 30 ; object,
why then introduced, 30 ; extent of its meaning, 30, 31 ; abſtraction,
qualities, relations, 31 , 32 ; innate ideas, 32 ; diſtinction ofneceſſary,
viii Contents.
and not neceſſary, 33, 34 ; names, 34 ; aſſumption of their correct
ufe a poftulate, 35 ; frequent vagueneſs of names, 35 ; the tendency
offcience to correct it, 35 ; definition, nominal and real, 36; the latter
purely objective, 36 ; reference of every name to every idea or object,
either as direct or contrary (i. e. contradictory), 37 ; the universe ofa
propofition, limitation of the term universe, 37, 38 ; Notationfor con-
traries, 38 ; remarks on the manner in which language furnishes con-
traries, 38, 39 ; conversion ofparticular into universal by invention of
Species, 39 ; the distinction ofA, E, I, O, not more than an accident of
language in any particular cafe, 40 ; the introduction of a limited uni-
verse gives positive meaning to contraries originally defined by negation,
40, 41 ; inference 41 , 42 ; qualities, how uſed in the forms of Logic,
42 ; formal Logic deals with names only, 42, 43 ; concluſion, ideal
and objective, remarks on the distinction of, 43, 44; Aſſertions
fometimes made on the ſtudy of neceſſary conſequences, 44, 45 ; vir-
tual incluſion of the neceſſary conſequence in the premiſes, remark
on, 45 ; Humble pofition of the logic treated in this work, 46.

CHAPTER III. On the abstract Form of the Propofition

(pages 46-54) .
Separation of logic from metaphyfics, 46, 47 ; particularly necef-
ſary as to the import of the propofition, 47 ; Ufual mode of repre-
ſenting abſtract terms, 47 ; the term may be nominal, ideal, or objec-
tive, 47 ; objection to quantitative expreſſions, as distinguished from
quantuplicitative, 48 ; objection to the notion of cumulation as an ade-
quate representation of combination, 48 , 49 ; Various meanings of the
copula is, 49 ; Abstraction of the logical characters of the word by
right of which all those meanings are properfor allinference, 50, 51 ;
meanings which only fatisfy some characteristics may be adapted to
Some inferences, 51 , 52 ; poſſibility of new meanings, 52 ; inadmif-
fibility of fome exifting meanings, 52, 53 ; some cases in which the
meanings may beshifted, 53, 54.

CHAPTER IV.- On Propofitions (pages 54-76) .

Formal uſe of names, 54 ; propofition defined, 54 ; Limited uni-
verse introduced, 55 ; Expreſſed ſtipulation that no name used fills
this universe, 55 ; distinction offimple and complex propofition, 56 ;
ſign, affirmative and negative, 56 ; relative quantity, univerſal and
particular, 56 ; Only relative quantity or ratio, definite in univerfals,
57; fubject and predicate, 57 ; predicate always quantified by poſition,
57 ; Distinction to be taken as to this quantification, 57 ; definite and
indefinite, ideal poſſibility ofperfect definiteness throughout, 58 ; order,
58 ; convertibles and inconvertibles, 58, 59 ; remark on the alter-
natives of logic, 59 ; uſual distinction of contrary and contradictory,
not made in this work, 60 ; fubcontrary andfupercontrary propofitions,
60 ; standard order of reference, which, as to claſſification, renders
 Contents. ix

figure unnecessary, 60 ; A, E, I, O, and their contranominals, 60 ;

theſe and their contranominals denoted by thesub-symbols and fuper-
Symbols A , E , I , O , A', Ε', I', O', 60 ; Meanings of X)Y, X.Y,
XY, and X : Y, 60 ; The eight standard forms ; reduction of all others
to them ; and representation by instances, 61 ; new term, contranominal,
and expreſſion by means of it, 62 ; meaning of the new forms ofaffer-
tion, E' and I', 62 ; representations of the eight forms, 62 ; Quanti-
ties of the direct and contrary terms, 63 ; Table ofrelations ofinclufion,
&c. , 63 ; Concomitants, 63 ; Reduction of the forms to one another,
by the orders of reference, XY, Xy, xy, xY, 63, 64 ; Investigation
of equivalences obtained by change of one or more of the four, fub-
ject, predicate, copula, and order, 64, 65 ; ſtrengthened and weak-
ened forms, 65 ; complex propofitions, 65 ; P, the complex particular,
66 ; D, the identical, 66 ; D₁, the fubidentical, D', thefuperidentical,
C, the contrary, C₁, the fubcontrary, C', the fupercontrary, 67 ; fub
and fuper affirmation and negation, 68 ; Table of relations between
thefimple and complex, 69 ; Table of connexion offimple and complex
propofitions by change of terms and orders, 70 ; Laws of this table,
70 ; Continuous interchange of complex relation, 70, 71 , 72 ; its
laws, 72 ; neceſſary, fufficient, actually poſſible, contingent, and their
contraries ; laws of connexion of these relations with thefimple and
complex forms, 72, 73 , 74 ; nomenclature in conjunction with, or
amendment of, that offubaffirmative, &c., 75 ; statement of the evi-
dent laws to which all syllogism might be reduced, 75 , 76.
Additions and corrections. Page 56, line 7, infert except only
one which confifts of four fimple propofitions. Page 62, line 23 ;
Say X and Y are not complements (instead of contraries) that is,
do not together either fill, or more than fill, the universe. Page
72, lines 4 and 3, from the bottom ; The oppofitions are incorrect.
It ought to be cannot do without and cannotfail with : must precede,
and must follow. The reader may easily identify the eight forms of
predication as having X for ſubject, Y for predicate, with the copulæ,
cannot be without, can be without, cannot be with, can be with,
cannot fail without, can fail without, cannot fail with, can fail with.

CHAPTER V.-On the Syllogism (pages 76-106) .

Definition of fyllogifm, premiſes, middle term, concluding terms, 76 ;
Distinction offimple and complex fyllogifm, 76 ; Reasons for beginning
with the latter, 76, 77 ; The common àfortiori fyllogifm is com-
plex, 76 ; Distinction offundamental and strengthenedsyllogism, 77 ;
Standard order of reference, the substituteforfigure, 77 ; Theforms
of the complex affirmatory and negatory syllogism, in symbols and in
language, 78 ; its limitingforms, 79 ; its rules, 79 ; the demonstration
of the affirmatory forms, by help of a diagram, 79, 80 ; their à for-
tiori character, 81 ; the demonstration of the negatory forms, 81 , 82 ;
reduction of all theforms of each kind to any one, and rules, 82, 83 ,
84; Complex forms in which P enters, 84, 85 ; doubt on the goodness
X Contents.

ofthe terms simple and complex, 85 ; Denial of thefimplicity of the

simple propofition, 85, 86 ; Are not disjunctive and conjunctive the
proper words ? 86 ; The denial ofa conclufion, coupled with one of
the premiſes, denies the other, 86 ; The fimple ſyllogiſm, 86 ; De-
monftration that a particular cannot lead to a univerſal, and that two
particulars are inconclufive, by help of the complexfyllogifm, 86, 87 ;
Opponent fyllogifms, 87, 88 ; Rules for the symbols of opponent
fyllogifms, 87, 88 ; Of fundamental fyllogiſms, there must be
twice as many particular as univerfal, 88 ; Deduction of the fun-
damental fimple syllogisms, eight universal, and fixteen particular,
from the eight affirmatory complex fyllogifms, 88, 89 ; Deduction of
the eightstrengthenedfyllogifms from the limitingforms ofthe affirma-
tory complex ones, 90, 91 ; Connexion ofthetwo modes ofstrengthening a
premise, 90, 91 ; The conclufion is neverstrengthened bystrengthening
the middle term, nor only weakened by weakening it, 91; Table of
connexion of thestrengthenedfyllogifms with the rest, 91 ; deduction
of the strengthened fyllogifms from the negatory complex ones, and
dismissal ofthe latter as of no more logical effect than theformer, 92 ;
Direct rule of notation, applying tofyllogifms which begin and conclude
with like quantity, 92 ; Inverse rule of notation, [N.B. the word
inverſe ſhould have been contrary] applying to fyllogifms which begin
and conclude with unlike quantity, 93 ; Rulesfor allthe retainedfyllo-
gifms, 93 ; Sub-rules for the particularfyllogifms [they would have
done as remarks, but are needleſs as rules] 94 ; Remarks, partly reca-
pitulatory, 94, 95, 96 ; In all fundamental fyllogifms, the middle
term is univerſal in one premiſe, and particular in the other, 95 ; dif-
tinction thence arifing, 95 ; rule for connecting the syllogisms which
are formed by interchanging the concluding terms, 96 ; conversion ofa
particular into a univerfal, 96 ; distinction of the particular quantity
in a conclufion into intrinsic and extrinfic, 97 ; the quantity of one
term always intrinsic, and hence thesyllogism can always be made uni-
verfal, 97 ; Nominal mode of notation for, and representation of, a
fundamental fyllogifm, 98 ; connexion of the nominalsystem with the
former (or proponent) ſyſtem, 99 ; mode of deriving concomitants and
weakened forms , 100 ; more abstract mode of representation derived
from the nominal, 100 ; nominal system of strengthened fyllogifms,
IOI ; mixed complex fyllogifm, 101 ; opponent forms, 102 ; verbal
description of the simple syllogism, 103 ; new view ofthesyllogism, in
which all is referred to the middle term, 104 ; rules thence derived,
105 ; compound names, and expulsion of quantity by reference of the
propofition to poſſibility or impoſſibility of a compound name, 105 ;
system offyllogifſm thence arifing, 106.
Additions and corrections. Page 79, in the firſt diagram, for
DDD, read DDD ; page 88 , line 23 , instead of has the other two
for its opponents, read has its opponents in the ſet ; page 90, line 4,
from the bottom, for premiſs read premiſe : the firſt ſpelling has been
common enough, but it seems ſtrange that the cognate words promife,
furmife, demife, &c. ſhould not have dictated the ſecond. Page 96 ;
 Contents . xi

The inverted forms of the ſtrengthened fyllogifms are omitted : of

thefe, four are their own inverſions , namely, AA'I', A'A I,, E'ΕΙ ,
and EEI' : of the remainder, AE'O' and E'A'O, are inverſions ;
and alſo A'EO and EAO'. Page 100, line 12, from the bottom ;
for -011 read - 011 ), the firſt time it occurs . Page 101 : Read
the ſymbols of the ſtrengthened fyllogifms ſo as to begin from the
middle in both premiſes : thus, Xyzl is y)x+y)z= Xz . Page 101 .
I might have faid a word or two on the cafe in which a complex
particular is combined with a univerſal ; to form the refults will be an
eafy exerciſe for the reader. Page 102 , line 7, from the bottom, for
IA'I read IAI .

CHAPTER VI. On the Syllogism (pages 107-126).

Remarks connected with the existence of the terms, 107, 108, 109,
110, 111 , 112, 113. The conclufion notfeparable from the premises
except as to truth, 107, 108 ; conditions, and conditional fyllogifm,
109 ; incompleteness of reduction of conditional to categorical, 109,
110; univerſe of propofitions, 110 ; existence of the terms of a pro-
pofition, III ; its afſumption infyllogifm, particularly as to the middle
term, 112, 113 ; poftulate more extensive than the dictum de omni et
nullo, involved as well in the formation of premises as in fyllogifm,
114, 115 ; Invention of names, 115 ; notation for conjunctive and
disjunctive names, 115, 116 ; expreſſion ofcomplex relations and their
contraries, 116 ; copulative and disjunctive fyllogifms and dilemma,
117 ; Conjunctive poftulate, 117 ; deduction of other evident propofi-
tions from it, 118, 119 ; The collective and, as conjunctive, opposed
to the disjunctives and and or distributively used in univerfals, and or
disjunctive (in the common sense) in particulars, 119 ; Disjunctives
may be rejectedfrom univerfals, and conjunctives from particulars, 119 ;
Tranſpoſition, introduction of, and rules for, 120 ; Table of the tranf-
posedforms of A and E with compound names, 121 ; Examples ofdif-
junctive ſyllogiſms, dilemmas, &c. treated by the above method, 122 ,
123 , 124; Sorites, 124 ; Extended rules for the formation of the
various claffes of Sorites, 125, 126.
Additions and corrections. Page 121 , line 8 , from the bottom.
For [x,y][p,q])u read [X,Y][p,q])u .

CHAPTER VII. On the Aristotelian Syllogism ( 127-141).

Limitations impoſed either by Ariftotle or his followers, 127 ;
Dictum de omni et nullo, 127 ; defect of this, 128 ; excluſion of
contraries, 128 ; Standard forms, 129 ; Major and minor terms, and
diſtinction of figure, 129 ; Selection of the Ariftotelian ſyllogiſmsfrom
among those ofthis work, 130131 ; Symbolic words, and meaning of
their letters , 131 ; Reduction to the firſt figure, 131 , 132 ; Old form of
the fourth figure, 132 , 133 ; Suggestion as to two figures fubdivided,
133 ; Poſſible use of the distinction of figure, 133, 134 ; Collection of
xii Contents.
the figures in detail, 134, 135, 136 ; Aldrich's verſes on the rules,
136 ; Explanation of theſe rules, andſubſtitutes for theſyſtem in which
contraries are allowed, 137, 138, 139 ; Method of determining what
terms are taken directfrom the premises, and what contrariwise. 140 ;
Reasonfor the duplication oftheſyſtem ofchapter V., 140, 141 .

CHAPTER VIII. — On the numerically definite Syllogism (pages

141-170).
Reasonfor its introduction, 141 ; definition ofnumericaldefiniteness,
141, 142 ; distinction between it and perfect definiteness, 142, 143 ;
Notationfor thefimple numericalpropofition, 144 ; Forms of inference
when only the direct middle term is numerically definite, 145, 146 ;
Canon of the middle term, 145 ; Double inference in the case of one
premise negative, 145, 146 ; This double inference is true in the Arif-
totelian fyllogifm Bokardo, 145 ; Application of the phraseology of
complex names to the relations of propofitions, 146, 147, 148, 149 ;
Identicalpropofitions, 146, 147 ; Neceffary consequence, 147 ; Reasons
for rejecting the usual distinction of Contrary and Contradictory, and
for introducingfubcontrary andfupercontrary, 148 ; Remarks on a uni-
verse ofpropofitions, 149 ; Abolition of the numerical quantification of
the predicate, 150, 151 ; The cafes in which it appears either identical
with those in which it does not appear, or spurious, 150, 151 ; numeri-
calforms of the usual propofitions, 151 ; Modes of contradicting the
numericalforms, 152 ; Definition ofSpurious propositions : reasonsfor
refusing their introduction, and excluding them when they appear, 153 ,
154 ; Note in defence ofthe wordSpurious, 153 ; Spurious conclufions
may result from premises not spurious, 153 , 154 ; Law ofinference,
154 ; Contranominalforms of numerical propofitions, partial, (which
are spurious) and complete, 155, 156 ; When one is impoſſible, the
other is fpurious, 157 ; Fundamentalform of inference, 157 ; Of two
contranominals, one is always partially fpurious, 158 ; deduction of the
remaining forms from the fundamental one, 158, 159 ; Equations of
connexion between the numerical quantities , 159 ; Enumeration of the
usefulfubdivisions of the numerical hypothefis, 160 ; Exhibition of the
fixteen varieties of numericalfyllogiftic inference, 161 ; Deduction of
all the ordinary syllogisms from them, 161 , 162 ; Cafes in which de-
finite particulars allow ofinference by description with respect to the
middle term, 163 ; Double choice in the mode ofexpreſſing thesefyllo-
gifms, 163 ; Exceptional syllogisms, afferted to be what are most
frequently meant when univerfals are used, 164 ; Formation of ab
infirmiorifyllogifms, their connexion with the ordinary ones, 165 ; For-
mation offyllogifms oftranſpoſed quantity, 166 ; Enumeration ofthem,
166, 167 ; Rules for their formation, 167, 168 ; Example of their
occurrence, 168 ; Example of the formation of an opponent numerical
fyllogifm, 168 ; Remark on what becomes of the second inference in a
partially definite ſyſtem, 169 ; Nonexistence ofdefinite numerical com-
plexfyllogifms, 169, 170.
 Contents. xiii

Additions and corrections. Page 143 , line 12 : Supply the propo-

ſitions X)M,P and Y)N,Q, as deducible from the numbers of in-
ſtances in the ſeveral names. Page 148, line 10, from the bottom :
for propofitions read prepofitions. Page 152, line 4 : for m read
m. Page 153 , line 22 : for will preſently ſhow us, read have ſhown
us in page 145. Page 154, line 2, from the bottom, for ys read zs.
Page 155 , line 6 from the bottom, for mXY read mXY . Page 162 ,
line 2, after the table : for laſt chapter read chapter V. Page 166,
line 17, for m'xy read m'xy. Page 167, line 24 : for 62 read 92 .

CHAPTER IX.-- On Probability (pages 170-191 ) .

Remark on old and new views of knowledge, 170 ; Neceſſary
truths not always identities, inſtance, two and two are four, 171 ;
degrees of belief or knowledge, 171 ; Degree of knowledge treated as
a magnitude, 172 ; Distinction of ideal and objective probability,
172, 173 ; Rejection of the latter, 173 ; Definition of probability as
referring to degree of belief, 173 ; Illuſtration ofdegree of belief as a
magnitude, 174 ; What is perception of magnitude, 174; Meafure-
ment of magnitude, 175 ; Illuſtration of various degrees of belief,
176 ; Difference of certain and probable, not that of magnitudes
of various kinds, but that of finite and infinite of the ſame kind,
176, 177 ; the real diſtinction not thereby abrogated, 178 ; Poftulate
on the acceptance of which the theory of probabilities depends, 179 ;
the affumption of this poftulate, in other cafes, not always ſo well
founded as is ſuppoſed, 179, 180, 181 ; the difficulties of this poſtu-
late intentionally introduced and inſiſted on, 181 , 182 ; Meaſure of
probability or credibility, and alfo of authority, 182, 183 ; Rule for
the formation of this meaſure, 184 ; Objective verification of a re-
mote conclufion of this rule, 184, 185 ; Probability of the joint hap-
pening of independent events, 186 ; Conſequences ofthis rule, 187 ;
Problem in which the primary cafes are unequally probable, 187,
188 ; Rule of inverſe probabilities, 188, 189, 190 ; this rule alſo
holds in calculating the probabilities of restricted cafes from the unre-
ſtricted ones, 190, 191 .

CHAPTER X.- On probable Inference (pages 191-210).

Argument and testimony, 191, 192 ; argument never the only vehi-
cle ofinformation except when demonstrative, 192 ; truth orfalsehood
not thefimple iſſue in argument, 192 , 193 ; difficulty thereby introduced
into thejudgment of truth or falsehood, 193 ; entrance of testimony,
194; remark on the precept to neglect authority, 194 ; Compofition
ofindependent teftimonies, 195 ; on the majority of witneſſes, 196 ;
the fame problem, when the event afferted has an antecedent proba-
bility, 197 ; queſtion of colluſion, 198, 199 ; extenſion of the laſt
problem to more complicated events, 2003; Compoſition of indepen
xiv Contents.
dent arguments on the fame fide, 201 ; manner in which the weak-
ness of an argument may become an argument or a testimony , 202 , 208 ;
Composition of arguments on contrary fides, 203 ; the fame on fubcon-
trary fides, 204 ; Composition of argument and testimony in a question
ofcontraryfides, 205 ; More weight due to argument than to testimony
of the same probability, 206 ; Utter rejection of authority, what it
amounts to, 207, 208 ; Effects of the fame arguments on different
minds, 209 ;Effect of probable conſequence upon an affertion, 209, 210 ;
Oldfuicidal affertion , explained by probability, 210.
Additions and corrections . Page 199, line 4, from the bottom :
for ( 1-1) read ( 1-1) . Page 201 , line 14, from the bottom :
for read

CHAPTER XI.-On Induction (pages 211-226).

Explanation of induction, 211 ; Reduction of the proceſs to a fyl-
logifm, 211 ; Induction by connexion, and inſtance, 212 ; Ordinary
induction not a demonſtrative proceſs, 212, 213 ; Pure induction,
incomplete, probability of it, 213 , 214 ; Ordinary mistakes on this
fubject, 215 ; Examination of Mr. T. B. Macaulay's enumeration of
inſtances in which ſcientific analyſis is uſeleſs, 216, 217, 218, 219,
220 , 221 , 222 , 223, 224 ; probability offyllogifms with particular
premises, 224, 225 , 226 ; Circumftantial evidence, 226.

CHAPTER XII.- On old logical Terms (pages 227-237) .

Dialectics, 227 ; fimple and complex terms, 227 ; apprehenfion,
judgment, diſcourſe, 227 ; Univerſal and fingular, 228 ; Individuals,
228 ; categories, predicaments, 228 ; ſubſtance, 228 ; firſt and fecond
ſubſtance, 229 ; quantity, continuous and difcrete, 229 ; Quality,
habit, difpofition, paffion, 229 ; Relation, 229 ; Action, paffion, imma-
nent, tranfient, univocal, equivocal, 229, 230 ; Remaining categories,
230 ; predicables, genus, ſpecies, 230 ; difference, property, acci-
dent, 231 ; cauſe, material, formal, efficient, final, 231 ; form, mo-
tion, ſubject, object, 231 ; Subjective, objective, adjunct, 232 ;
modals, ſubſtitution of the theory of probabilities for them, 232 ;
Their use in the old philofophy, 232, 233 ; Notions of old logicians
on quantity, 234 ; Intenſion or comprehenfion, and extenfion, ob-
jections to their oppoſition as quantities, and references to places in
this work where the diftinction has occurred, 234, 235, 236 ; In-
ſtance, 236 ; Enthymeme, Aristotle's, and modern, 236, 237.
Additions and corrections. Page 230, lines 16 and 15, from the
bottom; tranſpoſe the words former and latter. Page 234 line 2 from
bottom, for after read before. Page 237, note ; I find that etymolo-
gifts are decidedly of opinion that ῥῆσις, ſpeech, and ῾ρέω, flow, have
different roots, and that the former isspeech in its primitive meaning.
The reader muſt make the alteration, which however does not affect
my ſuggeſtion.
 Contents . XV

CHAPTER XIII.- On Fallacies (pages 237-286).

No claſſification offallacies, 237 ; Amusement derived from, 238 ;
fallacy, fophifm, paradox, paralogiſm, 238 ; Ariftotle's claſſification,
240 ; Poſition of ancients and moderns as to fallacies, 240 ;
Conſequences of the neglect of logic, 241 ; Aristotle's ſpecies of
fallacies enumerated, 241 ; Equivocation, 241 , 242 ; Change of
meanings with time, 243 ; Importance once attached to fucceſsful
equivocation, 244 ; Government fallacies, 244 ; Qualifications of
meaning, 244, 245 ; Phrases interpreted by their component words,
245 ; Aſſumption of right over words, 246, 247 ; Equivocating forms
of predication, 247 ; Amphibology, 247 ; Defects in the ſtructure
of language, 247 ; Compoſition and diviſion, 248 ; Accent, 248,
249; Fallacy of alteration of emphasis, 249, 250 ; diction, 250 ;
Accident and à dicto fecundum quid, &c., 250, 251 , 252 ; Examina-
tion of fome caſes of legal ſtrictneſs, 252, 253,254 ; Petitio prin-
cipii, 254 ; often wrongly imputed, 255 ; Aristotle's meaning of it,
256 ; Meaning of the old logicians, 256 ; derivation from the fyllo-
giſm of principle and example, 257 ; Charge of petitio principii
against all fyllogifms, 257, 258 , 259 ; Syllogifm ſometimes only re-
quired for diminution of comprehenfion, 259 ; Imperfect dilemma,
fophifm of Diodorus Cronus, 259, 260 ; Ignoratio elenchi, 260 ;
proof of negative, and negative proof, 261 , 262 ; afſſertions of difpu-
tants in their own favour, 262 ; Fallacy of tendencies and neceſſary
conſequences, 263 ; Fallacy of attributing refults of teftimony to ar-
gument, 264 ; Argumentum ad hominem, 265 ; Parallel cafes, 265,
266 : Fallacies ofilluſtration, 266, 267 ; Fallacia confequentis, 267 ;
Incorrect logical forms, 267, 268 ; Non caufa pro caufa, 268, 269 ;
Fallacia plurimum interrogationum, 269, 270 ; Practices ofbarrifters,
270 ; Incorrect uſe of univerſal form, 270, 271 ; Fallacy of the
extreme cafe, 271 ; Uſe of the extreme cafe, 271 , 272 ; Carriage of
principles, 272 ; Uſe ofthe word general, 272 ; Confufion of logic
and perſpective, 272, 273 ; General truths, 273 ; Implied univerſals
not fairly stated, 273 ; Fallacies of quantity, 274 ; Proverbs, 275 ;
Fallacies of probability , 275, 276 ; Fallacy of analogy, 276 ; Fallacy
of judging by reſults, 276, 277 ; Equivocations of ſtyle, 277 ; Fal-
lacy of ſynonymes, 277, 278 ; Fallacies arifing out ofconnection of
principles and rules, 279, 280, 281 ; Want of rule niſi in common
language, 280 ; Fallacy of importation of premiſes, 281 ; Fallacy of
retaining conclufions after abandoning premiſes, 282 ; Fallacies of
citation and quotation, 282, 283, 284, 285, 286.
Additions and corrections. Page 250, lines 3 and 5 ; for mil-
lenium read millennium, andfor Newtonion read Newtonian.

CHAPTER XIV . On the verbal Deſcription of the Syllogifm

(pages 286-296).
Conditions to be ſatisfied, 287 ; Double mode of deſcription and
xvi Contents.

reference of one to the other, 287 , 288 ; Language proposed, 288 ;

Description of the cases ofsyllogism in that language, 289, 290 ;
Connexion of the univerfal and concomitantfyllogifm with the complex
one, 291 , 292 ; Quantitativeformation ofthefyllogifm, 293, 294, 295 ;
Rules for theformation ofthe numericalfyllogifm, 295, 296.

APPENDIX I.-Account of a Controversy between the Author

of this Work and Sir William Hamilton of Edinburgh ; and
final reply to the latter ( pages 297-323) .

APPENDIX II. On fome Forms of Inference differing from

those of the Aristotelians (pages 323-336) .
 ELEMENTS OF LOGIC.

CHAPTER I.

First Notions.

HE firſt notion which a reader can form of Logic is by

Tviewing it as the examination of that part of reaſoning
which depends upon the manner in which inferences are formed,
and the inveſtigation of general maxims and rules for conſtruct-
ing arguments, ſo that the conclufion may contain no inaccuracy
which was not previously afſſerted in the premiſes. It has ſo far
nothing to do with the truth of the facts, opinions, or prefump-
tions, from which an inference is derived ; but ſimply takes care
that the inference ſhall certainly be true, if the premiſes be true.
Thus, when we say that all men will die, and that all men are
rational beings, and thence infer that ſome rational beings will
die, the logical truth of this ſentence is the ſame whether it be
true or falſe that men are mortal and rational. This logical truth
depends upon theſtructure ofthesentence, and not upon the par-
ticular matters ſpoken of. Thus,
Instead of Write,
All men will die. Every Y is X.
All men are rational beings. Every Y is Z.
Therefore ſome rational beings Therefore ſome Zs are Xs .
will die.

The ſecond oftheſe is the ſame propofition, logically confidered,

as the firſt ; the conſequence in both is virtually contained in,
and rightly inferred from, the premiſes. Whether the premiſes
be true or falſe, is not a queſtion of logic, but of morals, philoſo-
phy, hiſtory, or any other knowledge to which their ſubject-
B
2
First Notions of Logic.
matter belongs : the queſtion of logic is, does the conclufion
certainly follow if the premises be true ?
Every act of reaſoning muſt mainly confiſt in comparing to-
gether different things, and either finding out, or recalling from
previous knowledge, the points in which they reſemble or differ
from each other. That particular part of reaſoning which is
called inference, conſiſts in the compariſon of ſeveral and different
things with one and the fame other thing ; and ascertaining the
reſemblances, or differences, of the ſeveral things, by means of
the points in which they reſemble, or differ from, the thing with
which all are compared.
There muſt then be ſome propoſitions already obtained before
any inference can be drawn. All propoſitions are either aſſer-
tions or denials, and are thus divided into affirmative and negative.
Thus, X is Y, and X is not Y, are the two forms to which
all propoſitions may be reduced. These are, for our preſent
purpoſe, the moſt ſimple forms ; though it will frequently hap-
pen that much circumlocution is needed to reduce propofitions
to them. Thus, ſuppoſe the following aſſertion, ' If he ſhould
come to-morrow, he will probably ſtay till Monday ; ' how is
this to be reduced to the form X is Y ? There is evidently
ſomething ſpoken of, ſomething faid of it, and an affirmative
connection between them. Something, if it happen, that is, the
happening of fomething, makes the happening of another ſome-
thing probable ; or is one of the things which render the hap-
pening of the ſecond thing probable.
X is Y

The happening of his an event from which it may be

is inferred as probable that he
arrival to-morrow
will ſtay till Monday.
The forms of language will allow the manner of afſerting to
be varied in a great number of ways ; but the reduction to the
preceding form is always poſſible. Thus, ' ſo he ſaid ' is an affir-
mation, reducible as follows :
What you have juſt the thing which
is
faid (or whatever
elſe ' fo ' refers to) } 0{ he faid.
 First Notions of Logic. 3

By changing ' is ' into ' is not, we make a negative propoſi-

tion ; but care muſt always be taken to aſcertain whether a
propoſition which appears negative be really ſo. The principal
danger is that of confounding a propoſition which is negative
with another which is affirmative of ſomething requiring a nega-
tive to deſcribe it. Thus, he reſembles the man who was not
in the room,' is affirmative, and must not be confounded with
' he does not reſemble the man who was in the room.' Again,
' if he ſhould come to-morrow, it is probable he will not ſtay till
Monday,' does not mean the ſimple denial of the preceding pro-
poſition, but the affirmation of a directly oppoſite propoſition.
It is,
is Y

an event from which it may be

The happening of his is inferred to be improbable that
arrival to-morrow, he will ſtay till Monday :
whereas the following,
an event from which it may be
The happening of his isnot inferred as probable that he
arrival to-morrow,
} will ſtay till Monday,
would be expreſſed thus : ' If he ſhould come to-morrow, that is
no reason why he ſhould ſtay till Monday.'
Moreover, the negative words not, no, &c., have two kinds of
meaning which must be carefully diftinguiſhed. Sometimes they
deny, and nothing more : ſometimes they are uſed to affirm the
direct contrary. In caſes which offer but two alternatives, one
of which is neceſſary, theſe amount to the ſame thing, fince the
denial of one, and the affirmation of the other, are obviously
equivalent propoſitions. In many idioms of conversation, the
negative implies affirmation of the contrary in caſes which offer
not only alternatives, but degrees of alternatives. Thus, to the
queſtion, ' Is he tall ? ' the ſimple anſwer, ' No,' moſt frequently
means that he is the contrary of tall, or confiderably under the
average. But it must be remembered, that, in all logical reaſon-
ing, the negation is ſimply negation, and nothing more, never
implying affirmation of the contrary.
The common propoſition that two negatives make an affirm-
ative, is true only upon the ſuppoſition that there are but two
4 First Notions of Logic.
poſſible things, one of which is denied. Grant that a man muſt
be either able or unable to do a particular thing, and then not
unable and able are the ſame things. But if we ſuppoſe various
degrees of performance, and therefore degrees of ability, it is
falſe, in the common ſenſe of the words, that two negatives make
an affirmative. Thus, it would be erroneous to ſay, ' John is
able to tranſlate Virgil, and Thomas is not unable ; therefore,
what John can do Thomas can do,' for it is evident that the
premiſes mean that John is ſo near to the beſt ſort of tranſlation
that an affirmation of his ability may be made, while Thomas is
confiderably lower than John, but not ſo near to abſolute defi-
ciency that his ability may be altogether denied. It will generally
be found that two negatives imply an affirmative of a weaker
degree than the poſitive affirmation.
Each of the propoſitions, ' ✗ is Y,' and ' X is not Y,' may
be ſubdivided into two ſpecies: the universal, in which every
poſſible caſe is included ; and the particular, in which it is not
meant to be aſſerted that the affirmation or negation is univerſal.
The four ſpecies of propoſition are then as follows, each being
marked with the letter by which writers on logic have always
diftinguiſhed it.
Y
A Universal Affirmative Every X is
E Universal Negative No X is Y

I Particular Affirmative Some Xs are Ys

Ο Particular Negative Some Xs are not Ys

In common converſation the affirmation of a part is meant to

imply the denial of the remainder. Thus, by ' ſome of the apples
are ripe,' it is always intended to ſignify that ſome are not ripe .
This is not the caſe in logical language, but every propoſition is
intended to make its amount of affirmation or denial, and no
more. When we ſay, ' Some X is Y, or, more grammatically,
' Some Xs are Ys,' we do not mean to imply that ſome are not :
this may or may not be. Again, the word ſome means, one or ،

more, poſſibly all. The following table will ſhew the bearing
of each propofition on the reſt.
EveryXis Yaffirms SomeXs are Ys and denies No Xis r
Some Xs are not Ys
 First Notions ofLogic. 5

No Xis Yaffirms SomeXs are not Ys and denies Some

Every Xis Y
Xs are Ys
SomeXs are Ys doesnot contradi& Every X is Y but denies No X is Y
SomeXs are not YsS
No X is Y
(SomeXsare Ys}but denies Every Xis Y
SomeXsare not Ys doesnot contradictis
Contradictory propoſitions are thoſe in which one denies any
thing that the other affirms ; contrary propoſitions are thoſe in
which one denies every thing which the other affirms, or affirms
every thing which the other denies. The following pair are
contraries,
Every X is Y and No X is Y
and the following are contradictories,
Every X is Y to Some Xs are not Ys
No X is Y to Some Xs are Ys

A contrary, therefore, is a complete and total contradictory ;

and a little confideration will make it appear that the deciſive
distinction between contraries and contradictories lies in this,
that contraries may both be falſe, but of contradictories, one
muſt be true and the other falſe. We may say, ' Either P is true,
or fomething in contradiction of it is true ; but we cannot ſay,
' Either P is true, or every thing in contradiction of it is true.'
It is a very common mistake to imagine that the denial of a
propoſition gives a right to affirm the contrary; whereas it ſhould
be, that the affirmation of a propoſition gives a right to deny the
contrary. Thus, if we deny that Every X is Y, we do not affirm
that No X is Y, but only that Some Xs are not Ys ; while, if we
affirm that Every X is Y, we deny No X is Y, and alſo Some
Xs are not Ys .
But, as to contradictories, affirmation of one is denial of the
other, and denial of one is affirmation of the other. Thus,
either Every X is Y, or Some Xs are not Ys : affirmation ofeither
is denial of the other, and vice verſå.
Let the ſtudent now endeavour to fatisfy himself of the fol-
lowing. Taking the four preceding propoſitions, A, E, I, O,
let the ſimple letter ſignify the affirmation, the ſame letter in pa-
rentheſes the denial, and the abſence of the letter, that there is
neither affirmation nor denial.
6
First Notions of Logic.
0
From A follow (E), I, ( O) From (A) follow
From E .... ( A), (Ι ), Ο From (E) . .... I
From I (E)
....
From (I) .... ( Α), Ε, Ο
From O .... (Α) From (O) A, (E), I
...

Theſe may be thus ſummed up : The affirmation of a univerſal

propoſition, and the denial of a particular one, enable us to affirm
or deny all the other three ; but the denial of a univerſal propo-
ſition, and the affirmation of a particular one, leave us unable to
affirm or deny two of the others.
Infuch propofitions as ' Every X is Y, " Some Xs are not Ys,'
&c., X is called thesubject, and Y the predicate, while the verb
' is ' or ' is not,' is called the copula. It is obvious that the
words of the propoſition point out whether the ſubject is ſpoken
of univerſally or partially, but not ſo of the predicate, which it is
therefore important to examine. Logical writers generally give
the name of distributed ſubjects or predicates to thoſe which are
ſpoken of univerſally ; but as this word is rather technical, I ſhall
ſay that a ſubject or predicate enters wholly or partially, accord-
ing as it is univerſally or particularly ſpoken of.
1. In A, or ' Every X is Y, the ſubject enters wholly, but
the predicate only partially. For it obviously ſays, ' Among the
Ys are all the Xs,' ' Every X is part of the collection of Ys, ſo
that all the Xs make a part of the Ys, the whole it may be.'
Thus, ' Every horſe is an animal,' does not ſpeak of all animals,
but ſtates that all the horſes make up a portion of the animals.
2. In E, or ' No X is Y, both ſubject and predicate enter
،

wholly. No X whatsoever is any one out of all the Ys ; '

' ſearch the whole collection of Ys, and every Y ſhall be found
to be ſomething which is not X.'
3. In I, or ' Some Xs are Ys, both ſubject and predicate enter
partially. ' Some of the Xs are found among the Ys, or make
up a part (the whole poſſibly, but not known from the preceding)
of the Ys .'
4. In O, or ' Some Xs are not Ys,' the ſubject enters partially,
and the predicate wholly. ' Some Xs are none of them any
whatsoever of the Ys ; every Y will be found to be no one out
of a certain portion of the Xs.'
It appears then that,
In affirmatives, the predicate enters partially.
 First Notions of Logic. 7
In negatives, the predicate enters wholly.
In contradictory propoſitions, both ſubject and predicate enter
differently in the two.
The converse of a propoſition is that which is made by inter-
changing the ſubject and predicate, as follows :
The propofition. Its converſe.
A Every X is Y Every Y is X
E No X is Y No Y is X
I Some Xs are Ys Some Ys are Xs
O Some Xs are not Ys Some Ys are not Xs

Now, it is a fundamental and ſelf-evident propofition, that no

conſequence must be allowed to affert more widely than its pre-
miſes ; ſo that, for instance, an aſſertion which is only of fome
Ys can never lead to a reſult which is true of all Ys . But if a
propoſition affert agreement or disagreement, any other propofi-
tion which afſſerts the ſame, to the ſame extent and no further,
must be a legitimate conſequence ; or, if you pleaſe, muſt
amount to the whole, or part, of the original affertion in another
form. Thus, the converſe ofA is not true : for, in ' Every X
is Y, the predicate enters partially ; while in ' Every Y is X,'
the ſubject enters wholly. ' All the Xs make up a part of the
Ys, then a part of the Ys are among the Xs, or ſome Ys are Xs.'
Hence, the only legitimate converſe of ' Every X is Y ' is, ' Some
Ys are Xs. ' But in ' No X is Y, both ſubject and predicate enter
wholly, and ' No Y is X ' is, in fact, the ſame propoſition as
No X is Y.' And ' Some Xs are Ys ' is alſo the ſame as its con-
verſe Some Ys are Xs : ' here both terms enter partially. But
Some Xs are not Ys ' admits of no converſe whatever ; it is per-
fectly conſiſtent with all aſſertions upon Y and X in which Y is
the ſubject. Thus neither of the four following lines is incon-
ſiſtent with itſelf.

Some Xs are not Ys and Every Y is X

Some Xs are not Ys and No Y is X
Some Xs are not Ys and Some Ys are Xs
Some Xs are not Ys and Some Ys are not Xs .

Having thus diſcuſſed the principal points connected with the

ſimple aſſertion, I paſs to the manner of making two aſſertions
8
First Notions of Logic.
give a third. Every inſtance ofthis is called afyllogifm, the two
aſſertions which form the basis of the third are called premises,
and the third itſelf the conclufion.
If two things both agree with a third in any particular, they
agree with each other in the ſame ; as, ifX be ofthe ſame colour as
Y, and Z ofthe ſame colour as Y, thenX is ofthe ſame colour as
Z. Again, if X differ from Y in any particular in which Z
agrees with Y, then X and Z differ in that particular. If ✗ be
not of the fame colour as Y, and Z be of the ſame colour as Y,
then X is not of the colour of Z. But if X and Z both differ
from Y in any particular, nothing can be inferred ; they may
either differ in the ſame way and to the fame extent, or not.
Thus, if X and Z be both of different colours from Y, it neither
follows that they agree, nor differ, in their own colours.
The paragraph preceding contains the eſſential parts of all in-
ference, which conſiſts in comparing two things with a third, and
finding from their agreement or difference with that third, their
agreement or difference with one another. Thus, Every X is
Y, every Z is Y, allows us to infer that X and Z have all thoſe
qualities in common which are neceſſary to Y. Again, from
every X is Y, and ' No Z is Y,' we infer that X and Z differ
from one another in all particulars which are eſſential to Y. The
preceding forms, however, though they repreſent common reaſon-
ing better than the ordinary ſyllogiſm, to which we are now com-
ing, do not conftitute the ultimate forms of inference. Simple iden-
tity or non-identity is the ultimate ſtate to which every aſſertion
may be reduced ; and we ſhall, therefore, firſt aſk, from what
identities, &c., can other identities, &c., be produced ? Again,
ſince we name objects in ſpecies, each ſpecies confifting of a
number of individuals, and fince our aſſertion may include all or
only part of a ſpecies, it is further neceſſary to aſk, in every in-
ſtance, to what extent the conclufion drawn is true, whether of
all, or only of part ?
Let us take the ſimple aſſertion, ' Every living man reſpires ;'
or every living man is one of the things (however varied they
may be) which reſpire. If we were to encloſe all living men in
a large triangle, and all reſpiring objects in a large circle, the pre-
ceding aſſertion, if true, would require that the whole of the tri-
angle ſhould be contained in the circle. And in the ſame way we
 First Notions of Logic. 9
may reduce any aſſertion to the expreſſion of a coincidence, total
or partial, between two figures. Thus, a point in a circle may
repreſent an individual of one ſpecies, and a point in a triangle
an individual of another ſpecies : and we may expreſs that the
whole of one ſpecies is aſſerted to be contained or not contained
in the other by ſuch forms as, ' All the is in the ' ; ' None
of the A is in the '.
Any two aſſertions about X and Z, each expreſſing agreement
or diſagreement, total or partial, with or from Y, and leading to a
concluſion with reſpect to X or Z, is called a fyllogifm, of which
Y is called the middle term. The plaineſt ſyllogiſm is the follow-
ing :-
Every X is Y All the is in the
Every Y is Z All the is in the ☐
Therefore Every X is Z Therefore All the is in the

In order to find all the poſſible forms of fyllogifm, we must

make a table of all the elements of which they can confift ;
namely-
X and Y Z and Y
Every X is Y A Every Z is Y
No X is Y E No Z is Y
Some Xs are Ys I Some Zs are Ys
Some Xs are not Ys O Some Zs are not Ys
Every Y is X A Every Y is Z
Some Ys are not Xs O Some Ys are not Zs
Or their ſynonymes,
▲ and O and
All the is in the O A All the is in the
None of the is in the E None of the is in the
Some of the A is in the I Some of the ☐ is in the
Some ofthe A is not in the ☐ O Some of the is not in the
All the is in the A A All the is in the
Some of the is not in the A O Some of the ☐ is not in the
Now, taking any one ofthe fix relations between X and Y,
and combining it with either of thoſe between Z and Y, we
have fix pairs of premiſes, and the ſame number repeated for
every different relation of X to Y. We have then thirty-fix
10
First Notions of Logic.
forms to confider : but, thirty of theſe (namely, all but (A, A)
(E, E), &c.,) are halfofthem repetitions of the other half. Thus, ,

' Every X is Y, no Z is Y,' and ' Every Z is Y, no X is Y,

are of the fame form, and only differ by changing X into Z and
Z into X. There are then only 15+6, or 21 distinct forms,
ſome of which give a neceſſary conclufion, while others do not.
We ſhall ſelect the former of theſe, claſſifying them by their
conclufions ; that is, according as the inference is of the form
A, E, I , or O.
I. In what manner can a univerſal affirmative conclufion be
drawn ; namely, that one figure is entirelycontained in the other ?
This we can only aſſert when we know that one figure is entirely
contained in the circle, which itſelf is entirely contained in the
other figure. Thus,
Every X is Y All the is in the O A
Every Y is Z All the is in the A

Every X is Z All the is in the A

is the only way in which a univerſal affirmative concluſion can

be drawn .
II. In what manner can a univerſal negative concluſion be
drawn ; namely, that one figure is entirely exterior to the other ?
Only when we are able to affert that one figure is entirely within,
and the other entirely without, the circle. Thus,
Every X is Y All the is in the O A
No Z is Y None of the☐ is in the ☐ E
No X is Z None of the is in the E

is the only way in which a univerſal negative conclufion can be

drawn .
III. In what manner can a particular affirmative concluſion be
drawn ; namely, that part or all of one figure is contained in the
other ? Only when we are able to affert that the whole circle is
part of one of the figures, and that the whole, or part of the cir-
cle, is part ofthe other figure. We have then two forms.
Every Y is X All the is in the A A
Every Y is Z All the is in the A
Some Xs are Zs Some of the is in the I
 11
First Notions of Logic.
All the is in the A A
Every Y is X
Some Ys are Zs Some of the is in the I
Some Xs are Zs Some of the A is in the I

The ſecond of theſe contains all that is ſtrictly neceſſary to the

conclufion, and the firſt may be omitted. That which follows
when an aſſertion can be made as to fome, must follow when the
ſame aſſertion can be made of all .
IV. How can a particular negative propoſition be inferred ;
namely, that part, or all of one figure, is not contained in the
other ? It would seem at firſt ſight, whenever we are able to
aſſert that part or all of one figure is in the circle, and that part
or all ofthe other figure is not. The weakeſt ſyllogiſm from which
fuch an inference can be drawn would then ſeem to be as follows.

Some Xs are Ys Some of the is in the or

Some Zs are not Ys Some of the is not in the
... Some Zs are not Xs ... Some of the is not in the 0

But here it will appear, on a little confideration, that the con-

clufion is only thus far true ; that thoſe Xs which are Ys cannot
be thoſe Zs which are not Ys ; but they may be other Zs, about
which nothing is aſſerted when we say thatſome Zs are not Ys .
And further confideration will make it evident, that a conclufion
of this form can only be arrived at when one of the figures is
entirely within the circle, and the whole, or part of the other
without ; or elſe when the whole of one of the figures is without
the circle, and the whole or part of the other within ; or laſtly,
when the circle lies entirely within one of the figures, and not
entirely within the other. That is, the following are the distinct
forms which allow of a particular negative conclufion, in which
it ſhould be remembered that a particular propofition in the pre-
miſes may always be changed into a univerſal one, without affect-
ing the conclufion. For that which neceſſarily follows from
" fome," follows from " all. "

Every X is Y All the is in the A

Some Zs are not Ys Some of the is not in the Ο
... Some Zs are not Xs Some of the is not in the A
 12
First Notions of Logic.
No X is Y None of the is in the E
Some Zs are Ys Some of the is in the I
... Some Zs are not Xs Some of the ☐ is not in the A 0

Every Y is X All the is in the A A

Some Ys are not Zs Some of the ☐ is not in the 0
... Some Xs are not Zs Some of the A is not in the

It appears, then, that there are but fix diſtinct ſyllogiſms. All
others are made from them by ſtrengthening one of the premiſes,
or converting one or both of the premiſes, where ſuch converfion
is allowable ; or elſe by firſt making the converfion, and then
ſtrengthening one of the premiſes. And the following arrange-
ment will ſhow that two ofthem are univerſal, three of the others
being derived from them by weakening one of the premiſes in a
manner which does not deſtroy, but only weakens, the conclu-
fion.
1. Every X is Y 3. Every X is Y
Every Y is Z No Z isY .........

Every X is Z No X is Z

2. Some Xs are Ys 4. Some Xs are Ys 5. Every X is Y 6. Every Y is X

Every Y is Z No Zis Y Some Zs are not Ys Some Ys are notZs

Some Xs are Zs Some Xs are notZs Some Zs are not Xs SomeXs are not Zs

We may ſee how it ariſes that one of the partial fyllogifms is

not immediately derived, like the others, from a univerſal one .
In the preceding, A E E may be confidered as derived from
A A A, by changing the term in which Y enters univerſally into
a univerſal negative. If this be done with the other term instead,
we have

No X is Y) from which univerſal premiſes we cannot deduce a

Every Y is Zuniverſal concluſion, but only fome Zs are not Xs .
If we weaken one and the other of theſe premiſes, as they
ſtand, we obtain
Some Xs are not Ys No X is Y
Every Y is Z and Some Ys are Zs
No conclufion Some Zs are not Xs
 First Notions of Logic. 13
equivalent to the fourth of the preceding : but if we convert the
firſt premiſe, and proceed in the ſame manner,
From No Y is X we obtain Some Ys are not Xs
Every Y is Z Every Y is Z
Some Zs are not Xs Some Zs are not Xs

which is legitimate, and is the ſame as the laſt of the preceding

lift, with X and Z interchanged.
Before proceeding to ſhow that all the uſual forms are con-
tained in the preceding, let the reader remark the following rules,
which may be proved either by collecting them from the preceding
cafes, or by independent reaſoning.
1. The middle term muſt enter univerſally into one or the other
premiſe. If it were not ſo, then one premiſe might ſpeak of one
part of the middle term, and the other of another ; ſo that there
would, in fact, be no middle term. Thus, ' Every X is Y, Every
Z is Y,' gives no concluſion : it may be thus ſtated ;
All the Xs make up a part of the Ys
All the Zs make up a part of the Ys

And, before we can know that there is any common term of

compariſon at all, we must have ſome means of ſhowing that the
two parts are to fome extent the fame ; or the preceding premiſes
by themſelves are inconcluſive.
2. No term muſt enter the concluſion more generally than it
is found in the premises ; thus, ifX be ſpoken of partially in the
premiſes, it must enter partially into the conclufion. This is ob-
vious, fince the conclufion muſt aſſert no more than the premiſes
imply.
3. From premiſes both negative no conclufion can be drawn.
For it is obvious, that the mere aſſertion of diſagreement between
each of two things and a third, can be no reaſon for inferring
either agreement or diſagreement between theſe two things. It
will not be difficult to reduce any caſe which falls under this rule
to a breach of the firſt rule : thus, No X is Y, No Z is Y, gives
Every X is (fomething which is not Y)
Every Z is (fomething which is not Y)
14 First Notions of Logic.
in which the middle term is not ſpoken of univerſally in either.
Again, ' No Y is X, ſome Ys are not Zs,' may be converted into
Every X is (a thing which is not Y)
Some (things which are not Zs) are Ys
in which there is no middle term .
4. From premiſes both particular no conclufion can be drawn.
This is ſufficiently obvious when the firſt or ſecond rule is broken,
as in ' Some Xs are Ys, Some Zs are Ys.' But it is not immediately
obvious when the middle term enters one of the premiſes uni-
verſally. The following reaſoning will ſerve for exerciſe in the
preceding reſults. Since both premiſes are particular in form,
the middle term can only enter one ofthem univerſally by being
the predicate of a negative propofition ; conſequently (Rule 3)
the other premiſe must be affirmative, and, being particular, nei-
ther of its terms is univerſal. Conſequently both the terms as to
which the concluſion is to be drawn enter partially, and the con-
cluſion (Rule 2) can only be a particular affirmative propofition.
But if one of the premiſes be negative, the conclufion must be
negative (as we ſhall immediately fee). This contradiction ſhows
that the ſuppoſition of particular premiſes producing a legitimate
reſult is inadmiſſible.
5. If one premiſe be negative, the concluſion, if any, muſt be
negative. If one term agree with a ſecond and diſagree with a
third, no agreement can be inferred between the ſecond and
third.
6. If one premiſe be particular, the conclufion must be par-
ticular. This may be ſhown as follows. If two propoſitions
P and Q, together prove a third, R, it is plain that P and the
denial of R, prove the denial of Q. For P and Qcannot be true
together without R. Now if poſſible, let P (a particular) and Q
(a univerſal) prove R (a univerſal). Then P (particular) and
the denial of R (particular) prove the denial of Q. But two
particulars can prove nothing.
In the preceding ſet of ſyllogiſms we obſerve one form only
which produces A, or E, or I, but three which produce O.
Let an affertion be faid to be weakened when it is reduced
from univerſal to particular, and ſtrengthened in the contrary cafe.
Thus, ' Every X is Z' is called ſtronger than ' Some Xs are Zs . '
 First Notions of Logic. 15
Every uſual form of fyllogifm which can give a legitimate re-
ſult is either one of the preceding fix, or another formed from
one of the fix, either by changing one of the aſſertions into its
converſe, if that be allowable, or by ſtrengthening one of the
premiſes, without altering the conclufion, or both. Thus,
Some Xs are Ys Some Ys are Xs
Every Y is Z } may be written Every Y is Z
X
What follows will ſtill follow from Every Yis Z
Every Y is
for all which is true when ' Some Ys are Xs,' is not leſs true when
' Every Y is X.'
It would be poſſible alſo to form a legitimate ſyllogiſm by
weakening the conclufion, when it is univerſal, fince that which
is true of all is true ofſome. Thus, ' Every X is Y, Every Y
is Z,' which yields ' Every X is Z,' alſo yields ' Some Xs are Zs .'
But writers on logic have always confidered theſe ſyllogiſms as
uſeleſs, conceiving it better to draw from any premiſes their
ſtrongest conclufion. In this they were undoubtedly right ; and
the only question is, whether it would not have been adviſable
to make the premiſes as weak as poſſible, and not to admit any
ſyllogiſms in which more appeared than was abſolutely neceſſary
to the conclufion. If ſuch had been the practice, then
Every Y is X, Every Y is Z, therefore Some Xs are Zs
would have been confidered as formed by a ſpurious and unne-
ceſſary exceſs ofaſſertion. The minimum of aſſertion would be
contained in either of the following,
Every Y is X, Some Ys are Zs, therefore Some Xs are Zs
Some Ys are Xs, Every Y is Z, therefore Some Xs are Zs
In this chapter, ſyllogiſms have been divided into two claſſes :
firſt, thoſe which prove a univerſal conclufion ; ſecondly, thoſe
which prove a partial conclufion, and which are (all but one)
derived from the firſt by weakening one of the premiſes, in ſuch
manner as to produce a legitimate but weakened conclufion.
Thoſe of the firſt claſs are placed in the firſt column, and of the
other in the ſecond.
16 First Notions of Logic.
Univerſal . Particular.
Some Xs are Ys I
A Every X is Y
A Every Y is Z Every Y is Z A

Some Xs are Zs I
A Every X is Z
Some Xs are Ys I
No Y is Z E

A Every X is Y Some Xs are not Zs O

E No Y is Z
Every X is Y A
E No X is Z Some Zs are not Ys O

Some Zs are not Xs Ο

Every Y is X A
........ Some Ys are not Zs 0

Some Xs are not Zs

In all works on logic, it is cuſtomary to write that premiſe
firſt which contains the predicate of the conclufion. Thus,
Every Y is Z Every X is Y
Every X is Y would be written, and not Every Y is Z
Every X is Z Every X is Z
The premiſes thus arranged are called major and minor ; the pre-
dicate of the conclufion being called the major term, and its fub-
ject the minor. Again, in the preceding caſe we ſee the various
ſubjects coming in the order Y, Z , X, Y ; X, Z : and the num-
ber of different orders which can appear is four, namely-
YZ ZY YZ ZY
XY XY YX YX

XZ XZ XZ XZ

which are called the four figures, and every kind of fyllogifm in
each figure is called a mood. I now put down the various moods
of each figure, the letters of which will be a guide to find out
thoſe of the preceding lift from which they are derived. Co
means that a premiſe of the preceding lift has been converted ;
+ that it has been ſtrengthened ; Co+, that both changes have
taken place. Thus,
 First Notions of Logic. 17
A Every Y is Z A Every Y is Z
I Some Xs are Ys becomes A Every Y is X : (Co +)
I Some Xs are Zs I Some Xs are Zs

And Co + points out the following : If ſome Xs be Ys, then

ſome Ys are Xs (Co) ; and all that is true when Some Ys are Xs,
is true when Every Y is X (+) ; therefore the ſecond ſyllogifm
is legitimate, if the firſt be ſo.
First Figure.
A Every Y is Z A Every Y is Z
A Every X is Y I Some Xs are Ys

A Every X is Z I Some Xs are Zs

E No Y is Z E No Y is Z
A Every X is Y I Some Xs are Ys

E No X is Z O Some Xs are not Zs

Second Figure.
E No Zis Y (Co) E No Zis Y (Co)
A Every X is Y I Some Xs are Ys

E No X is Z O Some Xs are not Zs

A Every Z is Y A Every Z is Y
E No X is Y (Co) O Some Xs are not Ys

E No X is Z Some Xs are not Zs

Third Figure.
A Every Y is Z E No Y is Z
A Every Y is X (Co +) A Every Y is X (Co +)
I Some Xs are Zs O Some Xs are not Zs

I Some Ys are Zs (Co) O Some Ys are not Zs

A Every Y is X A Every Y is X
I Some Xs are Zs O Some Xs are not Zs
A Every Y is Z E No Y is Z
I Some Ys are Xs (Co) I Some Ys are Xs (Co)
I Some Xs are Zs O Some Xs are not Zs
C
 18 First Notions of Logic.
Fourth Figure.
A Every Z is Y (+) I Some Zs are Ys
A Every Y is X A Every Y is X
I Some Xs are Zs I Some Zs are Xs

A Every Z is Y E No ZisY (Co)

E No Y isX A Every Y is X (Co +)
E No X is Z O Some Xs are not Zs

E No Zis Y (Co)
I Some Ys are Xs ( Co)
O Some Xs are not Zs

The above is the ancient method of dividing fyllogifms ; but,

for the preſent purpoſe, it will be ſufficient to confider the fix
from which the reſt can be obtained. And fince ſome of the
fix have X in the predicate of the conclufion, and not Z, I ſhall
join to them the fix other fyllogifms which are found by tranf-
pofing Z and X. The complete liſt, therefore, offyllogifms with
the weakeſt premiſes and the ſtrongest concluſions, in which a
comparison of X and Z is obtained by compariſon of both with
Y, is as follows :
Every X is Y Every Z is Y Some Xs are Ys Some Zs are Ys
Every Y is Z Every Y is X No Yis Z No Yis X

Every X is Z Every Z is X Some Xs are not Zs Some Zs are not Xs

Every X is Y Every Z is Y Every X is Y Every Z is Y

No Y is Z No Y is X Some Zs are not Ys Some Xs are not Ys

No X isZ No Z is X Some Zs are not Xs Some Xs are not Zs

Some Xs are Ys Some Zs are Ys Every Y is X Every Y is Z
Every Y is Z Every Y is X Some Ys are not Zs Some Ys are not Xs

Some Xs are Zs Some Zs are Xs Some Xs are not Zs Some Zs are not Xs

In the lift ofpage 12, there was nothing but recapitulation of

forms, each form admitting a variation by interchanging X and
Z. This interchange having been made, and the reſults col-
lected as above, if we take every caſe in which Z is the predi-
cate, or can be made the predicate by allowable converfion, we
 First Notions of Logic. 19
have a collection of all poffible weakest forms in which the reſult
is one ofthe four ' Every X is Z, No X is Z, Some Xs are Zs,'
' Some Xs are not Zs ; ' as follows. The premiſes are written
in what appeared the moſt natural order, without distinction of
major or minor.
Every X is Y
Every Y is Z
Every X is Z
Some Xs are Ys Some Zs are Ys
Every Y is Z Every Y is X
Some Xs are Zs Some Xs are Zs

Every X is Y Every Z is Y
No Z is Y No X is Y

No X is Z No X is Z
Some Xs are Ys Every Z is Y Every Y is X
No Z is Y Some Xs are not Ys Some Ys are not Zs
Some Xs are not Zs Some Xs are not Zs Some Xs are not Zs
Every aſſertion which can be made upon two things by com-
pariſon with any third, that is, every ſimple inference, can be
reduced to one of the preceding forms. Generally ſpeaking, one
of the premiſes is omitted, as obvious from the conclufion ; that
is, one premiſe being named and the conclufion, that premiſe is
implied which is neceſſary to make the concluſion good. Thus,
if I say, " That race muſt have poſſeſſed ſome of the arts of life,
for they came from Afia," it is obviously meant to be aſſerted,
that all races coming from Afia must have poſſeſſed ſome ofthe
arts of life. The preceding is then a fyllogifm, as follows :
That race is ' a race of Afiatic origin : '
Every ' race of Afiatic origin' is ' a race which must
have poſſeſſed ſome of the arts of life : '
Therefore, That race is a race which must have poſſeſſed
ſome of the arts of life.
A perſon who makes the preceding aſſertion either means to
imply, antecedently to the conclufion, that all Afiatic races muſt
have poſſeſſed arts, or he talks nonſenſe if he aſſert the conclu-
20
First Notions of Logic.
ſion poſitively. '✗ must be Z,for it isY, can only be an inference
when ' Every Y is Z.' This latter propoſition may be called
the ſuppreſſed premiſe ; and it is in ſuch ſuppreſſed propoſitions
that the greatest danger of error lies. It is alſo in ſuch propofi-
tions that men convey opinions which they would not willingly
expreſs. Thus, the honeſt witneſs who ſaid, ' I always thought
him a reſpectable man-he kept his gig,' would probably not
have admitted in direct terms, ' Every man who keeps a gig muſt
be reſpectable.'

I ſhall now give a few detached illuſtrations of what precedes.

" His imbecility of character might have been inferred from
his proneneſs to favourites ; for all weak princes have this fail-
ing. " The preceding would ſtand very well in a hiſtory, and
many would paſs it over as containing very good inference.
Written, however, in the form of a ſyllogiſm, it is,
All weak princes are prone to favourites
He was prone to favourites
Therefore He was a weak prince
which is palpably wrong. (Rule 1.) The writer of ſuch a ſen-
tence as the preceding might have meant to ſay, ' for all who
have this failing are weak princes ; ' in which caſe he would have
inferred rightly. Every one ſhould be aware that there is much
falſe form of inference arifing out of badneſs of ſtyle, which is
juſt as injurious to the habits of the untrained reader as if the
errors were miſtakes of logic in the mind of the writer.
' X is leſs than Y ; Y is less than Z : therefore X is less than
Z.' This, at firſt ſight, appears to be a ſyllogiſm ; but, on re-
ducing it to the uſual form, we find it to be,
X is (a magnitude leſs than Y)
Y is (a magnitude leſs than Z)
Therefore X is (a magnitude leſs than Z)
which is not a ſyllogiſm, ſince there is no middle term. Evident
as the preceding is, the following additional propoſition must be
formed before it can be made explicitly logical. If Y be a mag-
nitude leſs than Z, then every magnitude leſs than Y is alſo leſs
than Z. ' There is, then, before the preceding can be reduced
to a ſyllogiſtic form, the neceſſity of a deduction from the ſecond
 First Notions of Logic. 21

premiſe, and the ſubſtitution ofthe reſult inſtead of that premiſe.

Thus,
X is less than Y
Leſs than Y is less than Z : following from Y is leſs than Z.
Therefore X is leſs than Z

But, if the additional argument be examined-namely, if Y be

leſs than Z, then that which is less than Y is less than Z-it will
be found to require preciſely the ſame confiderations repeated ;
for the original inference was nothing more. In fact, it may
eaſily be ſeen as follows, that the propoſition before us involves
more than any ſimple fyllogiſm can expreſs. When we say that
X is leſs than Y, we say that if X were applied to Y, every part
of X would match a part of Y, and there would be parts of Y
remaining over. But when we ſay, ' Every X is Y,' meaning
the premiſe of a common fyllogifm, we say that every inſtance of
X is an inſtance of Y, without ſaying any thing as to whether
there are or are not inſtances of Y ſtill left, after those which
are alſo X are taken away. If, then, we wish to write an ordi-
nary ſyllogiſm in a manner which ſhall correſpond with ' X is leſs
than Y, Y is less than Z, therefore X is less than Z,' we must
introduce a more definite amount of aſſertion than was made in
the preceding forms. Thus,
Every X is Y, and there are Ys which are not Xs
Every Y is Z, and there are Zs which are not Ys
Therefore Every X is Z, and there are Zs which are not Xs
Or thus :
The Ys contain all the Xs, and more
The Zs contain all the Ys, and more
The Zs contain all the Xs, and more
The moſt technical form, however, is,
From Every X is Y ; [Some Ys are not Xs]
Every Y is Z ; [Some Zs are not Ys]
Follows Every X is Z ; [ Some Zs are not Xs]
This fort of argument is called àfortiori argument, becauſe the
premiſes are more than ſufficient to prove the conclufion, and the
extent of the concluſion is thereby greater than its mere form
would indicate. Thus, ' ✗ is less than Y, Y is less than Z,
22
First Notions of Logic.
therefore, à fortiori, X is less than Z,' means that the extent to
which X is less than Z must be greater than that to which X is
leſs than Y, or Y than Z. In the ſyllogiſm laſt written, either
ofthe bracketted premiſes might be ſtruck out without deſtroying
the conclufion ; which laſt would, however, be weakened. As
it ſtands, then, the part of the conclufion, ' Some Zs are not
Xs,' follows àfortiori.
The argument àfortiori may then be defined as a univerſally
affirmative ſyllogiſm, in which both of the premiſes are ſhewn to
be leſs than the whole truth, or greater. Thus, in ' Every X is
Y, Every Y is Z, therefore Every X is Z, we do not certainly
imply that there are more Ys than Xs, or more Zs than Ys, ſo
that we do not know that there are more Zs than Xs . But if
we be at liberty to ſtate the ſyllogiſm as follows,
All the Xs make up part (and part only) of the Ys
Every Y is Z ;
then we are certain that

All the Xs make up part (and part only) of the Zs.

But if we be at liberty further to ſay that
All the Xs make up part (and part only) of the Ys
All the Ys make up part (and part only) of the Zs
then we conclude that

All the Xs make up part ofpart (only) of the Zs

and the words in Italics mark that quality of the conclufion from
which the argument is called àfortiori.
Moſt fyllogifms which give an affirmative concluſion are gene-
rally meant to imply àfortiori arguments, except only in mathe-
matics. It is ſeldom, except in the exact ſciences, that we meet
with a propofition, ' Every X is Z,' which we cannot immediately
couple with ' fome Zs are not Xs.'
When an argument is completely eſtabliſhed, with the excep-
tion of one aſſertion only, ſo that the inference may be drawn as
foon as that one aſſertion is eſtabliſhed, the reſult is ſtated in a
form which bears the name of an hypothetical fyllogifm. The
word hypotheſis means nothing but ſuppoſition ; and the ſpecies
of fyllogifm juſt mentioned firſt lays down the aſſertion that a
conſequence will be true if a certain condition be fulfilled, and
 First Notions of Logic . 23

then either afferts the fulfilment of the condition, and thence the
conſequence, or elſe denies the conſequence, and thence denies
the fulfilment of the condition. Thus, if we know that
When X is Z, it follows that P is Q ;
then, as foon as we can aſcertain that X is Z, we can conclude
that P is Q ; or, if we can ſhew that P is not Q, we know that
X is not Z. But if we find that X is not Z, we can infer no-
thing ; for the preceding does not aſſert that P is Qonly when
X is Z. And if we find out that P is Qwe can infer nothing.
This conditional fyllogifm may be converted into an ordinary
fyllogifm, as follows. Let K be any ' caſe in which X is Z,' and
V, a ' caſe in which P is Q;' then the preceding aſſertion amounts
to ' Every Kis V.' Let L be a particular inſtance, the X of
which may or may not be Z. If X be Z in the inſtance under
diſcuſſion, or if X be not Z, we have, in the one caſe and the
other,
Every K is V Every K is V
Lisa K L is not a K

Therefore Lisa V No conclufion

Similarly, according as a particular cafe (M) is or is not V, we
have
Every K is V Every K is V
Mis a V M is not a V
No conclufion M is not a K

That is to ſay : the aſſertion of an hypothefis is the aſſertion of

its neceſſary conſequence, and the denial of the neceſſary conſe-
quence is the denial of the hypothesis : but the aſſertion of the
neceſſary conſequence gives no right to affert the hypotheſis, nor
does the denial ofthe hypotheſis give any right to deny the truth
of that which would (were the hypotheſis true) be its neceſſary
conſequence.
Demonſtration is of two kinds : which ariſes from this, that
every propoſition has a contradictory ; and of theſe two, one
muſt be true and the other must be falſe. We may then either
prove a propoſition to be true, or its contradictory to be falſe.
' It is true that every X is Z,' and ' it is falſe that there are ſome
Xs which are not Zs,' are the ſame propoſition ; and the proof
ofeither is called the indirect proof of the other.
24 First Notions of Logic.
But how is any propoſition to be proved falſe, except by prov-
ing a contradiction to be true ? By proving a neceſſary conſe-
quence of the propoſition to be falſe. But this is not a complete
anſwer, ſince it involves the neceſſity of doing the ſame thing ;
or, ſo far as this anſwer goes, one propoſition cannot be proved
falſe unleſs by proving another to be falſe. But it may happen,
that a neceſſary conſequence can be obtained which is obviously
and ſelf-evidently falſe, in which caſe no further proof of the
falſehood of the hypotheſis is neceſſary. Thus the proof which
Euclid gives that all equiangular triangles are equilateral is of the
following ſtructure, logically confidered.
(1.) If there be an equiangular triangle not equilateral, it fol-
lows that a whole can be found which is not greater than its
part.*
(2.) It is falſe that there canbe any whole which is not greater
than its part (felf evident).
(3.) Therefore it is falſe that there is any equiangular triangle
which is not equilateral ; or all equiangular triangles are equila-
teral.

When a propofition is eſtabliſhed by proving the truth of the

matters it contains, the demonſtration is called direct ; when by
proving the falſehood of every contradictory propoſition, it is
called indirect. The latter ſpecies of demonſtration is as logical
as the former, but not of ſo ſimple a kind ; whence it is defira-
ble to use the former whenever it can be obtained.
The use of indirect demonſtration in the Elements of Euclid
is almoſt entirely confined to thoſe propoſitions in which the con-
verſes of fimple propoſitions are proved. It frequently happens
that an eſtabliſhed aſſertion of the form

Every X is Z ........... (I )
may be easily made the means of deducing,
Every (thing not X) is not Z .. (2)
which laſt gives
Every Z is X ............. (3)
* This is the propoſition in proof of which nearly the whole of the de-
monftration of Euclid is ſpent.
 First Notions of Logic. 25
The converſion of the ſecond propoſition into the third is
uſuallymadebyanindirectdemonſtration,in the following manner :
If poffible, let there be one Z, which is not X, (2) being true.
Then there is one thing which is not X and is Z ; but every .
thing not X is not Z ; therefore there is one thing which is Z
and is not Z : which is abſurd. It is then abſurd that there
ſhould be one fingle Z which is not X ; or, Every Z is X.
The following propoſition contains a method which is of fre-
quent uſe.
HYPOTHESIS. Let there be any number of propoſitions or
aſſertions,-three for instance, X, Y, and Z, of which it is the
property that one or the other must be true, and one only. Let
there be three other propofitions, P, Q, and R, of which it is
alſo the property that one, and one only, must be true. Let it
alſo be a connexion of thoſe aſſertions, that
When X is true, Pis true
When Y is true, Qis true
When Z is true, R is true
CONSEQUENCE : then it follows that
When P is true, X is true
When Qis true, Y is true
When R is true, Z is true
For, when P is true, then Qand R must be falſe ; conſequently,
neither Y nor Z can be true, for then Q or R would be true.
But either X, Y, or Z must be true, therefore X muſt be true ;
or, when P is true, X is true. In a ſimilar way the remaining
aſſertions may be proved.
Cafe 1. If When P is Q, X is Z
When P is not Q, X is not Z
It follows that When X is Z, Pis Q
When X is not Z, P is not Q
222

When X is greater than Z, P is greater than

Cafe 2. If When X is equal to Z, P is equal to
When X is less than Z, P is less than
When P is greater than Q, X is greater than Z
It follows that WhenP is equal to Q, X is equal to Z

When P is less than Q, X is less than Z

 26

CHAPTER II .
‫רן‬

On Objects, Ideas, and Names.

OGIC is derived from a Greek word (λόγος) which ſignifies
L communication of thought, uſually by ſpeech. It is the
name which is generally given to the branch ofinquiry (be itcalled
ſcience or art), in which the act of the mind in reaſoning is con-
fidered, particularly with reference to the connection of thought
and language. But no definition yet given in few words has
been found fatisfactory to any confiderable number of thinking
perſons.
All exiſting things upon this earth, which have knowledge of
their own exiſtence, poſſeſs, ſome in one degree and ſome in
another, the power of thought, accompanied by perception, which
is the awakening of thought by the effect of external objects
upon the ſenſes. By thought I here mean, all mental action, not
only that comparatively high ſtate of it which is peculiar to man,
but alſo that lower degree of the ſame thing which appears to be
poſſeſſed by brutes.
With reſpect to the mind, confidered as a complicated ap-
paratus which is to be ſtudied, we are not even ſo well off as
thoſe would be who had to examine and decide upon the me-
chaniſm of a watch, merely by obſervation of the functions of
the hands, without being allowed to see the inſide. A mechani-
cian, to whom a watch was preſented for the first time, would be
able to give a good gueſs as to its ſtructure, from his knowledge of
other pieces ofcontrivance. As foon as he had examined the law of
the motion of the hands, he might conceivably invent an inftru-
ment with ſimilar properties, in fifty different ways. But in the
caſe of the mind, we have manifeſtations only, without the
ſmalleſt power of reference to other fimilar things, or the leaſt
knowledge of ſtructure or proceſs, other than what may be
derived from thoſe manifeſtations. It is the problem of the watch
to thoſe who have never ſeen any mechaniſm at all.
 On Objects, Ideas, and Names. 27
We have nothing more to do with the ſcience of mind,
uſually called metaphysics,* than to draw a very few neceſſary
diſtinctions, which, whatever names we uſe to denote them, are
matters of fact connected with our ſubject. Some modes of
expreſſing them favor one ſyſtem of metaphyſics, and ſome
another ; but still they are matters of obſerved fact. Our words
muſt be very imperfect ſymbols, drawn from compariſon of the
manifeſtations of thought with thoſe of things in corporeal ex-
iſtence. For instance, I just now ſpoke of the mind as an
apparatus, or piece of mechaniſm. It is a ſtructure of ſome fort,
which has the means of fulfilling various purpoſes ; and ſo far it
reſembles the hand, which by the diſpoſition of bone and muſcle,
can be made to perform an immenſe variety of different motions
and graſps. Where the reſemblance begins to be imperfect, and
why, is what we cannot know. In all probability we ſhould
need new modes of perception, other ſenſes beſides fight, hear-
ing, and touch, in order to know thought as we know colour,
ſize, or motion. But the purpoſe of the preſent treatiſe is only
the examination of fome of the manifeſtations of thinking power
in their relation to the language in which they are expreſſed.
Knowledge of thought and knowledge of the reſults of thought,
* All ſyſtems make an aſſumption of the uniformity of proceſs in all
minds, carried to an extent the propriety of which ought to be a matter of
ſpecial diſcuſſion. There are no writers who give us ſo much must with ſo
little why, as the metaphyficians. If perſons who had only ſeen the outſide
of the timepiece, were to invent machines to anſwer its purpoſe, they might
arrive at their object in very different ways. One might uſe the pendulum
and weight, another the ſprings and the balance : one might diſcover the
combination of toothed wheels, another a more complicated action of lever
upon lever. Are we fure that there are not differences in our minds, ſuch
as the preceding inſtance may ſuggeſt by analogy ; if ſo, how are we fure ?
Again, if our minds be as tables with many legs, do we know that a weight
put upon different tables will be ſupported in the fame manner in all. May
not the fame leg ſupport much or all of a certain weight in one mind, and
little or nothing in another ? I have ſeen ſtriking inſtances of ſomething like
this, among those who have examined for themselves the grounds of the
mathematical ſciences .
I would not diſſuade a ſtudent from metaphyſical inquiry ; on the con-
trary, I would rather endeavour to promote the defire of entering upon fuch
ſubjects : but I would warn him, when he tries to look down his own throat
with a candle in his hand, to take care that he does not ſet his head on fire.
 28 On Objects, Ideas, and Names.
are very different things. The watch abovementioned might
have the functions of its hands diſcovered, might be uſed in find-
ing longitude (and even latitude) all over the world, without the
parties uſing it having the ſmallest idea of its interior ſtructure.
That our minds, ſouls, or thinking powers (uſe what name
we may) exiſt, is the thing of all others of which we are moſt
certain, each for himself. Next to this, nothing can be more
certain to us, each for himself, than that other things alſo exiſt ;
other minds, our own bodies, the whole world of matter. But
between the character of theſe two certainties there is a vaſt dif-
ference. Any one who ſhould deny his own exiſtence would,
if ſerious, be held beneath argument : he does not know the
meaning of his words, or he is falſe or mad. But if the ſame
man ſhould deny that any thing exiſts except himself, that is, if
he ſhould affirm the whole creation to be a dream of his own
mind, he would be abſolutely unanſwerable. If I (who know he
is wrong, for I am certain of my own exiſtence) argue with him,
and reduce him to filence, it is no more than might* happen in
his dream. A celebrated metaphysician, Berkeley, maintained
that with regard to matter, the above is the ſtate of the cafe :
that our impreſſions of matter are only impreſſions, communi-
cated by the Creator without any intervening cauſe ofcommuni-
cation.
Our moſt convincing communicable proof of the existence of
other things, is, not the appearance of objects, but the neceffity
of admitting that there are other minds beſides our own. The
external inanimate objects might be creations of our own
thought, or thinking and perceptive function : they are ſo ſome-
times, as in the caſe of infanity, in which the mind has frequently
the appearance of making the whole or part of its own external
world. But when we fee other beings, performing fimilar func-
tions to thoſe which we ourſelves perform, we come ſo irreſiſtibly
to the conclufion that there muſt be other ſentients like ourſelves,
that we ſhould rather compare a perſon who doubted it to one who
denied his own exiſtence, than to one who ſimplydenied the real
external exiſtence of the material world.

* It is not impoſſible that in a real dream of ſleep, ſome one may have
created an antagoniſt who beat him in an argument to prove that he was
awake.
 On Objects, Ideas, and Names. 29
When once we have admitted different and independent
minds, the reality of external objects (external to all thoſe minds)
follows as of courſe. For different minds receive impreſſions at
the ſame time, which their power of communication enables
them to know are ſimilar, ſo far as any impreſſions, one in each
of two different minds, can be known to be ſimilar. There muſt
be afomewhat independent of thoſe minds, which thus acts upon
them all at once, and without any choice of their own. This
Somewhat is what we call an external object : and whether it
ariſe in Berkeley's mode, or in any other, matters nothing to us
here.
We ſhall then, take it for granted that external objects actually
exiſt, independently of the mind which perceives them. And this
brings us to an important diſtinction, which we must carry with
us throughout the whole of this work. Besides the actual exter-
nal object, there is alſo the mind which perceives it, and what
(for want of better words or rather for want of knowing whether
they be good words or not) we must call the image of that object
in the mind, or the idea which it communicates. The termfub-
jett is applied by metaphyſicians to the perceiving mind : and
thus it is ſaid that a thing maybe confideredfubjectively (with re-
ference to what it is in the mind) or objectively (with reference
to what it is independently of any particular mind). But logicians
uſe the word ſubject in another ſenſe. In a propoſition ſuch as
' bread is wholeſome', the thing ſpoken of, ' bread', is called the
ſubject of the propoſition : and in fact the wordſubject is in com-
mon language ſo frequently confounded with object, that it is al-
moſt hopeleſs to ſpeak clearly to beginners about themſelves as
fubjects. I ſhall therefore adopt the words ideal and objective,
idea and object, as being, under explanations, as good as any
others : and better thanfubject and object for a work on logic.
The word idea, as here uſed, does not enter in that vague ſenſe
in which it is generally uſed, as if it were an opinion that might be
right or wrong. It is that which the object gives to the mind,
or the ſtate ofthe mind produced by the object. Thus the idea
of a horſe is the horse in the mind : and we know no other horſe.
We admit that there is an external object, a horſe, which may
give a horse in the mind to twenty different perſons : but no one
of theſe twenty knows the object ; each one only knows his idea.
30 On Objects, Ideas, and Names.
There is an object, becauſe each of the twenty perſons receives
an idea without communicating with the others : ſo that there is
ſomething external to give it them. But when they talk about
it, under the name of a horſe, they talk about their ideas. They
all refer to the object, as being the thing they are talking about,
until the moment theybegin to differ: and then they begin to ſpeak,
not of external horſes, but ofimpreſſions on their minds ; at leaſt
this is the cafe with those who know what knowledge is ; the poſi-
tive and the unthinking part ofthem ſtill talk of the horse. And
the latter have a great advantage * over the former with thoſe
who are like themselves .
Why then do we introduce the term object at all, ſince all our
knowledge lies in ideas ? For the ſame reaſon as we introduce the
term matter into natural philoſophy, when all we know is form,
ſize, colour, weight, &c., no one of which is matter, nor even all
together. It is convenient to have a word for that external
ſource from which ſenſible ideas are produced : and it is just as
convenient to have a word for the external ſource, material or
not, from which any idea is produced. Again, why do we ſpeak
of our power of confidering things either ideally or objectively,
when as we can know nothing but ideas, we can have no right
to ſpeak of any thing else ? The answer is that,just as in other
things, when we speak of an object, we ſpeak of the idea of an.
object. We learn to ſpeak of the external world, becauſe there
are others like ourſelves who evidently draw ideas from the ſame
ſources as ourſelves : hence we come to have the idea of thoſe
ſources, the idea of external objects, as we call them. But we do
not know thoſe ſources ; we know only our ideas of them.
We can even uſe the terms ideal and objective in what may
appear a metaphorical ſenſe. When we speak of ourſelves in the
manner of this chapter, we put ourſelves, as it were, in the pofi-
tion of ſpectators of our own minds : we ſpeak and think of our

* One man afferts a fact on his own knowledge, another afferts his full
conviction of the contrary fact. Both use the evidence of their ſenſes : but
the ſecond knows that full conviction is all that man can have. The firſt
will carry it hollow in a court of justice, in which perſons are conſtantly
compelled to ſwear, not only that they have an impreſſion, but that the im-
preſſion is correct ; that is to ſay, is the impreſſion which mankind in general
would have, and muſt have, and ought to have.
 On Objects, Ideas, and Names. 31
own minds objectively. And it must be remembered that by the
word object, we do not mean material object only. The mind
of another, any one of its thoughts or feelings, any relation of
minds to one another, a treaty of peace, a battle, a diſcuſſion
upon a controverted queſtion, the right of conveying a freehold,
-are all objects, independently of the perſons or things engaged
in them. They are things external to our minds, of which we
have ideas.
An object communicates an idea: but it does not follow that
every idea is communicated by an object. The mind can create
ideas in various ways ; or at least can derive, by combinations
which are not found in external exiſtence, new collections of
ideas. We have a perfectly distinct idea of a unicorn, or a flying
dragon : when we ſay there are no ſuch things, we speak ob-
jectively only : ideally, they have as much exiſtence as a horſe or
a ſheep ; to a herald, more. Add to this, that the mind can
ſeparate ideas into parts, in ſuch manner that the parts alone are
not ideas of any exiſting ſeparate material objects, any more than
the letters of a word are conſtituent parts of the meaning of the
whole. Hence we get what are called qualities and relations. A
ball may be hard and round, or may have hardness and round-
neſs : but we can not ſay that hardness and roundneſs are ſeparate
external material objects, though they are objects the ideas of
which neceſſarily accompany our perception of certain objects.
Theſe ideas are called abſtract as being removed or abſtracted from
the complex idea which gives them : the abſtraction is made by
compariſon or obſervation of reſemblances. If a perſon had never
ſeen any thing round except an apple, he would perhaps never
think of roundneſs as a diſtinct object of thought. When he saw
another round body, which was evidently not an apple, he would
immediately, by perception of the reſemblance, acquire a ſeparate
idea of the thing in which they reſemble one another,
Abſtraction is not performed upon the ideas of material objects
only. For instance, from conduct of one kind, running through
a number of actions, performed by a number of perſons, we get
the ideas of goodneſs, wickedness, talent, courage. But we must
not imagine that we can make ideally external repreſentation of
theſe words. They are objects, that is to ſay, the mind conſiders
them as external to itſelf: but they are not material objects.
32 On Objects, Ideas, and Names.
Some people deny their exiſtence, and look upon them as only
abſtract words, or words under which we ſpeak of minds or
bodies without ſpecifying any more than one of the ideas pro-
duced by theſe minds or bodies. For instance, they affert that
when we ſay ' knowledge gives power ' it is really that perſons
with knowledge are therefore able, or have power, to produce,
or to do, what perſons without it cannot. This is a queſtion
which it does not concern me here to diſcuſs .
Seeing that the mind poſſeſſes a power of originating new
combinations of ideas, and alſo of abſtracting from complex ideas
the more fimple ones of which, it seems natural to ſay, they are
compoſed, it has long been a queſtion among metaphyſicians
whether the mind has any ideas of its own which it poſſeſſes in-
dependently of all ſuggeſtion from external objects. It is not
neceſſary that I ſhould attempt to lead the ſtudent to any con-
clufion* on this ſubject : for our purpoſe, the diſtinction between
ideas and objects, though it were falſe, is of more importance
than that between innate and acquired ideas, though it be true.
But one of theſe two things muſt be true : either we have ideas
which we do not acquire from or by means of communication
with the external world (experience, trial of our ſenſes) or there
is a power in the mind of acquiring a certainty and a generality
which experience alone could not properly give. For instance,
we are ſatisfied as of our own exiſtence that ſeven and three col-
lected are the ſame as five and five, whatever the objects may be
* It has always appeared to me much ſuch a queſtion as the following.
There are hooks which certainly catch fish if put into the water ; and moſt
certainly they have been put into the water. There are then fiſh upon them.
But theſe fiſh might have been on ſome of them when they were put into the
water. It is to no purpoſe to inquire whether it was ſo or not, unleſs there
be ſomediſtinction between the fiſhes which may make it a queſtion whether
fome of them could have been bred in the river into which the hooks were
put. The mind has certainly a power of acquiring and retaining ideas,
which power, when put into communicationwith the external world, it must
exerciſe. There is no mind to experiment upon, except thoſe which have
had fuch communication. Are there found any ideas which we have reaſon
to think could not have been acquired by this communication ? any fiſhes
which could not have come out of the river ? Metaphyſicians ſeem to admit
that if any ideas be innate, they are thoſe of ſpace, time, and of cauſe and
effect : they ſeem alſo to admit, that if there be any ideas, which, not being
innate, are ſure to be acquired, they are theſe very ones.
 On Objects, Ideas, and Names. 33

which are counted: the thing is true of fingers, pebbles, counters,

ſheep, trees, &c. &c. &c. We cannot have aſſured ourſelves of
this by experience : for example, we know it to be true of peb-
bles at the North Pole, though we have never been there ; we
are as fure of it as of our own exiſtence. I do not mean that we
have a rational conviction only, fit to act upon, that it is ſo at the
North Pole, becauſe it is ſo in every place in which it has been
tried : ifwe had nothing elſe, we should have this ; but we feel
that this leſſer conviction is ſwallowed up by a greater. We have
the leſſer conviction that the pebbles at the pole fall to the ground
when they are let go : we are very ſure ofthis, without aſſerting
that it cannot be otherwiſe : we ſee no impoſſibility in thoſe peb-
bles being fuch as always to remain in air wherever they are
placed. * But that ſeven and three are no other than five and
five is a matter which we are prepared to affirm as poſitively of
the pebbles at the North Pole as of our own fingers, both that it
is ſo, and that it must be ſo. Whence ariſes this actual difference
in point of fact, between our mode of viewing and knowing
* Metaphyſicians, in their ſyſtems, have often taken this diſtinction to be
one of ſyſtem only, treating it as a thing to be accepted or rejected with
the ſyſtem, inſtead of an actual and indiſputable phenomenon which re-
quires explanation under any ſyſtem. Dr. Whewell, of all Engliſh writers
on natural ſcience I know, is the one who has made the fact, as a fact, per-
vade his writings, ſometimes attached to a ſyſtem, ſometimes not. The
following remarks on the general ſubject are worth confideration : " It is
indeed, extremely difficult to find, in ſpeaking of this ſubject, expreſſions
which are fatisfactory. The reality of the objects which we perceive is a
profound, apparently an inſoluble problem. We cannot but ſuppoſe that
exiſtence is ſomething different from our knowledge of exiſtence :-that what
exiſts, does not exiſt merely in our knowing that it does :---truth is truth
whether we know it or not. Yet how can we conceive truth, otherwise than
as fomething known ? How can we conceive things as exiſting, without
conceiving them as objects of perception ? Ideas and Things are conſtantly
oppoſed, yet neceſſarily coexiſtent. How they are thus oppoſite and yet iden-
tical, is the ultimate problem of all philoſophy. The ſucceſſive phaſes of
philoſophy have conſiſted in ſeparating and again uniting theſe two oppoſite
elements ; in dwelling ſometimes upon the one and ſometimes upon the other,
as the principal or original or only element ; and then in diſcovering that
fuch an account of the ſtate of the cafe was infufficient. Knowledge requires
ideas. Reality requires things. Ideas and things coexiſt. Truth is, and is
known. But the complete explanation of theſe points appears to be beyond
our reach ."

D
34 On Objects, Ideas, and Names.
different ſpecies of afſſertions ? the truth of the laſt named aſſer-
tion is not born with us, for children are without it, and learn it
by experience, as we know. The must be ſo cannot be acquired
from experience in the common way, for that ſame experience
on which we rely tells us that however often a thing may have
been found true, whatever rule may have been eſtabliſhed by re-
peated inſtances, an exception may at laſt occur. There ſeems
then to be in the mind a power of developing, from the ideas
which experience gives, a real and true diſtinction of neceſſary
and not neceſſary, poſſible and impoſſible. The things which
are without us always confirm our neceffary propofitions : but
how we derive that complete aſſurance that they will do ſo as
faithfully as hitherto they have done ſo, is not within our power
to ſay.
Connected with ideas are the names we give them ; the ſpoken
or written ſounds by which we think of them, and communicate
with others about them. To have an idea, and to make it the
ſubject of thought as an idea, are two perfectly distinct things :
the idea of an idea is not the idea itſelf. I doubt whether we
could have made thought itſelf the ſubject of thought without
language. As it is, we give names to our ideas, meaning by a
name not merely a ſingle word, but any collection of words which
conveys to one mind the idea in another. Thus a-man-in-a-
black-coat-riding-along-the high-road-on-a-bay-horſe is as much
the name of an idea as man, black, or horſe. We can coin
words at pleaſure ; and, were it worth while, might invent a
ſingle word to ſtand for the preceding phraſe.
Names are uſed indifferently, both for the objects which pro-
duce ideas, and for the ideas produced by them. This is a dif-
advantage, and it will frequently be neceſſary to ſpecify whether
we ſpeak ideally or objectively. In common converſation we
ſpeak ideally and think we speak objectively : we take for granted
that our own ideas are fit to paſs to others, and will convey to
them the ſame ideas as the objects themselves would have done.
That this may be the cafe, it is neceſſary firſt, that the object
ſhould really give us the ſame ideas as to others ; ſecondly, that
our words ſhould carry from us to our correſpondents the fame
ideas as thoſe which we intended to expreſs by them. How,
and in what caſes, the firſt or the ſecond condition is not ful
 On Objects, Ideas, and Names. 35
filled, it is impoffible to know or to enumerate. But we have
nothing to do here except to obſerve that we are only incidentally
concerned with this queſtion in a work of logic. We preſume
fixed and, if objective, objectively true ideas, with certain names
attached : ſo that it is never in doubt whether a name be or be not
properly attached to any idea. This method must be followed in
all works of ſcience : a conceivably attainable end is firſt pre-
ſumed to be attained, and the conſequences of its attainment are
ſtudied. Then, afterwards, comes the queſtion whether this end
is always attained, and if not, why. The way to mend bad roads
muſt come at the end, not at the beginning, of a treatise on the
art of making good ones.
Every name has a reference to every idea, either affirmative or
negative. The term horse applies to every thing, either poſitively
or negatively. This (no matter what I am ſpeaking of) either is
or is not a horſe. If there be any doubt about it, either the idea
is not preciſe, or the term horse is ill understood. A name ought
to be like a boundary, which clearly and undeniably either ſhuts
in, or ſhuts out, every idea that can be ſuggeſted. It is the im-
perfection of our minds, our language, and our knowledge of
external things, that this clear and undeniable inclufion or exclu-
fion is feldom attainable, except as to ideas which are well within
the boundary : at and near the boundary itſelf all is vague. There
are decided greens and decided blues : but between the two
colours there are ſhades of which it must be unſettled by uni-
verſal agreement to which of the two colours they belong. To
the eye, green paſſes into blue by imperceptible gradations : our
ſenſes will ſuggeſt no place on which all agree, at which one is
to end and the other to begin.
But the advance of knowledge has a tendency to ſupply means
of preciſe definition. Thus, in the inſtance above cited, Wol-
lafton and Fraunhofer have diſcovered the black lines which al-
ways exiſt in the ſpectrum of folar colours given by a glaſs priſm,
in the ſame relative places. There are definite places in the ſpec-
trum,by the help of which the place ofany ſhade ofcolour therein
exifting may be aſcertained, and means of definition given.
When a name is complex, it frequently admits of definition,

* It is quite within the poſſibilities of the application of ſcience to the

36 On Objects, Ideas, and Names.
nominal or real. A name may be ſaid to be defined nominally
when we can of right ſubſtitute for it other terms. In ſuch a
caſe, a perſon may be made to know the meaning of the word
without acceſs to the object of which it is to give the idea. Thus,
an island is completely defined in ' land ſurrounded by water.'
In definition, we do not mean that we are neceſſarily to have
very preciſe terms in which to explain the name defined : but, as
the terms of the definition ſo is the name which is defined ; ac-
cording as the firſt are preciſe or vague, clear or obſcure, ſo is the
ſecond. Thus there may be a queſtion as to the meaning of
land: is a marſh ſticking up out of the water an iſland ? Some
will ſay that, as oppoſed to water, a marſh is land, others may
confider marſh as intermediate between what is commonly called
[dry] land and water. If there be any vagueneſs, the term iſland
muſt partake of it : for iſland is but ſhort for ' land ſurrounded by
water,' whether this phrase be vague or precife. This fort of de-
finition is nominal, being the ſubſtitution of names for names. It
is complete, for it gives all that the name is to mean. An iſland,
as ſuch, can have nothing neceſſarily belonging to it except what
neceſſarily belongs to ' land furrounded by water. By real de-
finition, I mean ſuch an explanation of the word, be it the whole
of the meaning or only part, as will be ſufficient to ſeparate the
things contained under that word from all others. Thus the
following, I believe, is a complete definition of an elephant ; ' an
animal which naturally drinks by drawing the water into its noſe,
and then ſpirting it into its mouth. As it happens, the animal
which does this is the elephant only, of all which are known upon
the earth : ſo long as this is the caſe, ſo long the above definition
anſwers every purpoſe ; but it is far from involving all the ideas
which ariſe from the word. Neither ſagacity, nor utility, nor the
production of ivory, are neceſſarily connected with drinking by
help of the noſe. And this definition is purely objective ; we do
not mean that every idea we could form of an animal ſo drinking
is to be called an elephant. If a new animal were to be diſco-
vered, having the ſame mode of drinking, it would be a matter
of pure choice whether it ſhould be called elephant or not. It

arts that the time ſhould come when the ſpectrum, and the lines in it, will
be uſed for matching colours in every linen-draper's ſhop .
 On Objects, Ideas, and Names. 37
muſt then be ſettled whether it ſhall be called an elephant, and
that race of animals ſhall be divided into two ſpecies, with diſtinc-
tive definitions ; or whether it ſhall have another name, and the
definition above given ſhall be incomplete, as not ſerving to draw
an entire diſtinction between the elephant and all other things.
It will be obſerved that the nominal definition includes the real,
as ſoon as the terms of ſubſtitution are really defined : while the
real definition may fall ſhort of the nominal.
When a name is clearly understood, by which we mean when
of every object of thought we can distinctly ſay, this name does
or does not, contain that object-we have said that the name ap-
plies to everything, in one way or the other. The word man
has an application both to Alexander and Bucephalus : the firſt
was a man, the ſecond was not. In the formation of language,
a great many names are, as to their original fignification, of a
purely negative character : thus, parallels are only lines which do
not meet, aliens are men who are not Britons (that is, in our
country). If language were as perfect and as copious as we
could imagine it to be, we ſhould have, for every name which has
a poſitive ſignification, another which merely implies all other
things : thus, as we have a name for a tree, we should have an-
other to fignify every thing that is not a tree. As it is, we have
ſometimes a name for the poſitive, and none for the negative, as
in tree : ſometimes for the negative and none for the poſitive, as
in parallels : ſometimes for both, as in a frequent uſe of perfon
and thing. In logic, it is desirable to confider names of inclufion
with the correſponding names of exclufion : and this I intend to
do to a much greater extent than is uſual : inventing names of
excluſion by the prefix not, as in tree and not-tree, man and not-
man.
Let theſe be called contrary, or contradictory, names.
Let us take a pair of contrary names, as man and not-man.
It is plain that between them they repreſent everything imaginable
or real, in the univerſe. But the contraries of common language
uſually embrace, not the whole univerſe, but ſome one gene-
ral idea. Thus, of men, Briton and alien are contraries : every
man muſt be one of the two, no man can be both. Not-Briton
and alien are identical names, and ſo are not-alien and Briton.
* I intend to draw no diſtinction between theſe words ,
38 On Objects, Ideas, and Names.
The fame may be ſaid of integer and fraction among numbers,
peer and commoner among ſubjects of the realm, male and fe-
male among animals, and ſo on. In order to expreſs this, let us
ſay that the whole idea under confideration is the universe (mean-
ing merely the whole of which we are confidering parts) and let
names which have nothing in common, but which between them
contain the whole idea under confideration, be called contraries
in, or with respect to, that universe. Thus, the univerſe being
mankind, Briton and alien are contraries, as are ſoldier and civi-
lian, male and female, &c.: the univerſe being animal, man and
brute are contraries, &c .
Names may be repreſented by the letters ofthe alphabet : thus
A, B, &c., may fſtand for any names we are confidering, fimple or
complex. The contraries may be repreſented by not-A, not-B,
&c. , but I shall usually prefer to denote them by the ſmall letters
a, b, &c. Thus, everything in the univerſe (whatever that uni-
verſe may embrace) is either A or not-A, either A or a, either
Bor b, &c. Nothing can be both B and b ; every not-B is b,
and every not-b is B : and ſo on.
No language, as may well be ſuppoſed, has been conſtructed
beforehand with any intention of providing for the wants of any
metaphyfical ſyſtem. In moſt, it is ſeen that the neceffity of
providing for the formation of contrary terms has been obeyed.
Our own language has borrowed from the Latin as well as from
its parent : thus we have imperfect, disagreeable, as well as un-
formed and witless. There is a choice of contraries without very
well ſettled modes of appropriation : ſtanding for different de-
grees of contrariety. Thus we have not perfect which is not fo
ſtrong a term as imperfect ; and not imperfect, the contrary of a
contrary, which is not ſo ſtrong as perfect. The wants of com-
mon converſation have ſometimes retained a term and allowed
the contrary to fink into diſuſe ; ſometimes retained the contrary
and neglected the original term ; ſometimes have even introduced
the contrary without introducing any term for the original no-
tion, and allowed no means of expreſſing the original notion
except as the contrary of a contrary. If we could imagine a
perfect language, we ſhould ſuppoſe it would contain a mode of
fignifying the contrary of every name : this indeed our own lan-
guage maybe faid to have, though ſometimes in an awkward and
 On Objects, Ideas, and Names. 39
unidiomatic manner. One inflexion, or one additional word,
may ſerve to ſignify a contrary of any kind: thus not man is
effective to denote all that is other than man. But there is a
wider want, which can only be partially ſupplied, for its complete
fatisfaction would require words almoſt beyond the power of
arithmetic to count : and all that has been done to make it leſs
confifts, in our language and in every other, mostly in the forma-
tion ofcompound terms, be they ſubſtantive and adjective, dou-
ble ſubſtantives, or any others. A claſs of objects has a ſub-claſs
contained within it, the individuals of which are diftinguiſhed
from all others of the claſs by ſomething common to them and
them only. If the distinguiſhing characteriſtic have been ſepa-
rated, and a word formed to fignify the abſtract idea, that word,
or an adjective formed from it (if it be not an adjective) is joined
with the general name ofthe claſs. Thus we have ſtrong men,
white horſes , &c. Or it may happen that the individuals of the
ſub-claſs take, in right of the distinguiſhing characteriſtic, a per-
fectly new name, and by the moſt varied rules. A corn-grinding
man is called from the implement he uſes, a miller ; a meat-
killing man from the organ which he ſupplies, a butcher, (if the
firſt idea of the etymology of this word be correct). Other men
uſe mills and other trades feed the mouth : ſtill cuſtom has ſet-
tled theſe terms, though the firſt is only connected with its origin
by the ſpelling, and the ſecond by a derivation which must be
fought in another language. But again, it will more often hap-
pen that a diftinctive characteriſtic, belonging to ſome only, gives
no diſtinctive name to thoſeſome, which ſtill remain an unnamed
Some out of the whole, to be ſeparated by the deſcription of their
characteriſtic when wanted, instead of being the all of a name
invented to expreſs them, and them alone of their claſs. In ſuch
a predicament, for instance, are men who have never ſeen the
ſea, as diftinguiſhed from those who have seen it. Hence it ap-
pears that particular propoſitions are not ſo distinct from uni-
verſal ones in real character as they are generally made to be.
If I ſay ſome As are Bs ' the reader may well ſuppoſe that it is
not often neceſſary to advert to this fact : had it been ſo, a name
would have been invented ſpecially to fignify ' As which are Bs .'
Ifthis name had been C, the propofition would have been ' every
C is B.'
40 On Objects, Ideas, and Names.
The ſame convenience which dictates the formation of a name
for one fub-claſs and not for another, rules in the formation of
contrary terms, as already noted. And theſe caprices oflanguage
-for logically confidered they are nothing elſe, though their for-
mation is far from lawless-make it deſirable to include in a for-
mal treatiſe the moſt complete confideration of all propoſitions,
with reference not only to their terms, but alſo to the contraries
of thoſe terms. Every negative propoſition is affirmative, and
every affirmative is negative. Whatever completely does one of
the two, include or exclude, alſo does the other. If I say that
،

no A is B,' then, b being the name of every thing not B in the

univerſe of the propoſition, I ſay that ' every A is b : ' and if I
ſay that every A is B,' I ſay that ' no A is b .' Whether a lan-
guage will happen to poſſeſs the name B, or b, or both, depends
on circumstances of which logical preference is never one, ex-
cept in treatiſes of ſcience. The Engliſh may poſſeſs a term for
B, the French only for b : ſo that the ſame idea muſt be preſented
in an affirmative form to an Engliſhman, as in ' every A is B,'
and in a negative one to a Frenchman, as ' no A is b.' From
all this it follows that it is an accident of language whether a pro-
poſition is univerſal or particular, poſitive or negative. We,
having the names A and B, may be able to ſay ' every A is B : '
another language, which only names the contrary of B, muſt ſay
' no A is b. A third language, in which As have not a ſeparate
name, but are only individuals of the claſs C, muſt ſay ' ſome Cs
are Bs ; ' while a fourth, which is in the further predicament of
naming only b, must have it ' ſome Cs are not bs.' When we
come to confider the ſyllogiſm, we shall have full confirmation of
theItcorrectneſs and completeneſs of this view.
may be objected that the introduction of terms which are
merely negations of the poſitive ideas contained in other terms
is a ſpecies of fiction. I anſwer, that, firſt, the fiction, if it be a
fiction, exists in language, and produces its effects : nor will it
eaſily be proved more fictitious than the invention of ſounds to
ſtand for things. But, ſecondly, there is a much more effective
anſwer, which will require a little development.
When writers on logic, up to the preſent time, uſe ſuch con-
traries as man and not-man, they mean by the alternative, man
and everything elſe. There can be little effective meaning, and
 On Objects, Ideas, and Names. 41

no uſe, in a claſſification which, becauſe they are not men, in-

cludes in one word, not-man, a planet and a pin, a rock and a
featherbed, bodies and ideas, wiſhes and things wiſhed for. But
if we remember that in many, perhaps moſt, propofitions, the
range of thought is much leſs extenſive than the whole univerſe,
commonly ſo called, we begin to find that the whole extent of a
ſubject of diſcuſſion is, for the purpoſe ofdiſcuſſion, what I have
called a univerſe, that is to ſay, a range of ideas which is either
expreſſed or understood as containing the whole matter under
confideration. In ſuch univerſes, contraries are very common :
that is, terms each of which excludes every caſe of the other,
while both together contain the whole. And, it must be ob-
ſerved that the contraries of a limited univerſe, though it be a
ſufficient real definition of either that it is not the other, are fre-
quently both of them the objects from which poſitive ideas are
obtained. Thus, in the univerſe of property, perſonal and real
are contraries, and a definition of either is a definition of the
other. But though each be a negative term as compared with
the other, no one will ſay that the idea conveyed by either is that
of a mere negation. Money is not land, but it is ſomething. And
even when the contrary term is originally invented merely as a
negation, it may and does acquire poſitive properties. Thus
alien is strictly not-Briton : but ſuppoſe a man taken in arms
against the crown on ſome ſpot within its dominions, and claim-
ing to be a priſoner of war. The answer that he is a Britiſh ſub-
ject is a negation : to eſtabliſh his poſitive claim he firſt muſt
prove himself an alien, and moreover that he is in another poſitive
predicament, namely, that he is the ſubject of a power at war
with Great Britain. Accordingly, of two contraries, neither
muſt be confidered as only the negation ofthe other : except when
the univerſe in queſtion is ſo wide, and the poſitive term ſo li-
mited, that the things contained under the contrary name have
nothing but the negative quality in common.
Perception of agreements and diſagreements is the foundation
of all affertion : the acquirement of ſuch perception with reſpect
to any two ideas by the comparison of both with a third, is the
proceſs of all inference. To infer, by comparison of abſtract ideas,
is the peculiar privilege of man ; to need inference is his imper-
fection. To what point man would carry inference if he wanted
 42 On Objects, Ideas, and Names.
language, how much further the lower animals could carry what
they have of it if they had language, are queſtions on which it is
vain to ſpeculate. The words is and is not, which imply the
agreement or diſagreement of two ideas, muſt exiſt, explicitly or
implicitly, in every aſſertion. And what we call agreement or
diſagreement, may be reduced to identity or non-identity. When
we ſay John is a man, we have the firſt and moſt objective form of
aſſertion. Looked at in the moſt objective point of view it is
only this, John is one of the individual objects who are called
man.
Looked at ideally, the propofition is more general. The
idea of man, gathered from inſtances, preſents itſelfas a collective
maſs of ideas, ofwhich we can figure to ourſelves an inſtance
without neceſſarily calling up the idea of any man that ever ex-
iſted. In the ideal conception of man, Achilles is a man as
much as the Duke of Wellington, whether the former ever ex-
ifted objectively or no : of all the ideas of man which the mind
can imagine, the former is one as well as the latter.
The ſeparation of ideas, or formation of abſtract ideas, and
aſſertion by means of them, preſents nothing, for our purpoſe,
which differs from the former cafe. If we ſay ' this picture is
beautiful, the mere phraſe is incomplete, for ' beautiful' is only an
attribute, a purely ideal reference to a claſſification which the
mind makes, dictated by its own judgment. The picture being
a material object, cannot be anything but an object, cannot be-
long to any claſs of notions, unleſs that claſs contain objects.
What the propoſition may mean is to a certain extent dependent
upon the implied ſubſtantive to which beautiful belongs : that is,
to the claſs of objects which the propoſition implies the mind to
have ſeparated into beautiful and not beautiful. It may be that
the picture is a beautiful picture : or a beautiful work of art, tak-
ing its place in that diviſion by which not only pictures, but ſta-
tues, buildings, reliefs, &c. are ſeparated into beautiful and other-
wife : or a beautiful creation of human thought, placed among
works of art, imagination, or ſcience, &c. in the ſubdiviſion beau-
tiful : or finally, it may be a beautiful thing, placed with all ob-
jects of perception in a fimilar ſubdiviſion.
In all affertions, however, it is to be noted, once for all, that
formal logic, the object of this treatiſe, deals with names and not
with either the ideas or things to which theſe names belong. We
 On Objects, Ideas, and Names. 43

are concerned with the properties of ' A is B ' and ' A is not B '
ſo far as they preſent an idea independently of any ſpecification
of what A and B mean : with ſuch ideas upon propoſitions as
are preſented by theirforms, and are common to all forms of the
fame kind. The reality of logic is the examination of the use of
is and is not : the tracing ofthe conſequences of the application
of theſe words. The argument ' when the fun ſhines it is day :
but it is not day, therefore the fun does not ſhine,' contains a
theory and two facts, the latter of which is made to follow from
the former by the theory. That inference is made is ſeen in the
word therefore : and the ſentence is capable of being put upon its
trial for truth or falſehood by logical examination. But this exa-
mination rejects the meaning of fun and day, the truth of the
theory and of the facts ; and only inquires into the right which
the ſentence, of its own ſtructure, gives us to introduce the word
therefore. It merely enters upon ' when A is, Bis ; but Bis
not ; therefore A is not : ' and decides that this is a correct junc-
tion of precedents and conſequent, an exhibition ofneceſſary con-
nexion between what goes before and after therefore, and a de-
velopment, in the latter, ofwhat is virtually, though not actually,
expreſſed in the former. What A and B may mean is of no
conſequence to the inference, or right to bring in ' A is not.'
Thus A and B, diveſted of all ſpecific meaning, are really
names as names, independently of things : or at leaſt may be fo
confidered. For the truth of the propoſition, under all mean-
ings, gives us a right to ſuppoſe, if we like, that names are the
meanings that is to ſay, that we may put it thus, ' When the
name A is, the name B is : but the name B is not; therefore the
name A is not.
It is not therefore the object of logic to determine whether
conclufions be true or falſe ; but whether what are aſſerted to
be concluſions are conclufions. By a concluſion is meant that which
is and must be ſhut in with certain other preceding things put in
firſt : it is that which must have been put into a ſentence becauſe
certain other things were put in. To infer a concluſion is to bring
in, as it were, the direct ſtatement of that which has been virtu-
ally ſtated already has been shut in. When we ſay ' A is B,
Bis C ; we conclude A is C' ; it would be more correct to ſay
A is B, B is C ; we have concluded A is C ' . We ſhould never
44 On Objects, Ideas, and Names.
think of faying ' we have put into a box a man's upper dreſs of
the colour of the trees ; therefore we must put in a green coat ' ;
-we ſhould say ' we have put in.' To infer the concluſion then
is to bring in a ſtatement that we have concluded.
Inference does not give us more than there was before : but it
may make us fee more than we saw before : ideally ſpeaking,
then, it does give us (in the mind) more than there was before.
But the homely truth that no more can come out than was in,
though accepted as to all material objects even by metaphyſicians
-who are generally well pleaſed to find the key of a box which
contains what they want, though ſure that it will put in no more
than was there already has been applied to logic, and even to
mathematics, in depreciation of their rank as branches of know-
ledge. Those who have made this ſtrangeſt of human errors
must have aſſumed an ideal omniſcience, and looked at human
imperfection objectively. Omniſcience need neither compare
ideas, nor draw inferences : the concluſion which we deduce
from premiſes, is always preſent with them ; truths are concomi-
tants, not confequences. When we say that one aſſertion follows
from another, we ſpeak purely ideally, and deſcribe an imperfec-
tion of our own minds : it is not that the conſequence follows
from the premiſes, but that our perception of the conſequence fol-
lows our perception of the premiſes : the conſequence, objectively
ſpeaking, is in, and with, and of, the premiſes. We ſpeakwrongly
ifwe ſpeak ideally, when we say that ' A is C,' is in ' A is B and
B is C' : in fact, it is only by giving an objective view to the
argument, that we can even aſſert that it will be ſeen. To un-
cultivated minds, this ſimple conclufion is never concomitant
with the premiſes, and only with ſome difficulty a conſequence.
From the certainty that a conſequence may be made to come
out, which is an allegorical uſe of the word out, we aſſume a right
to declare, by the ſame ſort of allegory,* that it was in. The
premiſes therefore contain the conclufion : and hence ſome have
ſpoken as if in ſtudying how to draw the conclufion, we were
ſtudying to know what we knew before. All the propofitions
of pure geometry, which multiply ſo faſt that it is only a ſmall
* I am of opinion that it is more conſiſtent with analogy to ſay that the
hypothefis is contained in its neceſſary conſequence, than to ſaythat the for-
mer contains the latter. My reaſon will appear in the courſe of the work.
 On Objects, Ideas, and Names. 45
and iſolated claſs even among mathematicians who know all that
has been done in that ſcience, are certainly contained in, that is
neceſſarily deducible from, a very few ſimple notions. But to be
knownfrom theſe premiſes is very different from being known with
them.
Another form of the aſſertion is that conſequences are virtually
contained in the premiſes, or (I ſuppoſe) as good as contained in
the premiſes. Perſons not ſpoiled by ſophiſtry will smile when
they are told that knowing two ſtraight lines cannot encloſe a
ſpace, the whole is greater than its part, &c. they as good as
knew that the three interſections of oppoſite ſides of a hexagon
inſcribed in a circle must be in the ſame ſtraight line. Many
of my readers will learn this now for the first time : it will com-
fort them much to be aſſured, on many high authorities, that they
virtually knew it ever ſince their childhood. They can now pon-
der upon the distinction, as to the ſtate of their own minds, be-
tween virtual knowledge and abſolute ignorance.
There muſt always be ſome contention as to the relative value
of their knowledge between the ſtudents of the things which we
can ſee must have been, and of the things which, for what
we can fee, might have been otherwise. How much of the dif-
tinction is due to our ignorance, no one can tell. In the mean
time, it is of more uſe to point out the advantage, as things are,
of ſtudying both kinds of knowledge, than to attempt to inſtitute
a rivalry between them. Those who have undervalued the ſtudy
of neceſſary conſequences, have allowed themselves, in illuſtrat-
ing their argument, phrases * which taken literally, mean more
perhaps than they intended.
* We might ſometimes take them to mean that the ſtudy of neceſſary
connexion in logic, mathematics, &c., is at leaſt uſeleſs, if not pernicious.
Now we should ſuppoſe, if this be what they mean, that cloſe connexion,
ſhort of abſolute neceſſity, muſt partake ſomewhat of the fame character. If
the abſolute mathematical neceſſity that three angles of a triangle are equal
to two right angles is therefore to be avoided, the ſtudy ofphyſics, in which
there are the neceffities which we expreſs by the term laws ofnature, muſt
do ſome harm. Hiſtory, in which we may ſo often count upon the actions
which motives will produce, cannot be quite faultleſs : and there are laws of
formation in language which might as well be kept out of fight, for they
act almoſt with the uniformity of laws of nature. True knowledge muſt
conſiſt in the ſtudy of the actions of madmen : that a certain man imagined
46 On Objects, Ideas, and Names.
The study oflogic, then, confidered relatively to human know-
ledge, ſtands in as low a place as that of the humble rules of
arithmetic, with reference to the vaſt extent of mathematics and
their phyſical applications. Neither is the leſs important for its
lowlineſs : but it is not every one who can fee that. Writers on
the ſubject frequently take a ſcope which entitles them to claim
for logic one of the highest places : they do not confine them-
ſelves to the connexion of premiſes and conclufion, but enter
upon the periculum et commodum of the formation of the premiſes
themselves. In the hands of Mr. Mill, for example (and to fome
extent in thoſe of Dr. Whateley) logic is the ſcience of diftin-
guiſhing truth from falſehood, ſo as both to judge the premiſes
and draw the conclufion, to compare name with name, not only
as to identity or difference, but in all the varied aſſociations of
thought which arife out of this compariſon.

CHAPTER III .

On the Abstract Form of the Propofition.

N the preceding chapter, I have endeavoured to put together
ſuch notions on the actual ſources of our knowledge as may
give the reader the means of thinking upon points which any
ſyſtem of logic, however restricted, muſt neceſſarily ſuggeſt.
We cannot attempt to connect our use of words with our notions
ofthings, without the occurrence of a great many difficulties, a
great many ſources of adverſe theories, and of never-ending dif-
putes. We cannot even repreſent phænomena, as phænomena,
except in the language of ſome ſyſtem, and it may be of a wrong
one. The confidence which the favourers of theſe ſeveral the-
ories place in their correctneſs is a ſufficient reaſon for keeping
the account of the process of the understanding, ſo far as it can
be made an exact ſcience, as diſtinct as poſſible from all of them :
for they differ widely, and if they agree in anything which can
himſelf to be Cæfar, when he might juſt as well have been Newton or Ne-
buchadnezzar, muſt be a real bit of knowledge, not virtually contained in
anything else, wholly or partially.
 On the Abstract Form of the Propofition. 47
be diſtinctly apprehended, it is only in having names of great
authority enrolled among the partiſans of every one.
In order to examine the laws of inference, of the way of dif-
tinctly perceiving the right to ſay ' therefore,' ' ſo that,' ' whence
it must be,' &c., &c., in a manner which may be admitted, ſo
far as we go, by all, we must make this ſeparation very complete.
All admit propofitions, as 'man is animal, ' no man is faultleſs ;'
all are, after a little thought, agreed upon the modes of inference :
but upon the import of a ſimple propofition, there is every kind
of difference. How much we mean, when we ſay ' man is ani-
mal,' and how we arrive at our meaning, is matter for volumes
on different fides of unſettled queſtions.
In order properly to examine the laws of inference, or of any
thing elſe, we must firſt endeavour to arrive at a diſtinct abſtrac-
tion of ſo much ofthe idea we are concerned with, as is itſelfthe
precedent reaſon, if it be right ſo to ſpeak, of the law in queſtion.
This is an eaſy proceſs upon familiar things. We do not give
the carriers of goods much credit for profundity, in ſeeing that,
on a given road, there is only the difference of weight by which
they are concerned to know how one parcel differs from another ;
and further that, as long as they have to carry a pound, it matters
nothing whether it be of ſugar or iron. It is this proceſs which
we want to perform to the utmoſt, upon the ſimple propofition.
Writers on logic, from Aristotle downwards, have made a large
and important ſtep in ſubſtituting for ſpecific names, with all their
ſuggeſtions about them, the mere letters of the alphabet, A, B,
C, &c. Theſe letters areſymbols, and general ſymbols : each of
them ſtands for any one we pleaſe of its claſs. But what are
they ſymbols of, names, ideas,* or the objects which give thoſe
ideas ? The answer is, that this is preciſely one of thoſe confi-
derations which we may leave behind, in abſtracting what is
neceſſary to an examination of the laws of inference. The only
condition is, that we are to confine ourſelves to one or the other.
When we ſay man is animal, it may be that the name man is
contained in the name animal, that the idea of man is contained
in that of animal, or that the object man is in the object animal.
Or if there were twenty more different appropriations of the
* Meaning of courſe (page 30) ideas of ideas, and ideas of objects.
48 On the Abstract Form
ſymbols, the fame thing might be ſaid of each. This is, I believe,
the firſt uſe of the general ſymbol in order of time ; the algebrai-
cal uſe of letter or other ſymbol, to deſignate number, being both
ſubſequent and derived .
When therefore we say ' Every X is Y', we underſtand that
X is a ſymbol which represents an inſtance of a name, idea, ob-
ject, &c., as the cafe may be. There may be more or fewer of
ſuch inſtances ; they may be numerable or innumerable. And the
ſame of Y. The language of logicians has generally been unfa-
vourable to the distinct perception of their terms being diſtribu-
tively applicable to claſſes of inſtances. They have rather been
quantitative than quantuplicitative: expreſſing themſelves as if, in
ſaying that animal is a larger or wider term than man, they would
rather draw their language from the idea of two areas, one of
which is larger than the other, than from two collections of indi-
viſible units, one of which is in number more than the other.
They have even carried this ſo far as to make it doubtful, except
from context, whether their distinction between univerſal and par-
ticular is that of all andfome, or of the whole and part. Iftheir
inſtances had been whiteſquares, their ' all A is B' and ' fome A
is B' might have applied as well to ' All the ſquare is white' and
'Some of the ſquare is white' as to ' All the ſquares are white' and
' Some of the ſquares are white.' I ſhall take particular care
to uſe numerical language, as diftinguiſhed from magnitudinal,
throughout this work, introducing of courſe, the plurals Xs, Ys,
Zs, &c.
I may mention here another mode of ſpeaking, which will, I
think, appear objectionable to all who are much uſed to confi-
deration of quantity. When a compound idea contains two or
more ſimpler ones, ſome logicians have ſpoken as if the com-
bination were legitimately repreſented by arithmetical addition .
Thus the combination of the ideas of animal and rational muſt
give the idea of man : for the two notions co-exiſt in nothing
elſe that we know of. Accordingly, ſome write animal + ratio-
nal= man . Ifthis be intended as an abſtraction of the notation of
arithmetic, for the purpoſe of fitting to it entirely different mean-
ing, there is of courſe no objection which I need conſider here :
but it ſeems to me that more is meant, and that those who have
uſed this notation imagine a great reſemblance between combining
 ofthe Proposition. 49
ideas, and cumulating them. What the difference is, I cannot
pretend to ſay, any more than I can pretend to say what the dif-
ference is between chemically combining volumes of oxygen and
hydrogen, ſo as to produce water, and ſimple cumulation ofthem
in the fame veſſel, ſo as to produce a mixed gas : every beginner
knows that the electric ſpark, or ſome other inexplicable agency,
is neceſſary to turn the mixed gas into a new chemical combina-
tion. But that the difference exiſts in the former caſe alſo, ſeems
to me as clear as any thing I can imagine. Even in chemiſtry the
cumulative notation,whichwas once thought an all-ſufficient mode
of expreffing the reſults of the atomic theory, has failed with the
progreſs ofknowledge. To a confiderable extent, the introduc-
tion of modes of cumulation as yet anſwers the purpoſe : but
there ſtill remain isomeric compounds, differing in properties, but
of the ſame compoſition. For example, the tartaric and racemic
acids : of which Profeſſor Graham ſays (Elements of Chemistry p.
158), " A nearer approach to identity could ſcarcely be con-
ceived than is exhibited by theſe bodies, which are, indeed, the
ſame both in form and compoſition. But by no treat-
ment can the one acid be tranſmuted into the other." If the
above mode of confounding cumulation and combination be ad-
miſſible, I ſuppoſe we might easily give ourſelves a right to ſay that
2 + 2 + addition = 4
an equation at which the mathematician would ſtare.
So much for the characteriſtics of the terms of a propoſition,
as wanted for the abſtract forms of inference. It remains to
confider thoſe of the connecting copulæ is and is not.
The complete attempt to deal with the term is would go to
the form and matter of every thing in existence, at least, if not to
the poſſible form and matter of all that does not exiſt, but might.
As far as it could be done, it would give the grand Cyclopædia,
and its yearly ſupplement would be the history of the humanrace
for the time. That logic exiſts as a treated ſcience, ariſes from
the characteriſtics of the word, requiſite to be abſtracted in ſtudy-
ing inference, being few and easily apprehended. It may be uſed
in many fenfes, all having a common property. Names, ideas,
and objects, require it in three different ſenſes. Speak of names,
and ſay ' man is animal' : the is is here an is of applicability ; to
E
50 On the Abstract Form
whatsoever (idea, object, &c.) man is a name to be applied, to
that ſame (idea, object, &c.) animal is a name to be applied. As
to ideas, the is is an is of poſſeſſion of all eſſential characteriſtics ;
man is an idea which poſſeſſes, contains, preſents, all that is con-
ſtitutive of the idea animal. As to abſolute external objects, the
is is an is of identity, the most common and poſitive uſe of the
word. Every man is one of the animals ; touch him, you touch
an animal, destroy him, you destroy an animal.
Theſe ſenſes are not all interchangeable. Take the is of iden-
tity, and the name man is not, as a name, the name animal : the
idea man is not, as an idea, the idea animal. Now we muſt aſk,
what common property is poſſeſſed by each of theſe three notions
of is, on which the common laws of inference depend. Common
laws of inference there certainly are. If the applicability of the
name A be always accompanied by that of B, and that of B by
that of C, then that of A is always accompanied by that of C. If
the idea A contain all that is eſſential to the idea B, and B all that
is eſſential to C, then A contains all that is eſſential to C. If the
object A be actually the object B, and if B be actually C, then
A is actually C.
The following are the characteriſtics of the word is which, ex-
ifting in any propoſed meaning of it, make that meaning fatisfy
the requirements of logicians when they lay down the propofition
' A is B.' To make the ſtatement diſtinct, let the propofition
be doubly fingular, or refer to one inſtance of each, one A and
one B : let it be ' this one A is this one B.'
First, the double fingular propoſition above mentioned, and
every ſuch double-fingular, must be indifferent to converfion : the
' A is B,' and the ' B is A' must have the fame meaning, and be
both true or both falſe .
Secondly, the connexion is, exiſting between one term and
each of two others, muſt therefore exiſt between thoſe two
others ; ſo ' A is B' and ' A is C' must give ' B is C.'
Thirdly, the eſſential diſtinction of the term is not is merely
that is and is not are contradictory alternatives, one must, both
cannot, be true.
Every connexion which can be invented and ſignified by the
terms is and is not, ſo as to ſatisfy theſe three conditions, makes
all the rules of logic true. No doubt abſolute identity was the ſug-
 of the Propofition. 51
geſting connexion from which all the others aroſe : juſt as arith-
metic was the medium in which the forms and laws of algebra
were ſuggeſted. But, as now we invent algebras by abſtract-
ing the forms and laws of operation, and fitting new meanings to
them, ſo we have power to invent new meanings for all the
forms of inference, in every way in which we have power to
make meanings of is and is not which fatisfy the above condi-
tions. For instance, let X, Y, Z, each be the ſymbol attached
to every inſtance of a claſs of material objects, let is placed be-
tween two, as in ' X is Y' mean that the two are tied together,
ſay by a cord, and let X be confidered as tied to Z when it is
tied to Y which is tied to Z, &c. There is no fyllogifm but
what remains true under theſe meanings. Thus
The fyllogifm Is true in the ſenſe
Every X is Y Every X is tied to a Y
Some Zs are not Ys Some Zs are not tied to Ys
... Some Zs are not Xs ... Some Zs are not tied to Xs

This laſt inſtance might be confidered as a material reprefen-

tation of attachment together of ideas in the mind.
We must distinctly obſerve that it is not every caſe of infe-
rence which demands all the characteriſtics to be fatisfied. Thus
in the most common caſe of all, ' Every A is B, every B is C,
therefore every A is C,' of all the three conditions only the ſe-
cond is wanted to ſecure the validity ofthis caſe. Though it be fel-
dom thought worth while to make this obſervation, yet it is uni-
verſal practice to act upon it, and ſo as to introduce into formal
logic apparent contradictions of its own rules. For example,th e
following are allowed to paſs for fyllogifms, in the ordinary defi-
nition of that word .
' Every A is greater than ſome one B ; every B is greater than
ſome one C, therefore every A is greater than ſome one C. '
And the ſame when inſtead of greater than is read equal to or
less than. The form which most commonly appears is the
pair of doubly ſingular propofitions, ' A (one thing) is greater
than B ; B is greater than C ; therefore A is greater than C.'
Here ' greater than greater' is ' greater,' the ſecond rule is fatisfied,
and no other is wanted. But this meaning for is (or this ſubſti-
tute for it, if the reader like it better) will not fatisfy all the con
52 On the Abstract Form
ditions, and therefore will not apply to all the forms of inference.
But is in the ſenſe ' is equal to' does fatisfy all the conditions.
This ſenſe of is, namely agreement in magnitude, is the copula of
the mathematician's ſyllogiſm, when he is reaſoning on quantity
only.
It will probably be affirmed that the generalization thus made,
or ſhown to be poſſible, in the conception of the word is for
purpoſes of inference, amounts only to a very frequent, if not
moſt uſual, uſe of the word, namely, as ſignifying a certain mode,
not of identity, but of agreement in quality. As when we fay
' theſe two things are the ſame-in colour' or ' the one thing is
the other-in colour :' that the name man is the name animal,
in a certain reſpect, namely, in what the latter can be applied to :
that the idea man is animal, in both poſſeſſing certain charac-
teriſtics : that every object man is an object animal, in actual
ſubſtance : that A is B in magnitude, when we ſay A equals
B; and ſo on. But I admit only the converſe, namely, that all
theſe uſes fatisfy the conditions. It would hardly be for any one
to ſay, that every poſſible uſe of is which ſatisfies three ſuch fim-
ple requirements, has been or can be exhausted. Even the
material example which was just now given, cannot be iden-
tified with any common uſe, or eaſily imaginable one, of the
common verb. But if no invented meaning, proper to fatisfy
the conditions, can be found, other than already exiſts in more
or leſs of uſe, ſtill, theſe conditions are the laws to which the
word muſt ſubmit in its logical acceptation.
There are common uſes of the word which are not admitted
in logic : and among them, one of the most common, connection
of an object with its quality, and of an idea with one of its con-
ſtituent or aſſociated ideas. As when we ſay, the roſe is red,
prudence is defirable. Here the logical conditions are not fatis-
fied. For example, ' red is the roſe,' though a poetical inverſion
of the firſt aſſertion, is not logically true. It is uſual to confider
ſuch propoſitions, in logic, as elliptical ; thus ' the roſe is red' is
confidered as ' the roſe is a red object, or an object of red colour ;'
in which the is now takes one of the ſenſes which allows of con-
verſion. Similarly, in all other caſes, the ſubject and predicate
are made to take the ſame character ; both names, both ideas, or
both objects. This reduction renders unneceſſary both the ſtudy
 ofthe Propofition. 53
of the varieties ofmeaning of the word is (meaning varieties out
of the pale ofthe conditions above enumerated), and alſo that of
the tranſitions of meaning within the circle of which the infe-
rence remains good.
The most common uſes of the verb are ;-firſt abſolute iden-
tity, as in ' the thing he ſold you is the one I fold him : ' ſecondly,
agreement in a certain particular or particulars understood, as
in ' He is a negro' faid of a European in reference to his colour :
thirdly, poſſeſſion of a quality, as in the roſe is red :' fourthly,
reference of a ſpecies to its genus, as in ' man is an animal.' All
theſe uſes are independent of the use of the verb alone, denoting
exiſtence, as in ' man is [i. e. exiſts] .' In all theſe ſenſes, and in
all which might be added confiftently with the conditions in page
50, ſome propofitions ſometimes admit of having the ſenſe of is
ſhifted, and ſome do not. Thus, in negative propofitions, the is
of agreement in particulars may be lawfully converted into that
of identity : if ' No A is B in colour, then abſolutely ' No A is
B.' But ' Every A is B' in colour, does not prove ' Every A
is B. But the firſt pair might be connected by a fyllogifm .
The is of agreement in particulars may always be reduced to
the is of identity, by alteration of the predicate ; thus ' Every A
is B in colour' is ' Every A is a thing having the colour of one
of the Bs.' When a fyllogifm has a negative conclufion, and
the middle term is, or can be made, the predicate of both pre-
miſes, then the whole ſyllogiſm can be transformed from one in
which there is only the is of agreement to one in which there is
no is but that of identity. For example, ſuppoſe the premiſes to
be ' No X is Y (in colour) ; every Z is Y (in colour),' not
meaning neceſſarily that all the Ys are of one colour, but reading
it as ' No X is of the colour of any one of the Ys ; every Z is
of the colour of one of the Ys.' The conclufion is that ' no Z
is X (in colour),' or ' no Z is of the colour of any one of the
Xs.' But from this it follows that no Z is X, for if any one Z
were abſolutely X, it would have * the colour of that X. This
* The reader muſt not paint any of the letters during the proceſs . The
ſenſe in which we say a door is the ſame door as before, after it has been
painted of a different colour, is not the ſenſe of logical identity : it is the
ſame in all but colour and colouring matter ; and the is is one of agreement.
Except as a joke in ſufficient anſwer to a captious objection or a trap, no
54 On Propofitions.
laſt conclufion can be brought directly from altered premiſes :
thus, is being that of identity, we have ' No X is [a thing hav-
ing the colour of one of the Ys] ; every Z is [a thing having
the colour of one of the Ys] ; therefore no Z is X.' But ſup-
poſe we take the following premiſes, ' Some Ys are not Xs (in
colour) ; every Y is Z (in colour). From this it follows that
ſome Zs are not Xs (in colour), and thence that ſome Zs are
not Xs. But we cannot now alter the premiſes, ſo as to produce
the laſt concluſion from X, Z, and a middle term.

CHAPTER IV .

On Propofitions.

A NAME is a ſymbol which is attached to one or more

objects of thought, on account of ſome reſemblance, or
community of properties. Or elſe it is a ſymbol attached to
fome one or more objects of thought, to diftinguiſh them from
others having the ſame properties. Objects of the ſame name
are, ſo far as that name is concerned, undiftinguishable. And
one object may have many names, as being one in each of many
claſſes of objects of thought.
Names, as explained in chapter II, are excluſively the ob-
jects of formal logic. The identity and difference of things is
deſcribed by aſſerting the right to affert, or the right to deny, the
application of names. And names, whether fimple or complex,
will be repreſented by letters of the alphabet, as X, Y, Z.
A propoſition is the aſſertion of agreement, more or leſs, or
diſagreement, more or leſs, between two names. It expreſſes
that of the objects of thought called Xs, there are ſome which
are, or are not, found among the objects of thought called Ys :

change whatever muſt take place in the terms of conclufion, during infe-
rence . The American calculating boy, Zerah Colburn, was asked how
many black beans it would take to make ten white ones ; to which he very
properly anſwered ' Ten, if you ſkin 'em :' but the ten ſkinned beans would
not be the fame beans as before : except, indeed, to thoſe to whom black is
white.
 On Propofitions. 55
that there are objects which have both names, or which have
one but not the other, or which have neither.
For the moſt part, the objects of thought which enter into a
propoſition are ſuppoſed to be taken, not from the whole univerſe
of poffible objects, but from ſome more definite collection of
them. Thus when we say " All animals require air," or that
the name requiring air belongs to every thing to which the name
animal belongs, we ſhould underſtand that we are ſpeaking of
things on this earth : the planets, &c., of which we know no-
thing, not being included. By the universe of a propofition, I
mean the whole range of names in which it is expreſſed or un-
derstood that the names in the propofition are found. If there
be no ſuch expreſſion nor underſtanding, then the univerſe ofthe
propoſition is the whole range of poſſible names . If, the uni-
verſe being the name U, we have a right to ſay ' every X is Y,'
then we can only extend the univerſe ſo as to make it include all
poſſible names, by saying ' Every X which is U is one of the Ys
which are Us,' or ſomething equivalent.
Contrary names, with reference to any one univerſe, are thoſe
which cannot both apply at once, but one or other of which al-
ways applies . Thus, the univerſe being man, Briton and alien
are contraries ; the univerſe being property, real and perſonal are
contraries. Names which are contraries in one univerſe, are
not neceffarily ſo in a larger one. Thus in geometry, when the
univerſe is one plane, pairs of ſtraight lines are either parallels or
interfectors, and never both : parallels and interſectors are then
contraries. But when the ſtudent comes to folid geometry, in
which all space is the univerſe, there are lines which are neither
parallels nor interfectors ; and theſe words are then not contra-
ries. But names which are contraries in the larger and contain-
ing univerſe, are neceſſarily contraries in the ſmaller and contained,
unleſs the ſmaller univerſe abſolutely exclude one name, and then
the other name is the univerſe .

In future, I always underſtand ſome one univerſe as being that

in which all names uſed are wholly contained : and alſo (which
it is very important to bear in mind) that no one name mentioned
in a propoſition fills this univerſe, or applies to everything in it.
Nothing is more eaſy than to treat the ſuppoſition of a name
being the univerſe as an extreme caſe. And I ſhall denote con
56 On Propofitions.
traries by large and ſmall letters : thus, X being a name, x is the
contrary name. And everything (in the univerſe understood) is
either X or x : and nothing is both.
A propoſition may be either fimple and incomplete, or complex
and complete. The ſimple propofition only aſſerts that Xs are
Ys, or are not Ys : the complex propoſition, which always con-
ſiſts of two ſimple ones, diſpoſes in one manner or the other of
every X and every Y. Thus ' Every X is Y' is a ſimple pro-
poſition : but it forms a part of two complex propofitions. It
may belong either to ' every X is Y and every Y is X,' or to
Every X is Y and ſome Ys are not Xs.'
The propoſitions advanced in common life are usually com-
plex, with one ſimple propofition expreſſed and one understood :
but books of logic have hitherto confidered only the ſimple pro-
poſition. And this laſt ſhould be confidered before the complex
form.
The fimple propoſition must be confidered with reſpect to
fign, relative quantity, and order.
Simple propoſitions are of twoſigns : affirmative and negative.
It is either Xs are Ys,' or ' Xs are not Ys.' The phrases are
and are not, or is and is not, which mark the distinction, are
called copula.
The relative quantity of a propofition has reference to the
numbers of inſtances of the different names which enter it. The
distinctions of quantity uſually recognized are all and fome * :
leading to the diſtinction of univerſal and particular. Thus
' Every X is Y' and ' Every X is not Y' are the univerſal affir-
mative and negative propoſitions: the latter is usually ſtated as
No X is Y.' And ' ſome Xs are Ys' and ' ſome Xs are not
Ys' are the particular affirmative and negative propofitions. And
when the propoſitions are reduced ſtrictly to theſe four forms,
* Some, in logic, means one or more, it may be all. He who ſays that
Some are , is not to be held to mean that the rest are not. 'Some men
breathe,' 'fome horſes are diftinguishable by ſhape from their riders' would
be held falſe in common language. The reaſon is, as above noted, that
common language uſually adopts the complex particular propoſition, and
implies that ſome are not in ſaying that ſome are. The ſtudent cannot be
too careful to remember this diſtinction. A particular propoſition is only a
' may be particular, '
 On Propofitions. 57
the first named, X, is called the subject, and the ſecond named,
Y, the predicate.
It has been propoſed to conſider the univerſal propoſitions as
definite with reſpect to quantity : but this is not quite correct.
The phrafe ' all Xs are Ys' does not tell us how many Xs
there are, but that, be the unknown number of Xs in existence
what it may, the unknown number mentioned in the propofition
is the fame. That which is definite is the ratio of the number
ofXs of the propoſition to the Xs of the univerſe. So under-
ſtood, however, the ' definite quantity,' as an abbreviation, may be
faid to belong to univerſals. And the indefiniteneſs of the parti-
cular propofition is only hypothetical. It is in our power to fup-
poſe thefome to be one half of the whole, or two-thirds, or any
other fraction .
The quantity of the ſubject is expreſſed ; that of the predicate,
though not expreſſed, is neceſſarily implied by the meaning of
language. The predicate of an affirmative is particular : the
predicate of a negative is univerfal. If I say ' Xs are Ys,' even
though I speak of all the Xs, I only really ſpeak of ſo many Ys
as are compared with Xs and found to agree : and theſe need
not be all the Ys. Every horſe is an animal,' declares that ſo
many horſes as there are to ſpeak of, ſo many animals are ſpoken
of: and leaves it wholly unſettled whether there be or be not
more animals left. But if I ſhould say ' Xs are not Ys,' though
it ſhould be only one X, as in ' this X is not a Y,' yet I ſpeak of
every Y which exiſts. The aſſertion is ' this X is not any one
whatsoever of all the Ys in exiſtence.' A person who ſhould
wish to verify by actual inſpection, ' theſe 20 Xs are Ys' might,
perchance, be enabled to affirm the reſult upon the examination
of only 20 Ys, if he came firſt upon the right ones. Buthe
could not verify ' this one X is not a Y' until he had examined
every Y in exiſtence. This is the common doctrine, but though
admitting of courſe that the affirmative propoſition only enables
us to infer of ſome inſtances ofthe predicate, yet I think it more
correct to ſay that the predicate itſelf is ſpoken of univerſally, but
indiviſibly, and that in the negative propoſition the predicate is
ſpoken of univerſally and diviſibly. Some Xs are Ys' tells
us that each X mentioned is either the firſt Y, or the ſecond Y,
or the third Y, &c., no Y being excluded from compariſon. But
 58 On Propofitions.
' Some Xs are not Ys' tells us that each X mentioned is abſo-
lutely not the firſt Y, nor the ſecond, nor the third, &c ; is not,
in fact, any one of all the Ys. Still, however, the predicate of
an affirmative yields no more than it would do if the Ys finally
accepted as Xs were ſpecially ſeparated, and confidered as the
only Ys ſpoken of.
The relation of the univerſal quantity to the whole quantity of
inſtances in exiſtence is definite, being that whole quantity itſelf.
But the particular quantity is wholly indefinite : ' Some Xs are
Ys' gives no clue to the fraction of all the Xs ſpoken of, nor to
the fraction which they make of all the Ys. Common language
makes a certain conventional approach to definiteneſs, which has
been thrown away in works of logic. ' Some,' uſually means a
rather ſmall fraction of the whole ; a larger fraction would be
expreſſed by ' a good many' ; and ſomewhat more than half by
( moſt' ; while a ſtill larger proportion would be a great majo-
rity' or ' nearly all'. A perfectly definite particular, as to quan-
tity, would expreſs how many Xs are in exiſtence, how many
Ys, and how many of the Xs are or are not Ys : as in '70 out
of the 100 Xs are among the 200 Ys'. In this chapter I ſhall
treat only the indefinite particular, leaving the definite particular
for future confideration.
The order of a propoſition has relation to the choice of fub-
-ject and predicate. Thus ' Every X is Y' and ' every Y is X'
though both eſtabliſh a univerſal affirmative relation between X
and Y, yet are in fact two different propoſitions. They are called
converse forms. When the ſubject and predicate are of the ſame
fort of quantity, both univerſal or both particular, the converſe
forms give the ſame propoſition. Thus ' No X is Y' and ' No
Y is X' are the ſame ; neither has any meaning, except perhaps
of emphasis, which the other has not. And Some Xs are Ys'
is the fame as ' Some Ys are Xs' . The univerſal negative, then,
in which both terms are univerſal, and the particular affirmative,
in which both are particular-are neceſſarily convertible propofi-
tions. But the univerſal affirmative, in which the ſubject is uni-
verſal and the predicate particular, and the particular negative, in
which the ſubject is particular, and the predicate univerſal-are
not neceffarily convertible, and are generally called inconvertible.
They may be convertible, in one caſe, and inconvertible in an
 On Propofitions. 59
other. But the term inconvertible is not incorrect, for the fol-
lowing reaſon .
The agreements and diſagreements which are treated in logic
are of this character ; there can only be agreement with one, but
there may be diſagreement with all. If this X be a Y' it is
one Y only : it is this X is either the firſt Y, or the ſecond
Y, or the third Y, &c. If there be 100 Ys, there is, to thoſe
who can know it, 99 times as much negation as affirmation in
the propofition : and yet moſt aſſuredly it is properly called affir-
mative. But if it be ' this X is not a Y, we have ' this X is
not the firſt Y, and it is not the ſecond Y, and it is not the third
Y, &c.' The affirmation is what is commonly called disjunctive,
the negation conjunctive. A disjunctive negation would be no
propofition at all, except that one and the ſame thing cannot
be two different things : any X is either not the firſt Y or not
the ſecond Y. And in like manner a conjunctive affirmation
would be an impoffibility : it would ſtate that one thing is two
or more different things .
We must be prepared, then, to confider caſes of oppofition in
which on the one fide there is fixed neceffity, and on the other
ſide poffibility of alternatives : and we must be prepared to de-
note theſe by oppoſite terms, which, looking to etymology only,
denote fixed neceffities of oppoſite characters. This happens in
the cafe above : convertible means abſolutely and neceſſarily con-
vertible, inconvertible means convertible or inconvertible as the
cafe may be. Taking the four forms of one order, we find that
each of the univerſals cannot exiſt with either propofition of op-
poſite form. Thus ' Every X is Y ' cannot be true if either
' No X is Y ' or ' Some Xs are not Ys : ' while ' No X is Y
cannot be true if either ' Every X is Y ' or ' Some Xs are Ys. '
But each ofthe particulars is neceſſarily inconſiſtent with nothing
but the univerſal of oppofite form. That ' Some Xs are Ys ’
cannot be true if ' No X is Y ' but it may be true if ' Some Xs
are not Ys .' And ' Some Xs are not Ys ' cannot be true if
' Every X is Y, but it may be true though ' Some Xs are
Ys .'
The pair ' Every X is Y' and 'fome Xs are not Ys ' are called
contradictory : and ſo are the pair ' No X is Y' and ' Some Xs
are Ys .' Of each pair of contradictories, one must be true and
60. On Propofitions.
one must be falſe ſo that the affirmation of either is the denial
of the other, and the denial of either is the affirmation of the
other. The pair ' Every X is Y ' and ' No X is Y' are uſually
called contraries ; contrariety implying the utmoſt extreme of
contradiction. Contraries may both be falſe, but cannot both be
true. The pair ' Some Xs are Ys,' and ' Some Xs are not Ys,'
which may both be true, but cannot both be falſe, are uſually
called fubcontraries. But, for reaſons hereafter to be given, I
intend to abandon the diſtinction between the words contrary
and contradictory, and to treat them as ſynonymous. And the
propoſitions uſually called contraries, ‘ Every X is Y' and ' No
X is Y' I ſhall callfubcontraries : while thoſe uſually calledfub-
contraries ' Some Xs are Ys' and ' Some ✗s are not Ys ' I ſhall
call fupercontraries. ,
I fhall now proceed to an enlarged view of the propofition,
and to the ſtructure of a notation proper to repreſent its different
cafes.
As usual, let the univerſal affirmative be denoted by A, the par-
ticular affirmative by I, the univerſal negative by E, and the par-
ticular negative by O. This is the extent ofthe common ſym-
bolic expreſſion ofpropofitions : I propoſe to make the following
additions for this work. Let one particular choice of order, as
to ſubject and predicate, be ſuppoſed eſtabliſhed as a ſtandard of
reference. As to the letters X, Y, Z, let the order be always
that of the alphabet, XY, YZ, XZ. Let x, y, z, be the con-
trary names of X, Y, Z ; and let the ſame order be adopted
in the ſtandard of reference. Let the four forms, when choice
is made out of X, Y, Z, be denoted by A₁, Ει, Ιι, Οι; but when
the choice is made from the contraries, let them be denoted by
Α', Ε' , Ι', Ο'. Thus, with reference to Y and Z, ' Every Y is
Z' is the A of that pair and order : while ' Every y is z ' is the
A'. I ſhould recommend A. and A' to be called the fub-A and
the fuper-A of the pair and order in queſtion : the helps which
this will give the memory will preſently be very apparent. And
the ſame of I, and I', &c.
Let the following abbreviations be employed ;-
X)Y means ' Every X is Y' X.Y means ' No X is Y '
X :Y -

' Some Xs are not Ys' XY ' Some Xs are Ys '

 On Propofitions. 61

There are eight distinct modes, independent of contraries, in

which a fimple propoſition may be made by means of X and Y.
Theſe eight modes are X)Y and Y)X, X : Y and Y: X, X.Y and
Y.X, and XY and YX. But the eight are equivalent only to
fix : for X.Y and Y. X are the ſame, and ſo are XY and YX.
-Again, there are fix fimple propofitions between x and y, fix be-
tween X and y, fix between x and Y. Taking in contraries,
there are then twenty-four apparent modes of forming a ſimple
propoſition from X and Y : but theſe are not all distinct. Eight
of them contain all the reſt : theſe eight being the A1, Ει, Ιι, Οι,
A', E', I' , O', above deſcribed. This is ſeen in the following
table, the ſtudy of which ſhould be carefully made,
A. X)Y = X.y = y)x A' x)y = x.Y = Y)X
Ο X:Y =
Xy =
y: x O' x :y = xY =Y:X
E. X.YX)y Y)x
= =
E' x.y = x)Y = y)X
IXY = X:y = Y:x I' xy = x:Y = y: X
I ſuppoſe moſt readers will readily fee the truth ofthe identities
here affirmed : if not, the following mode of illuſtration (which
will be very useful when I come to treat of the fyllogifm) may be
tried. Let U be the name which is the univerſe of the propo-
ſition : and write down in a line as many Us as there are diftinct
objects to which this name applies. A dozen will do as well for
illuſtration as a million. Under every U which is an X write
down X : and x, of course, under all the reſt. Follow the ſame
plan with Y. The occurrence of letters in the ſame column
ſhows that they are names of the ſame object. The following
are ſpecimens of the eight ſtandard varieties of aſſertion, to which
all the reſt may be referred.
A, UUUUUUUUUUUU A' UUUUUUUUUUUU
XXXXXxxxxxxx XXXXXXXXxxxx
YYYYYYYYyyyy YYYYY y y ууууу
OD UUUUUUUUUUUU O' ) UUUUUUUUUUUU
XXXXXXXxxxxx XXXXXxxxxxxx
Lyyyy YYYYYY y y I') YYyyyyy y YYYY
E. UUUUUUUUUUUU E' UUUUUUUUUUUU
XXXX x xxxxxxx XXXXXXXXxxxx
yyyyyyyYYYYY yyyyy YYYYYYY
 62 On Propofitions.
In the firſt ſcheme, A1, there exiſt twelve Us, the first five of
which are both Xs and Ys, the next three Ys but not Xs, the laſt
four neither Xs nor Ys. This caſe, ſo conſtructed that X)Y is
true, ſhows X.y and y)x.
The propoſitions A and A', X)Y and x)y, may be called con-
tranominal, as having each names contrary ofthoſe in the other.
It appears, then, that as to inconvertibles, contranominal and con-
verſe are terms of the ſame meaning, for X)Y and y)x are the
fame, and x:y and Y:X. And fince it is more natural to ſpeak
of direct names than of their contraries, it will be beſt to attach
to A' and O' the ideas of Y)X and Y:X ; but not ſo as to forget
their derivation from x)y and x:y. Obſerve alſo that each uni-
verſal propofition has converted contranominals for its affirmative
forms. Thus X)Y = y)x : and though X.Y is not y.x, yet if
we make X.Y take the affirmative form X)y, it is equivalent to
Y)x. In particular propofitions, the negative forms have the
ſame property. The contranominals of the convertible propo-
ſitions E. and I are of totally different meaning. They have
never till now been introduced into logic, and a few words of
explanation are wanted.
Firſt as to I' or xy. We here expreſs that ſome not-Xs are
not-Ys, or that there are things in the univerſe which are neither
Xs nor Ys. That is, X and Y are not contraries. Next as to
E' or x.y. We here expreſs that no not-X is not-Y, or that
everything in the univerſe is either X or Y, or both. Theſe laſt
words are important : by omitting them, we ſhould imagine that
x.y ſignifies that X and Y are contraries ; which is not necef-
ſarily true.
Accordingly, the eight ſtandard forms of expreffion, with re-
ference to the order XY, and exhibited in the form in which it
will be moſt convenient to think and ſpeak of them, are as
follows,
(A, or X)Y Every X is Y A'or isX
Oor X: Y Some Xs are not Ys EveryYs are not Xs
{ 0 ' or Y:X Some
E or X.Y No X is Y E' or x.y Everything is eitherXor Y
I or XY Some Xs are Ys I' or xy Some things are neither Xs norYs .

Returning to the table, we now ſee the following general laws.

1. Each triad ofequivalents contains two inconvertibles and one
convertible. 2. Of the four, X, Y, x, y, each of the eight forms
 On Propofitions. 63
ſpeaks univerſally of two, and particularly of two. 3. A pro-
poſition ſpeaks in different ways of each name and its contrary ;
univerſally of one and particularly of the other. 4. The propo-
ſitions called contradictory, from the common meaning of this
word, may be ſo called in another ſenſe : for they ſpeak in the
fame manner of contraries. Thus X)Y ſpeaks univerſally of X,
and particularly of Y: its denial, X: Y or y:x, ſpeaks univerſally
of x, and particularly of y.
Any two of the eight forms being taken, it is clear either that
they cannot exiſt together, or that one must exiſt when the other
exifts, or that one may exiſt either with or without the other.
The alternatives of each caſe are preſented in the following table.
Con- Isindif- Iscon- Isindit-
Denies tains ferent to Denies tained in ferent to
AOEE' II' A'O ' ΟΑΕΕ' | Α'Ο'ΙΙ'
Α' Ο'Ε'Ε, I'I AO Ο' Α' Ε'Ε ΑΟΙΙ,
Ε ΙΑΑ' 00' Ε'Ι' I E AΑ' Ε'Ι'ΟΟ '
Ε' Ι'Α'Α, Ο'Ο ΕΙ Ι' Ε' Α'Α. ΕΙΟΟ

Let the concomitants of a propoſition be thoſe to which it is

wholly indifferent. Then it appears that each univerſal has for
concomitants its contranominal and the contradictory of the laſt :
but each particular has all for concomitants except only its own
contradictory. Each univerſal denies, beſides its own contra-
dictory, the two univerſals of oppoſite name ; and contains the
two particulars of the ſame name. The two concomitants of a
univerſal may be deſcribed as its univerſal and its particular con-
comitant.
There is a certain fort of repetition in our choice of the four
forms, combined with the four ſelections XY, Xy, xy, xY. If
any one of the four forms A,E A' E' be applied to all the above,
it will give the four forms derived from XY. Thus the A of
XY, Xy, xy, xY, are ſeverally the A1, E1, A' and E' of XY ;
and the E' of XY, Xy, xy, and xY are ſeverally the E', A', Ει ,
and A of XY : and ſo on. It will ſerve for exerciſe to verify
the above, and ſtill more the caſes contained in the following.
There are four things in a propoſition, each of which may be
changed into its contrary : ſubject, predicate, order, and copula.
Let S be the direction to change the ſubject into its contrary : P
64 On Propofitions.
the fame for the predicate : let T be the direction to transform
the order : and F the direction to change the form, from affirma-
tive to negative, or from negative to affirmative. When T enters,
let it be done laſt, to avoid confufion. Thus SPT performed
upon X)Y gives x)Y from S, x)y, from P, and y)x from T ;
which is X)Y, ſo that in this caſe alteration of ſubject, predicate,
and order, is no alteration at all. Let L be the repreſentation of
no alteration at all. To inveſtigate equivalent alterations, ob-
ſerve, firſt, that F and P, ſingly, are identical : thus F performed
on X.Y gives X)Y, and P on X.Y gives X.y. And X)Y =
X.y. This perfect identity of F and P in effect, remains in
all combinations into which T does not enter. But when T
enters, it is S and F which are identical. Thus ST performed
on Y)X gives X)y or X.Y : and FT performed on Y)X gives
X.Y. The reaſon is, that T interchanges ſubject and predicate ;
ſo that F, after T, makes a change which is counterbalanced by
a change in what was the ſubject. Accordingly, remembering
that each operation performed twice is no operation at all (thus
PP is L, and TT is L), we have in all caſes
P= F, SP = SF, PF = L, SPF = S
ST = FT, SPT = FPT, SFT = T, SPFT = PT
all which ſhould be tried for exerciſe. Again, in a convertible
propoſition, transformation is no alteration or T =L : in an incon-
vertible one, transformation changes it into its contranominal ;
or T = SP. Now ſet out as follows ;-L, in convertible propo-
ſitions is T ; which in inconvertibles, is SP ; which, in convertibles
again, is SPT ; which, in inconvertibles again, is TT, or L.
Put theſe down as follows, writing under them the operations
which are always equivalent to them, as ſhewn above,
L T SP SPT L
PF SFT SF PFT PF

The combinations written under one another are always the

ſame in effect : thoſe ſeparated by double lines have the ſame
effect on convertibles : thoſe ſeparated by ſingle lines, have the
fame effect on inconvertibles. Again P, for convertibles, is the
fame as PT; which, for inconvertibles is the fame as PSP, or
S; which, for convertibles again, is the fame as ST ; which, for
inconvertibles, is SSP or P. Theſe treated as before, give the
table
 On Propofitions. 65
P PT S STP
FSPFT
SPFT SPF FT F

In theſe two cycles there are L and all the fifteen ſelections
which can be made out of S, P, F, T. And every poſſible caſe
of equivalent changes is contained in theſe two tables. Thus
PT is in all caſes equivalent to SPFT ; in convertible caſes, to
Pand to F ; in inconvertible ones, to S and to SPF. And no
other combination is in any caſe equivalent to PT. In verifica-
tion of theſe tables, obſerve that the operation F always occurs
in the lower line, and never in the upper ; and that this opera-
tion changes convertibles into inconvertibles, and vice versa.
We ought then to expect, that the equivalences which, con-
taining F, apply to inconvertibles, will be thoſe which when F is
ſtruck out, apply to convertibles ; and vice versa. And ſo we shall
find it : for instance, SPFT and SPF are equivalent when per-
formed on inconvertibles ; ſtrike out F and we have SPT and
SP, which are equivalent when performed on convertibles.
It appears, then, that any change which can be made on a
propofition, amounts in effect to L, P, S, or PS. This is another
verification of the preceding table : for all our forms may be de-
rived from applying thoſe which relate to XY in the caſes of
Xy, xY, and xy .
We have seen that A, and A' both contain I and I' ; and that
E and E' both contain O and O ' . Hence each of the univerſals
may be faid to be theſtrengthened form of either of its particulars
of the ſame ſign: and each of the particulars the weakenedform
of its univerſals of the ſame ſign. The only distinction which
appears between the two forms of the convertible particulars,
XY and YX, xy and yx, is that the ſtrengthened forms derived
from extending the ſubjects are different. Thus xy gives x)y or
Y)X ; but yx gives y)x or X)Y.
A complex propoſition is one which involves within itſelf the
aſſertion or denial of each and all of the eight ſimple propoſitions.
If theſe eight propoſitions were all concomitants, or if any num-
ber of them might be true, and the reſt falſe, there would be
256 poſſible caſes of the complex propoſition. As it is, owing
to the connexion eſtabliſhed in the table of page 63, there are
butfeven.
F
66 On Propofitions.
Firſt, let the names X and Y be ſo related that neither of the
four univerſals are true. Then all the four particulars are true :
and this is the firſt caſe. Let it be called a complex particular,
and denoted by P. Then, denoting coexistence of ſimple pro-
poſitions by writing + between their ſeveral letters, we have
P = O' + O + I' + I

This caſe is ofthe leaſt frequent mention in the theory of the

fyllogifm.
Next, let one of the univerſal propoſitions be true. Then five
of the other propoſitions are ſettled, either by affirmation or de-
nial. There remain the two concomitants, which are contra-
dictory ; ſo that only one is true. Accordingly, with the excep-
tion of the complex particular juſt deſcribed, every complex pro-
poſition muſt conſiſt of the coexiſtence of a univerſal and one of
its concomitants. But there are not therefore eight more ſuch
propoſitions : for A' +A and A +A' are the ſame, and ſo are
E1 + E' and E' + E.. The remaining number is then reduced
to fix, which are
A + O ', A +A', A' +O
E + I', Ει +Ε', E' +11,
Theſe must be ſeparately examined.
Firſt, take A + A' (the order XY always understood). We
have then X)Y and Y)X. That is, there is no object whatſo-
ever which has one of theſe names, but what alſo has the other.
The names X and Y are then identical, not as names, but as
ſubjects of application. Where either can be applied, there can
the other alſo. Thus, in geometry (the univerſe being plane
rectilinear figure) equilateral and equiangular are identical names.
Not that they agree in etymology nor in meaning : more than
this, a few words would explain the firſt to many who could not
comprehend the ſecond without difficulty. But they agree in
that what figure ſoever has a right to either name, it has the ſame
right to the other. It will tend to uniformity of language, if we
call X, in this caſe, an identical of Y, and Y an identical of X.
Let the ſymbol of an identical be D : then we have
D=A + A'
 On Propofitions. 67
Next, take A + O'. We have then X) Y and Y: X. Every X
is Y, and ſo far there is a character of identity. But ſome Ys
are not Xs ; there are more Ys than Xs, and X ſtops ſhort ofa
complete claim of identity with Y. Let X be called afubiden-
tical of Y (thus man is a fubidentical of animal), and let D. de-
note this cafe. Then
D =A +O'

Let A' + O exiſt. We have then Y)X and X:Y. Every Y

is X, and ſo far there is identity. But ſome Xs are not Ys, there
are more Xs than Ys, orX goes beyond a claim of identity with
Y. Let X be now called a fuperidentical of Y, and let it be
denoted by D'. Then
D' = A' +O
The terms ſuperidentical and ſubidentical are obviouſly correla-
tive. If X be either of Y, Y is the other of X. Now let us
conſider E1+ E'. We have then X.Y and x.y. There is no-
thing which is both X and Y, there is nothing which is neither.
Conſequently X and Y are contraries, or juſt fill up the univerſe.
Let C be the mark of this relation. Then
C=E + E'

Next, take E + I'. We have then X.Y and xy. Nothing is

both X and Y, but there are things which are neither. X and
Y are clear of one another, but do not amount to contraries,
for they do not fill up the univerſe. Let them be calledfubcon-
traries, (thus in the univerſe metal, gold and silver are ſubcontra-
ries, and let C. denote the relation. Then
C =E + I'

Laſtly, take E' + I. We have x.y and XY. The names fill
the univerſe ; for there is nothing but what is either X or Y.
But they overfill it ; for ſome things are both Xs and Ys. There
is then all the completeness of a contrary and more. LetX and
Y be calledfupercontraries,* and let C' denote the relation. Then
we have
C'E' + I

* The fupercontrary relation, though eſſential to a complete ſyſtem of

fyllogifm, is not frequently met with. The other extreme of the ſupercon-
68 On Propofitions.
To completeour language, let A₁ or X)Y, with reference to the
order XY, be calledfub-affirmative ; and A' or Y)X,fuperaffirma-
tive. Let E or X.Y be calledfubnegative; and E'or x.y,fuperneg-
ative. Let the particulars I., I', and O₁, O', have alſo theſe ſeveral
names. This extenfion of our language will require a little ex-
planation.
When I say that X is a ſubidentical of Y, I mean that the
etymological ſuggeſtions are actually satisfied. The whole name
X, and more, is contained in Y. But when I say that X is a
univerſal ſubaffirmative of Y, or X)Y, I mean no more than that
we have the propoſition whoſe form is not fuperaffirmative, ac-
cording to the etymology of that word. An algebraiſt would
well underſtand the diſtinction at a glance. He has often to
diftinguiſh the caſe in which a is leſs than b from that in which
a is leſs than or equal to b : the caſe in which the extreme limit
of the affertion is not included from that in which it is included.
Again, the word negative had better be viewed as not ſo much
preſenting excluſion for its firſt idea, as incluſion in the contrary.
Thus a fubnegative, when univerſal, is to ſuggeſt complete in-
cluſion in the contrary, meaning the extreme cafe, poſſibly ;
namely, that the ſubnegative names may be contraries. Again,
ſupernegative is to ſuggeſt the idea of fupercontrary, with the
loweſt extreme, the relation of contrary, poſſibly included.
For exerciſe in this language, and in the ideas which it is
meant to preſent, I now ſtate the following reſults .
Univerſal affirmation, though as a general term, it is to include
fuper and fub affirmation, yet looked at as one of the three, and
diftinguiſhed from the reſt, it means identity. The fame of ne-
gation and contrariety. Subidentity requires univerſal fubaffir-
mation and particular ſupernegation. Identity is univerſal ſub
and ſuper affirmation, both. Superidentity requires univerſal
ſuperaffirmation and particular ſubnegation. Subcontrariety re-
quires univerſal ſubnegation and particular fuperaffirmation. Con
trary, or the fubidentical, is ſo much the eaſieſt of all our complex relations,
that the latter rarely allows the former to appear. The firſt inſtance that
ſuggeſted itſelf to me was man and irrational (as deſcriptive of the quality of
the individual and not of the ſpecies) in the univerſe animal. Theſe more
than fill that univerſe, idiot being common to both. But it is more natural
to ſay that rational (in this ſenſe) is ſubidentical of man.
 On Propofitions. 69
trariety is univerſal ſub and ſuper negation, both. Supercontra-
riety requires univerſal ſupernegation and particular ſubaffirma-
tion. Again, univerſal ſubaffirmation is either ſubidentity or iden-
tity : particular fubaffirmation is a denial of contrariety and ſub-
contrariety. Univerſal ſuperaffirmation is either ſuperidentity or
identity : particular ſuperaffirmation denies contrariety and ſuper-
contrariety. Univerſal ſubnegation is either ſubcontrariety or
contrariety : particular ſubnegation denies ſubidentity and iden-
tity. Univerſal ſupernegation is either ſupercontrariety or con-
trariety : particular ſupernegation denies ſuperidentity and iden-
tity. All this is expreſſed in the following table,
Da affirms A and O' | A affirms Dior D
D -

A and A' A -

Dior Dor D'

D' -

A' and O Α' -

D' or D
C -

E and I' E -

Cor C
C -

E and E' E -

Cor Cor C'

C' -

E' and I E' -

C' or C
Denial ofDi A' or O O denies D₁ and D
-
D -

O' or O Ο -

D₁ and D, or D' and D

-
D' -

Aor O' Ο' -

D' and D
-
C -

E' or I, I -

C and C
C I' or I I C and C or C' and C
-
C' -

E or I' I' -
C' and C

Every ſubidentical of a name is the ſubcontrary ofits contrary ;

every fubcontrary is the ſubidentical of the contrary. Treat the
word contrary as negative, the word identical as poſitive ; and
the two as of different ſigns. Then the algebraical rule ' like
ſigns give a poſitive, unlike ſigns a negative,' holds in every caſe :
including the variety of it ſo well known as ' two negatives make
an affirmative.' When the modifying prepoſition comes firſt it
must be retained ; when it comes ſecond, it must be changed.
Thus the fubcontrary of a contrary is a ſubidentical : but the con-
trary of a fubcontrary is a fuperidentical. In putting two rela-
tions together, however, we have got into fyllogifm, as we ſhall
preſently fee.
The following tables will ſhow a connexion between the ex-
preſſions, for different orders and ſelections, which it may be useful
to verify.
 70 On Propofitions.
XY YX xY Yx Xy yX xy yx
AO'D A'OD' E' I, C'E'I, C'E, I' C, E, I' C. A'OD' AO'D
A'OD' AO'D, E, I' C, E, I' C. E' I, C'E' I, C' AO'D A'OD'
E. I' CE, I' C A'OD' AO'D AO'D A'OD' E' I, C'E' I, C'
E' I, C'E' I, C' AO'D A'OD' A'OD' AO'DE I' CE. I' C
This table only contains ſome of the rules already laid down
in pp. 64, 65. It expreſſes that, for instance, the A1, O', and D
of XY, are ſeverally the ſame as the E., I', and C. ofyX. This
table may be exhibited thus, the identicals counting as inconvert-
ibles, the contraries as convertibles.
In Convertibles, In Inconvertibles,
Change of changes changes

Subject Sign and Prepoſition Sign and Prepofition

Predicate Sign Sign
Subject and Predicate Prepofition Prepofition
Order Neither Prepofition
Subject and Order Sign Sign and Prepofition
Predicate and Order Sign and Prepoſition Sign
Subject, Predicate, and Order | Prepofition Neither

In all cafes, change of ſubject is change both ofſign and pre-

poſition ; change of predicate is change of ſign ; change of fub-
ject and predicate is change of prepoſition. Theſe three cafes
are of great importance in the fyllogifm : and the reader would
do well to connect in his mind
Subject with Sign and prepofition
Subject and Predicate Prepofition
Predicate -

Sign
It is defirable to confider the ſeveral complex relations as to
the continuous tranſition from one into another : the growth of
names concerns not only the etymologiſt, but the logician alſo.
With the analogies and affinities by which the dominion of
one name is extended to inſtance after inſtance, and claſs after
claſs-and ſometimes, in ſcientific language at least, deprived of
a part of what it has held-I have here nothing to do. It is
enough that the phenomena exiſt which may be deſcribed as the
gradual transformation of one relation into another. The words
butt and bottle, for example, are now ſubcontraries in the uni-
verſe receptacle : but the etymology of the ſecond word ſhows
 On Propofitions. 71
that it was a ſubidentical of the firſt, being a diminutive. And
ifwe were to take the whole claſs butt, buſs, boot, buſhel, box,
boat, bottle, pottle, &c, which are all of one origin, the number
of tranſitions would be found to be very large.
I aſſume that all the inſtances of a name are counted and
arranged in its univerſe : a conceivable, though not attainable,
ſuppoſition. Alſo, that the inſtances of the name are arranged
contiguoufly, as in page 61. Whatever the reaſon may be
which dictates the particular arrangement choſen, it will generally
happen that the inſtances near to the boundary poſſeſs the cha-
racteriſtics of the name in a ſmaller degree than thoſe nearer the
middle. Let the contiguous arrangement be made of all the in-
ſtances of the name Y, the univerſe being U. Let another name
X begin to grow, commencing with one inſtance, that is, being
applied to one of the objects in the univerſe U, be it a Y or not ;
then to another contiguous, and ſo on. We are to enumerate the
ways in which ſuch changes, whether of increaſe or diminu-
tion, may cauſe one name to change its relation to another.
According as the change is made by acceffion or retrenchment,
it may be denoted by ( + ) or ( - ) .
Let the name X begin within the limits of the name Y: its
initial relation to Y is then Da •
And the poſſibility of the
following continuous changes is obvious :
D. ( + ) D ( + ) D' D. ( + ) P ( + ) C'
D. ( + ) P ( + ) D' D. ( + ) P ( − ) С
Hence Dı may become D' through either D or P, but C₁ or C'
only through P. Next, let X begin without the limits of Y :
the initial relation is C. We may have then
C. ( + ) C ( + ) C' C. ( + ) P ( + ) D'
C. ( + ) P ( + ) C' C. ( + ) P ( 1) D
Let X begin both within and without Y : its initial relation is
then P. And we have

P ( +) D', P ( + ) C', P ( 1 ) D., P ( - ) C

But when ( - ) follows D. or D, C, or C, we have nothing
except

D. ( -) D. , D ( - ) D. , C. ( -) C₁, C ( -) С
72 On Propofitions.
If we begin at the other extreme, with the name U, we have
U ( - ) D' U ( - ) C'
Beginning from D' and C' we have
D' ( −) D ( −) D. D' ( − ) P ( − ) С
D' ( − ) P ( −) D D' ( − ) P ( + ) C'
C' ( -) C ( − ) С C' ( − ) P ( 1 ) D
C' ( -) P ( -) С C' ( −) P ( + ) D'
But when ( + ) follows D' or D, C' or C, we have only
D' ( +) D' D ( + ) D', C' ( + ) C', C ( + ) C'
From the above lift it appears that the tranſition which is ac-
companied by a change of prepofition only can be made either
through the letter without prepoſition or through P : and in all
caſes with one continued mode of alteration. But when the tranf-
ition involves change of letter, it can only be made through P :
with continuation of the mode of alteration when the prepoſi-
tions are different, and change in the mode when they are the
fame. The following ſucceſſions contain the arrangement of the
refults.

With one altera- With one altera- With two altera-

tion (+) tion (-) tions (+-)
DDD' D'D D DP C
DP D' D'PD CP D
CC C' C' CC
C, PC' C' PC (- + )
D'PC'
D. P C' D'PC C
' P D'
C. P D' C' PD

The following confiderations will further ſerve to illuſtrate the

want of the extenſion of the doctrine of propoſitions made in
this chapter, and alſo the completeness of it. Among our most
fundamental distinctions is that of neceſſity and fufficiency ; of
what we cannot do without, and what we can do with ; of that
which must precede, and that which can follow. The contraries
of theſe are non-neceſſity and non-fufficiency. In theſe four words,
applied to both Y and y, we have the deſcription of the eight re
 On Propofitions. 73
lations of X to Y. For inſtance A or X)Y tells us that to
have an X, we must take a Y, or to be X, it is necessary to be
Y. Treating all in the ſame way, we have
A. X ) Y To take an X it is necessary to take a Y
A' Y ) X X Y
•

Sufficient •

E X.Y • X•
neceffary
•

y •

E' x.y ...X fufficient

• •

y • •

I XY ...X not necessary

•

y •

I' xy .. X notfufficient y •

OX : Y • X •

not necessary • • Y
O' Y : X • X • notfufficient • Y

And the convertibility of the ordinary mode of deſcription with

this new one may be easily ſhown in any cafe. For example,
what can we mean by ſaying that to take a X, it is not ſufficient
to take what is not Y ? Clearly that by taking not Y, or y, we
may at the fame time take a x, or that there are xs which are ys.
And fo on for the reft.
Of the four pairs XY, Xy, xy, xY, we know that each
propoſition may be expreſſed by three, and refuſes to be expreſſed
by one. If we now admit the two words impoſſible and contingent,
meaning by the latter that which, as the caſe may be, is poffible
or impoſſible, we shall eaſily ſee the following table for the uni-
verſals :
XY Xy xy xY
A X)Y N I S C
EX.Y I NC S
A' Y)X S CNI
E' x.y C S IN

The letters N, I, S, C, are the initials of neceſſary, &c. And

we read in the firſt line, that ifX ) Y, then to be X it is necef-
ſary to be Y ; to be X, it is impoffible to be y ; to be x it is ſuf-
ficient to be y ; and to be x, it is contingently poſſible or impoſſible
to be Y. Again, if by n and s we mean not neceſſary and not
fufficient ; by P, actually poſſible ; and by C, as before (C being
its own contrary), we have the following table for the parti-
culars :
74 On Propofitions.
XY Xy xy xY
OX : Y n P S C
IXY P n C S
O' Y : X S C n P
I' xy C S P n

Ofthe four contrary pairs, n, P, s, C, are related to the par-

ticulars preciſely as N, I, S, C, are to the univerſals. The inter-
change of Y and y is always accompanied by the interchange of
N and I, S and C, n and P, s and C ; the interchange of X
and x is that of N and C, Sand I, n and C, s and P ; of both
X and x, Y and y, is that of N and S, C and I, n and s, C
and P.
The complex relations may be thus deſcribed. According as
X is ſubidentical, identical, or ſuperidentical of Y, to be X it is
neceſſary and not ſufficient, neceſſary and ſufficient, or not ne-
ceſſary and ſufficient, to be Y : according as X is ſubcontrary,
contrary, or fupercontrary of Y, to be X it is neceſſary and not
ſufficient, neceſſary and ſufficient, or not neceſſary and ſufficient,
to be y. Or, as in the following table :
XY Xy xy xY

D Ns I Sn P
C I Ns P Sn
D' Sn P Ns I
C' P Sn I Ns

D NS I NS I
C INS I NS
P nspnspnspnsP

Instead of IC and PC, write I and C : for " impoffible, and

poſſible or impoſſible as the cafe may be " is " impoffible" &c.
The names of the complex relations, ſubidentity, identity, &c
I ſuppoſe will be held tolerably fatisfactory : thoſe ofthe ſimple
relations ſuggeſted in page 68, ſubaffirmative &c. have nothing in
their favor except analogy with the former, and cloſe connexion
with the notation. A little practice in their uſe might ren-
der theſe laſt names available : but it will be adviſable to con-
 On Propofitions. 75
nect them with names more deſcriptive of the meaning, and to
adopt theſe laſt, whether we reject or maintain their ſynonymes.
When X ) Y, the relation of X to Y is well understood as
that of the species to the genus. We may adopt theſe words,
with the understanding that the word ſpecies includes the
extreme caſe in which the ſpecies is as extenſive as the genus.
When X : Y, we may call X a non-ſpecies ofY, and Y a non-genus
of X. When X. Y we may call X an excluſive or excludent of
Y, or elſe a non-participant ; and alſo Y of X. When XY, we
may ſay that each is participant, or non-excluſive, of the other.
When x . y, which means that X and Y together fill up, or more
than fill up, the univere, we may ſay that they are complemental
names. When xy, which only means that X and Y do not be-
tween them contain the univerſe, we may call them non-comple-
mental. We have then

Inconvertibles . Name of X with reſpect to Y.

A X)Y ſpecies, or ſubaffirmative.
OX:Y non-ſpecies, or particular ſubnegative.
1 A' Y)X genus, or fuperaffirmative.
O'Y:X non-genus, or particular ſupernegative.
Convertibles. Name ofX and Y with reſpect to each other.
EX.Y Excluſives, or non-participants, or ſubnegatives .
IXY Non-excluſives, or participants, or particular fubaffir-
E' x.y Complements, or fupernegatives. [matives.
I' xy Non-complements, or particular ſuperaffirmatives.
The following exerciſes in theſe terms, really contain the de-
ſcription of all the ſyllogiſms in the next chapter.
Incluſion in the ſpecies is incluſion in the genus ; and inclufion
of the genus is incluſion of its parts (ſpecies or not).
Excluſion from the genus is excluſion from the ſpecies; and ex-
cluſion of the genus is excluſion of its parts (ſpecies or not).
Incluſion or excluſion ofthe ſpecies is part incluſion or exclu-
ſion of the genus.
When the ſpecies is complemental, ſo is the genus : and when
the genus is not complemental, neither is the ſpecies.
Exclufion from one complement is inclufion in the other.
Complements of the fame are participants.
 76 On the Syllogifm.
Two ſpecies of one genus, are not complements ; neither are
two exclufions from the fame .
The complement of a genus is a non-ſpecies ; and the com-
plement is a non-ſpecies of the non-complement.

CHAPTER V.

On the Syllogifm.
SYLLOGISM is the inference of the relation between
A two names from the relation of each of thoſe names to a
third. Three names therefore are involved, the two which ap-
pear in the conclufion, and the third or middle term, with which
the names, or terms, of the conclufion are ſeverally compared.
The ſtatements expreſſing the relations of the two concluding
terms to the middle term, are the two premises. In this chapter,
no ratio of quantities is confidered except the definite all and the
indefinite fome.
A fyllogifm may be eitherfimple or complex. A fyllogifm is
fimple when in it two ſimple propoſitions produce the affirmation
or denial of a third : or the affirmation of a third, we may ſay,
ſince every denial of one ſimple propoſition is the affirmation
of another. A complex fyllogifm is one in which two complex
propofitions produce the affirmation or denial of a third complex
propofition.
It might be ſuppoſed that we ought to begin with the ſimple
!
fyllogifm, and from thence proceed to the complex. On this
point I have ſome remarks to offer, in juſtification of following
preciſely the reverſe plan.
Hitherto the complex fyllogifm has never made its appearance
in a work on logic, except in one particular caſe, in which it is
allowed to be treated as a ſimple fyllogifm, though most obviously
it is not fo. I allude to the common à fortiori argument, as in
' A is greater than B, B is greater than C, therefore A is greater
than C. ' There is no middle term here : the predicate of the
firſt propofition is ' a thing greater than B,' the ſubject of the
ſecond propofition is ' B.'
Admitting fully that the quality of the premiſes, that which
 On the Syllogifm. 77
entitles the concluſion to be made, as it is ſaid, àfortiori-marks
this argument out as, if anything, ſtronger, clearer, and (could 1

ſuch a thing be) truer, than a ſimple ſyllogifm ; yet it is plain

that the very additional circumſtance on which this additional
clearneſs depends, takes the argument out of a fyllogifm, as de-
fined by all writers. By beginning with the complex fyllogifm,
and thence deſcending to the ſimple one, it will be ſeen that we
begin with cafes which preſent this àfortiori and clearer charac-
ter. I think I ſhall ſhew that the complex fyllogifm is eaſier
than the fimple one.
Next, the fyllogifm hitherto confidered has never involved any
contrary terms ; the conſequence of which has been that various
legitimate modes of inference have been neglected. Moreover,
ſeveral of the uſual fyllogifms are more ſtrong than need be in
the premiſes, in order to produce the conclufion. Thus Y)X
and Y)Z being admitted as premiſes, the neceſſary conclufion is
XZ. But if Y)X be weakened into YX, the fame conclufion
follows. If we call a ſyllogiſm fundamental, when neither of its
premiſes are ſtronger than is neceſſary to produce the conclufion,
it is obvious that every fundamental ſyllogiſm which has a parti-
cular premiſe, gives at least as ſtrong a conclufion when that
particular is ſtrengthened into a univerſal. But, except when
ſtrengthening the premiſe alſo enables us to ſtrengthen the con-
cluſion, in which caſe we have a new and different fyllogifm, it
ſeems hardly ſyſtematic to mix with fundamental arguments ſyl-
logiſms which have quality or quantity more than is neceſſary for
the conclufion .
The uſe of the complex fyllogifm will, as we ſhall fee, give
an independent and ſyſtematic derivation to theſe ſtrengthened
fyllogifms , as well as to the reſt.
Let X and Z be the terms of the concluſion ; and let Y be
the middle term. Let the premiſe in which X and Y are com-
pared come firſt of the two. Let the order of reference in each
cafe be that of the alphabet
XY YZ XZ

So that by ſtating what X is with reſpect to Y, and what Y is

with reſpect to Z, our ſyllogiſm involves the ſtatement of what
X therefore muſt be, or therefore cannot be, with respect to Z.
We can, in every caſe, expreſs the reſult in ſimple words. Thus,
78 On the Syllogifm.
one of our fyllogifms being what I ſhall repreſent by DiD.D. is
as follows. If X be a ſubidentical of Y, and Y a ſubidentical of
Z, then X is a fubidentical of Z. But all this merely amounts
to the following ' A ſubidentical of a ſubidentical is a ſubidentical.'
We have then to examine every way in which Dior D' or C.
or C' can be combined with Da or D' or C, or C', giving fixteen
caſes in all, and all concluſive in one way or the other. Instead
oftaking an accidental order, and afterwards claſſifying the re-
ſults, it will be better to predict the order which will give claffi-
fication. That order will be to take 1. a D followed by another
of the ſame prepoſition 2. a C followed by another of different
prepofition 3. a D followed by another of a different prepofition.
4. a C followed by another of a like prepofition. This arrange-
ment gives us
1. DADA D'D' DC D'C' 3. D.D' D'D, DC' D'C
2. CD' C'D CC' C'C 4. CD. C'D' CC. C'C'
Each oftheſe caſes will be examined by a method fimilar to that
propoſed in page 61. But a clear perception ofthe meaning of
the words will at once dictate the fixteen reſults, which are as
follows, preceded by the mode in which the fyllogifms are to
be expreſſed.
DDD Subidentical of ſubidentical is ſubidentical.
D'D'D' Superidentical of ſuperidentical is ſuperidentical.
DC.C. Subidentical of ſubcontrary is ſubcontrary.
D'C'C' Superidentical of ſupercontrary is ſupercontrary.
CD'C. Subcontrary of ſuperidentical is ſubcontrary.
C'DC' Supercontrary of ſubidentical is ſupercontrary.
C.C'D. Subcontrary of ſupercontrary is ſubidentical.
C'CID' Supercontrary of fubcontrary is ſuperidentical.
D.D': C' Subidentical of ſuperidentical is not ſupercontrary.
D'D₁: C. Superidentical of ſubidentical is not fubcontrary.
D.C':D' Subidentical of ſupercontrary is not fuperidentical.
D'C :D. Superidentical of fubcontrary is not fubidentical.
CD:D' Subcontrary of ſubidentical is not fuperidentical.
C'D':D. Supercontrary of ſuperidentical is not ſubidentical.
CC: C' Subcontrary of fubcontrary is not fupercontrary.
C'C':C. Supercontrary of ſupercontrary is not fubcontrary.
 On the Syllogifm. 79
In the denials, the extreme limit is included: in the affirma-
tions it is not. Thus ' not fuperidentical' and ' not ſubidentical'
both include ' not identical ;' and the ſame of contraries. In the
affirmations, extreme limitation of one premiſe does not alter the
conclufion : but that of both reduces the conclufion to its extreme
limit. Thus

Subcontrary of identical is ſubcontrary.

Contrary of fuperidentical is ſubcontrary.
Contrary of identical is contrary.
and ſo on. The rules of this ſpecies of ſyllogiſm are as follows.
For affirmatory concluſions ;-(1. ) Like names in the premiſes give
D in the concluſion, and unlike names C. (2.) D in the firſt
premiſe requires premiſes of the ſame prepofition ; C in the firſt
premiſe, of different prepoſitions. (3.) The prepofition of the
conclufion agrees with that of the firſt premiſe. For negatory
conclufions, the preceding rules are reversed. Theſe rules will do
for the preſent, but they afterwards merge in others.
The fixteen forms of complex conclufion above given are of
the clearness of axioms, as soon as the terms are diftinctly appre-
hended. The following diagrams will aſſiſt, and ſhould be uſed
until the propoſitions ſuggeſt their own meaning. Though there
be four, yet theſe four are really but one, as will be ſhown.
1 2 3
X X
DDD Y Y D'D'D'
Z Z

2 3

X X

DCC Y Y D'C'C'
Z Z

1 2 3
X X

CD'C Y Y C'D,C'
Z Z

1 2 3
X
CC'D Y Y C'CID'
Z Z
80 On the Syllogifm.
In each diagram are three lines, partly thick and partly open :
theſe are meant tobe laid over one another, but are kept ſeparate
for diftinctneſs. A point on the firſt line ſignifies a X or ax ;
and one on the ſecond or third, a Y or a y, and a Z or az .
The univerſe of the propoſitions is ſuppoſed to be the whole
breadth. Points which come under one another are ſuppoſed to
repreſent the fame object of thought, variouſly named. Thus
in the firſt diagram, when the thick lines contain the points
named X, Y, and Z, it is ſhown that we mean to ſay there are
objects to which all the three names apply : for there are points
under one another in the thick part of all the three lines.
When we read by the letters on the left, the thick lines are
meant to repreſent the parts in which the Xs, Ys, and Zs muſt
be placed : and when by thoſe on the right, the open lines.
Accordingly, looking at the third diagram, and at the left, we fee
CD'C: while in the diagram, it is clear that X is a fubcon-
trary of Y, or that X. Y and xy ; and that Y is a fuperidentical
of Z , or that Z ) YandY : Z. And the conclufion is equally
manifeſt, namely, that X is a fubcontrary of Z. But, looking at
the left, and ſeeing C' D₁ C', we take the open parts to repreſent
the ſpaces in which Xs, Ys, and Zs are found, and the thick
parts for thoſe in which xs, ys, and zs are found. Here then we
ſee that X is a ſupercontrary of Y, that Y is a ſubidentical of Z,
and that, confequently, X is a ſupercontrary of Z.
Some attempts at laying down the premiſes ſo as to evade the
conclufions, will be inſtructive to any one who does not imme-
diately ſee the latter. And formal demonſtration is always prac-
ticable. Thus if X be a ſubcontrary of Y, that is, if X and Y
do not fill the univerſe, and have nothing in common ; and if Y
be a ſuperidentical of Z, or entirely contain Z, without being
filled by it : then it is clear that X must be more a ſubcontrary
of Z than of Y, by all the inſtances which there are of a Y not
being a Z. The diagram, however, is ſo much clearer than this
fort of demonſtration, that the reader, until he has great com-
mand of the language, may as well look to the former to fee that
he is right in the latter.
It may be convenient, as a matter of language, to ſpeak of a
name as a kind of collective whole, conſiſting of inſtances. And
thus we may talk of one name being entirely in another, or
partly in and partly out &c, as in fact we have already done.
 On the Syllogifm. 81

All the complex fyllogifms which conclude by affirmation are

obviously of the à fortiori character : I ſhould rather ſay, thoſe
of the first three diagrams properly and obviously, thoſe of the
fourth by an eaſy extenfion of language. The marks 1 2 3 in the
middle of the diagrams ſhow how this is. In the firſt, on the
left, X is more of a ſubidentical of Z than it is of Y : the in-
ſtances in which itsfub-identity appears confift of all thoſe which
prove the ſubidentity of X to Y, together with all thoſe which
prove the ſubidentity of Y to Z. In the third, read from the
right, X is more fupercontrary to Z than it is to Y, by all the
inſtances which ſhow the ſubidentity of Y to Z. In the fourth
diagram (from the left) we cannot ſay that X is more ſubidentical
of Z than of ſomething elſe, ſimply becauſe there is no previous
ſubidentity among the relations. But ſtill the diftinguiſhing
characteriſtic of the concluſion takes its quantity from the addi-
tion of thoſe of both the premiſes .
If either of the premiſes be brought to the limit which ſepa-
rates it from the relation of an oppoſite prepoſition ; that is, if
C' or Cı be changed into C, or elſe D'or Da into D : the nature
of the conclufion is not altered, except by the loss of the àfor-
tiori character. One of the quantities which have hitherto con-
tributed to the quantity of the conclufion, now disappears. Thus
CD gives C₁ as well as C. D' ; and CD' gives Ci as well as
C. D' ; C. C gives D, as well as C. C' .
Let one of the premiſes paſs over the limit, and take the oppo-
ſite prepoſition. Chooſe C. D', which gives Cı, and continues
togive it, though weakened, when the firſt C,becomes C. Then
let Ca become C' : ſo that our premiſes are C' D' . The dia-
gram is then as follows
2 1 3

X
C'D' Y
Z

The quantity of the conclufion now depends upon the differ-

ence between the number of inſtances in (12) and (23) and its
quality upon whether (12) has fewer inſtances than (23), or the
ſame number, or more. As I have drawn it, C₁ is the conclufion,
ſtill : ſtrengthen the firſt premiſe ſtill more, and the conclufion
G
 82 On the Syllogifm
will paſs through C into C' or elſe into P, and in the ſecond
cafe may paſs into D', as in the following diagram
X
C'D' Y
Z

Nothing is impoſſible except Da or D. Hence C' D' enables

us only to deny D. and its limit D. Treat the other caſes in the
ſame manner, and, remembering that denial is to include denial
up to the limit (while affirmation only affirms to any thing ſhort
of the limit) we have
D. D' denies C' D' Da denies C
D. C' : D' D' C • D
C D D' C' D' D
C C • C' C' C' •
C

The rules given above in page 79 may be collected from the

instances .
As long as we keep contraries out of view, the ultimate ele-
ment of inference is of a twofold character. It is either ' X and
Z are both Y ; therefore X is Z ' or elſe X is Y and Z is not
Y ; therefore X is not Z ' : X, Y, Z, being ſingle inſtances of
three names ; and Y the fame inſtance in both premiſes. But the
uſe ofcontraries enables us to give an affirmative form to the latter
cafe. It is ' X is Y, and not-Z is Y' ; therefore ' X is not-Z'.
Connected with this change of expreffion is the following
theorem : that all the eight affirmatory complex ſyllogiſms are
reducible to any one among them : and the ſame of the negatory
ones. The reader may trace this theorem to the order of the
figures 1 , 2, 3, being the ſame in all the four diagrams. Taking
DDD as the moſt ſimple and natural form, and looking at the
diagram of CD'C , we see the last as DiDiD in ' X is ſubi-
dentical of y ; y is ſubidentical of z ; therefore X is ſubidentical
of z .' If we write the terms of the ſyllogiſm after its defcriptive
letters, as in DDD ( XYZ) we have the following reſults ;-
DDD (XYZ) = DDD (XYZ) | D'D'D' (XYZ) = DiDiDi (xyz)
DCC ( XYZ) = DDD (XYz) D'C'C' (XYZ)= DDDı (xyZ)
CD'C. (XYZ) = DDD (Xyz) C'DC' (XYZ) = DDD (XYZ)
CC'D (XYZ) = DDD (XYZ) | C'CD' (XYZ) = DDD (xYz)
 On the Syllogifm. 83
Thinking of the firſt deſcription only as to relations, and of the
ſecond only as to terms, we ſee the following rules of connexion.
In the firſt and ſecond premiſes and terms, there are X and Y
in the terms, or their contraries, according as there are ſub-
accents or fuperaccents in the relations. But in the concluſion,
the term is Z for D. and C', z for D' and C₁. And we may
thus reduce any ſyllogiſm involving any one ofthe eight varieties
of relation combined with any one of the varieties of terms,
either to DDD or to XYZ. Thus CD'C. (XyZ) is DIDD
(XYz), or DCC (XYZ). Not to load the ſubject with de-
monſtration of forms, I will give at once the general rules by
which changes of accent and letter are governed : remarking
that they apply throughout the whole of my ſyſtem.
The varieties in queſtion are eight :
XYZ, xyz ; XYZ, Xyz ; XyZ, xYz ; XYz, xyZ .
in which (thinking of XYZ) all are kept ; or all changed ; or
one only kept ; or one only changed. Learn to connect each
letter with the propoſitions in which it occurs ; marking the pro-
pofitions, premiſes and conclufion, as 1, 2, 3. Connect X with
1,3 ; Y with 1,2 ; Z with 2,3. Keeping all, or changing all,
makes no alteration of letters : keeping only one, or changing
only one, alters the letters in the premises in which that one
occurs. Thus, be the accents what they may, if in DDD we
change only the firſt letter into its contrary, the fyllogifm becomes
CDC ; and the ſame if we keep only the firſt letter unchanged.
As to accents, remember that change of Z produces no effect :
look then only at X and Y. When either letter is changed into
its contrary, change the accents belonging to the premiſes in
which that letter comes firſt ; 13 for X, 2 for Y, 123 for XY.
For example, what is CC'D. (Xyz). Here, as to letters, X
alone ( 1,3) is unchanged : then CCD becomes DCC. As to
accents, Y is changed, which comes firſt only in 2 : change C'
into C1. Hence CC'D (Xyz) = DCC (XYZ). Here we
have paſſed from a ſyllogiſm in Xyz to the correſponding equi-
valent in XYZ : the rules equally hold for the inverſe proceſs,
and for all combinations of letters. For the change of XYZ
into Xyz, and that of Xyz into XYZ, have only one deſcription :
the firſt only left unchanged. Now ſuppoſe it required to know
84 On the Syllogifm.
what ſyllogiſm in xYz anſwers to DCC(Xyz). The key words
are, the third only unchanged. Alter then DCC into DDD by the
A
firſt rule, and change all the accents. Thus DiCC (Xyz) =
D'D'D'(xYz). The independent rules are that change of fub-
ject only, changes both letter and accent ; predicate only, letter ;
ſubject and predicate, accent. Thus to find what D'C'C'(xYz)
is, expreſſed in XYz, the changes are, in the three premiſes S,
neither, S, and D'C'C'(xYz)=C,C'D (XYZ). The following
table may be verified for exerciſe : it ſhows the effect of all
changes except that of the middle term.
XYZ XYZ XYz xYz
DDD C'DC' DCC C'CID'
C'DC' DDD C'CD' DCC
DCC C'CID' DDD C'DC'
C'CID' DCC C'DC' DDD

Similarly, D'D'D' would have CD'C, D'C'C' &c. When

the middle term only is changed, the table may ſtand thus ;-
XYZ DDD C'D.C' DCC C'CID'
XyZ CC'D D'C'C' CD'C D'D'D'

It will of courſe have been obſerved that the eight fyllogifms

go in pairs, each one of a pair differing from the other in accen-
tuation, and nothing else. When we take ſets of four, the ones
put together ſhould be thoſe in which the firſt premiſe, or the
ſecond, or the conclufion (whichever we take for a ſtandard)
has D. and C', or elſe has D' and C..
The ſame rules of transformation apply to negatory complex
fyllogifms ; thus D'D₁: C (XYZ) is C'D':D.(Xyz). In fact theſe
rules do not depend upon the character of the inference, nor even
upon its validity, but merely on the effects produced in the ſingle
propofitions by changes of term. Thus the ſtatement D'D.C.
(XYZ), an invalid inference, is the ſame ſtatement (equally in-
valid of courſe) as is expreſſed in D.C'D'(xyZ).
An examination of the complex particular relation P= I + I'
+ O + O', whether by the diagram or by unaſſiſted thought, will
ſhow that when this relation exiſts between X and Y, it alſo exiſts
between x and Y, X and y, x and y. Hence PC, CP, PD, DP,
 On the Syllogifm. 85
give P. Moreover, two complex particulars give no poſſibility
ofany conclufion, all being equally poſſible. Thus PP may give
C or C or C', or D or D or D'.
Now combine one of the others, as Di, with P : examine PD
and D.P. It will be found that the complex particular of a fub-
identical may be either complex particular, ſubidentical, or ſuper-
contrary ; or that PD, may be either P, D₁ or C'. Examine all
the cafes, and the rules will be found in
(D.C.)P P(D.C')
(D'C' )P P(C.D' )
thus interpreted. Either premiſe from between the parentheſes,
with P, in order as written, may have either, and must have one,
of the three for its conclufion. That DP must give either D
C or P, and ſo muſt CAP : but PC, must have either P, Cı, or D'.
Before proceeding to the fimple ſyllogiſm, as I have called it,
I will ſtate that I much doubt the propriety of the terms fimple
and complex. Undoubtedly the phrases are hiſtorically juſt, for
each of the fyllogifms which I propoſe to call complex is, as
we shall fee, neceſſarily composed of three of thoſe which are
always called simple. But in another point of view, the phrase-
ology ought to be reverſed ; the ſimple ſyllogiſm is the affirma-
tion of the existence of one out of ſeveral of the complex ones.
Thus X)Y+Y)Z=X)Z, or AAA, is really (Di or D, not
known which) (D. or D, not known which) (D, or D, not
known which) and afferts that there is either DaDaD or DDD
or DDD or DDD.
But it will be faid, ſurely the complex propoſition requires the
conjunctive exiſtence of two ſimple ones : D=A+0' ; and is
therefore compound at least. I anſwer that, on the other hand,
the fimple propoſition requires the disjunctive existence of two
complex ones : as A=D or D. Which is moſt ſimple, both, or
one or the other ? to me, I think, the firſt. Certainly the fyllo-
giſm DDD is one which I more readily apprehend than AAA .
Indeed, to moſt minds, the latter is the former, if they are left
to themſelves : and the cases DDD1, &c. are only admitted when
produced and inſiſted on.
But further, is the ſimple propofition properly called simple?
Is there in it but one aſſertion to deny or admit ? Is but one
86 On the Syllogifm.
queſtion anſwered ? When I affirm ' Every X is Y,' I affirm
1. Comparison of X and Y. 2. Coincidences. 3. The greatest
poſſible amount of them. 4. That every X has been uſed in ob-
taining them. In ' Some Xs are Ys the firſt two of the preced-
ing are employed. In ' No X is Y,' we have, 1. Comparison
of Xs and Ys. 2. Excluſions. 3. The greatest amount. 4. The
comparison of every X with every Y. And Some Xs are not
Ys' omits the third, and ſubſtitutes Xs for every X in the fourth.
Now the ſubidentical, for instance, only contains, beſides what
is in the ſubaffirmative, the notion that there are more Ys than
Xs in exiſtence. The ſubcontrary conſiſts, over and above what
is in the fubnegative, in that Xs and Ys are not every thing that
the propofition might have applied to : and ſo on. On theſe
confiderations, I think it may be allowed to treat the words fim-
ple and complex as only of hiſtorical reference, and to confider
the firſt as disjunctively connected with the ſecond, the ſecond
as conjunctively connected with the firſt, in the manner above
noted. I think I ſhall make it clear enough, that the paſſage
from the conjunctions to the disjunctions is better ſuited to a
demonſtrative ſyſtem than the converſe. If the plan which I
propoſe ſhould gain any reception, I ſhould imagine that disjunc-
tive and conjunctive would be the names given to the claſſes
which I have called ſimple and complex : the conjunctive com-
poſed of ſeveral of the disjunctive, the disjunctive confifting of
one or the other out of ſeveral of the conjunctive.
When a propoſition R, is the neceſſary conſequence of two
others, P and Q, it neceſſarily follows that the denial of R, muſt
be the denial of one at least of P and Q. For every propofition
admits but of affirmation or denial : and he who affirms both P
and must affirm R. If then P be affirmed and R denied, the
denial of Q must follow : if Q be affirmed and R denied, the
denial of P must follow.
Afimple ſyllogism is one, the two premiſes and conclufion of
which are to be found among the ſimple propoſitions A., Ει, Ιι,
Οι, Α', Ε', Ι', O'. Thus we have AEE or X)Y + Y.Z =
X.Z, as an inſtance. The order of reference is always XY,
YZ, XZ .
The following theorems will be neceſſary ;-1 . A particular
premise cannot befollowed by a univerſal conclufion.
 On the Syllogifm. 87
If poffible, let AI,for example, have a univerſal conclufion.
Take the complex premiſes DAP or (A1 + O')( I + I' + O + O').
All that can be inferred is that one of three conclufions (page 85)
is valid, and neither D nor C : either Da or Por C1. But if a
univerſal be true, one of two concluſions must be valid (page 69)
and one of them D or C. If then A. and I alone yielded a
univerſal conclufion, quite as much muſt DP : or a form which
is indifferent to three conclufions, and not having D nor C, is ne-
ceſſarily productive of one of two conclufions, one of which is
D or C. This contradiction cannot exiſt : or AI cannot yield
a univerſal conclufion .
2. From two particular premiſes no concluſion can follow.
If poſſible, let II.yield a concluſion ; which by the laſt the-
orem, muſt be only particular. Now PP or ( Ii + I' + O + O')
(I +I' + O + O') is indifferent to all complex conclufions : quite
as much is II . But if theſe premiſes yield a particular conclu-
ſion, two complex concluſions are denied (page 69). This con-
tradiction cannot exiſt : or particular premiſes can yield no
conclufion.
Let a ſimple ſyllogiſm with premiſes and concluſion all univer-
fal, be called univerfal: and with either premiſe (and therefore
the conclufion) particular, be called particular. Then every
univerſal fyllogifm has two particular ſyllogiſms deducible from it.
Thus if AEE be valid, then A joined with the denial of E
gives the denial of E₁ : or AII ſeems to be valid. But the altera-
tion of the places of the propoſitions requires us to ſay that it is
A'II.which is valid : and this point requires cloſe attention.
Take AEE or X) Y + Y.Z = X.Z . Then X)Y with the
denial of X.Z(or XZ) gives the denial of Y.Z(or YZ) ; and
we have
X) Y + XZ = YZ
This is valid, if the firſt be (as it is) valid : but its ſymbol is not
AII. For the middle term is, in our notation, made middle in
the order of reference, which is therefore YX, XZ, YZ : and
the ſyllogiſm is ALI. Similarly we have
XZ + Y.Z = X : Y

produced by coupling the denial of X.Z with Y.Z. But this is

LEO : for the order of reference is now XZ, ZY, XY, and
88
On the Syllogifm.
E.is not changed by change of order. The rule is as follows.
When the denials ofthe concluſion and of a premiſe are made to
take the places of that premiſe and the conclufion, the order of
reference remains undisturbed as to the tranſpoſed terms, and is
changed as to the ſtanding term. This laſt muſt therefore have
the prepofition of the inconvertible propofition changed ; but
not that of the convertible propoſition.
Thus E'A.E', if valid, gives E'I'O, and I'A'I'. Again, in a
ſimilar way it may be ſhown that from each particular fyllogifm
follows a univerſal : thus IE'O', if valid, ſhows that denial of
O' , and E', give denial of I or A'E'E.. In this caſe neither is
valid. And E'I'Ο., befides E'A,E', alſo gives AI'I'.
Such claſſification of theſe opponent forms as is useful, will pre-
ſently be given.
Since there are eight forms of afſſertion, with reference to each
of the orders XY YZ, it follows that there are fixty-four com-
binations of a pair of premiſes each. But of theſe the only ones
which have a chance of yielding a concluſion are, 1. fixteen
with premiſes both univerſal ; 2. thirty-two with one univerſal
and one particular. If, for a moment, U ſtand for univerſal and
P for particular, the form of a ſyllogiſm is either UUU, PUP,
UPP, or UUP. Of theſe, the firſt, ſecond, and third are ſo
related that each form has the other two for its opponents : but
the fourth has its own form in each of its opponents.
Now examine one of the complex affirmative ſyllogiſms, ſay
DiDiDi, by the diagram in page 79. The premiſes are A + O'
and A + O', giving the four combinations AA , AO', O'A, and
O'O'. The coneluſion is A + O' : but it is not merely twofold,
but threefold : for the àfortiori character explained in page 81,
ſhows that O' is obtainable on two different grounds, and is the
ſum, as it were, of two different and neceſſary parts of the con-
clufion. That every X is Z, follows from X)Y and Y)Z, or
we have the fyllogifm.
AAA X) Y + Y)Z = X)Z
But as far as the Zs which are below ( 12) are concerned, it
follows that they are not Xs becauſe they are the Ys which are
not Xs : or we have
O'AO' Y: X + Y)Z =Z : X
 On the Syllogifm. 89
and as to the Zs below (23) they are not Xs becauſe they are
not Ys, among which are all the Xs. Accordingly we have
AO'O ' X)Y + Z:Y = Z : X
or DDD requires the coexiſtence of AAA , O'AO', A.O'O'.
Apply this reaſoning to the contraries x, y, z, or elſe examine
D'D'D' in the fame way, and we find that D'D'D' requires the
coexiſtence of A'A'A', OA'OΟΙ, Α'ΟΙΟΙ.
By applying the preceding reſults to x, Y, Z, &c. as in page
82, or, as is better at firſt, by examining all the caſes of the dia-
gram in page 79, we get the following table of derivations from
the eight affirmatory complex fyllogifms. The firſt column
ſhews the terms which must be uſed, to deduce all from DDD
AAA X) Y + Y)Z = X)Z
XYZ DDD... O'AO' Y: X + Y)Z = Z: X (12)
AO'O' X)Y + Z : Y = Z : X (23)
A'A'A' Y)X + Z) Y = Z )X
xyz D'D'D' . . OAOX : Y + Z ) Y = X : Z ( 12 )
A'OO Y)X + Y : Z = X: Z ( 23)
E'AE' x.y + Y)Z = x.z
xYZ C'D.C' .. {
IAL XY + Y)Z = XZ ( 12)
LE'O'L x.y + Z :Y = XZ (23)
EAE X.Y + Z ) Y = X.Z
Xyz CD'C... I'A'I' xy + Z)Y =xz (12)
EOI' X.Y + Y: Z =xz (23)
AEEX) Y + Y.Z = X.Z
XYZ DCC... O'EI' Y: X + Y.Z =xz (12)
AI'I' X)Y + yz = xZ (23)
A'E'E' Y)X + y.z = x.z
xy Z D'C'C' .. OEI X : Y + y.z = XZ (12)
Α'ΙΙ Y)X + YZ = XZ ( 23)
E'EA' x.y + Y.Z = Z)X
xYz C'CD' . LEO XY + Y.Z = X :Z ( 12)
E'I'O. x.y + yz = XZ (23)
EE'A. X.Y + y.z = X)Z
Ι'Ε'Ο'
XyZ CC'D...
{ ELO' xy +y.z = Z : X
X.Y + YZ = Z: X
( 12)
(23)
Before forming any rule, or making any remark, I proceed to
१० On the Syllogifm.
collect the reſults of the remaining caſes. And firſt, let a pre-
miſe be brought to its limit, D or C : ſay that DDD becomes
DD.D. In the diagram it immediately appears that one of the
particular concluſions is loſt ; not contradicted, but nullified : for
(12) diſappears, becauſe X and Y are identical names. That is,
AAA remains, and AO'O' : but the concluſion of O'AO' is
nullified. But this very circumſtance creates, not a new conclu-
ſion, for it is only a part of one already exiſting, but a new form
of deduction. The premiſes are now A +A' and A1+ O' , and
the conclufion is A1+ O'. The ſyllogiſms AAA and AO'O'
are as before, and for the fame reaſons : but there is now the
combination A'A, among the premiſes, which produces the con-
clufion Iı, and we have
A'AI Y)X + Y)Z = XZ
This fyllogifm, though new as far as DDD is concerned, is
only a ſtrengthened form of IAI, a concomitant of E'AE' .
For (page 65) I is true whenever A' is true, ſo that A'A, in-
cludes IA and its neceſſary conſequence I. But if I had been
ſtrengthened into A instead of A', we should have had AAI
which though perfectly valid, yet admits of a ſtronger conclu-
ſion, as ſeen in AAA .
Of the two modes of ſtrengthening a particular propofition
(as I into A or A') there is one which ſtrengthens the quantity
of the firſt form of the propofition, and another that of the ſecond.
Thus XY or I becomes X)Y or A.when the firſt form, and
Y)X or A' , when the ſecond form, is ſtrengthened. Similarly
O or X :Y becomes X.Y or E., and y.x or E' , according as
the form ſtrengthened is X:Y or y:x. The prepoſition remains
the fame, or changes, according as the firſt or ſecond form is
ſtrengthened. If the first form of thesecond premiſe of a fyllo-
giſm, or thesecond form of the first premiſe, be ſtrengthened, no
ſtrength is added to the concluſion. Thus, as far as the fyllo-
giſms in this chapter are concerned, I.A gives as much as A'Aı,
and EO as EE.. But if the firſt form of the firſt premiſs, or
the ſecond form of the ſecond, be ſtrengthened, the conclufion
has its firſt form ſtrengthened.
A very fimple and obvious theorem contains all theſe reſults.
The concluding terms are, in our order of reference, the firſt
 On the Syllogifm. 91
term of the firſt premiſe and the ſecond term of the ſecond. The
conclufion is never ſtrengthened by augmenting the quantity of
the middle term, nor only weakened (it may be altogether de-
ſtroyed) by weakening the middle term. A wider field of com-
pariſon does not by itſelf give more comparisons : nor can more
compariſons ariſe except by augmenting the number of things
compared in that field. Since the conclufion can obviously
ſpeak of no more than was in the premiſes, no term of that con-
clufion can be augmented in quantity, until the ſame thing has
taken place in its premiſe. But no ſtrengthening of a propofi-
tion ſtrengthens both terms : conſequently, to make ſuch a thing
effective, it must be the concluding, and not the middle, term
which is ſtrengthened.
The following table is only worth inſerting as a collection of
exerciſes. The fourth column ſhows the eightstrengthened par-
ticularfyllogifms, as I will call them, having univerſal premiſes but
only a particular conclufion, not ſtronger than might have been
inferred from the particular fyllogifm itſelf.
Alteration and fub- ſtrengthened occurring
of into removes ſtitutes from in

DDD DDD O'AO' A'ALLAL C'D.C'

D'D'D' D'DD' A'OO A'LI D'C'C'
DDD DDD AO'O') AA'I' I'A'I' CD'C
D'D'D' DD'D' OLAOS AI'I' DCC
DCC DCC O'EI' A'EO LEO ) C'CD'
D'C'C' D'CC' A'LI A'OOS D'D'D'
DCC DCC AI'I' ΑΕ'Ο' Ι'Ε'O'CC'D
D'C'C' DC'C' OLE'I AO'O'S DD.D.
CD'C CD'C, I'A'I' Ε'Α'Ο, ΟΙΑ'O) D'D'D'
C'D.C' C'DC' E'O'I E'I'O. C'C.D'
CD'C CDC EOI' EAO' O'AO' DDD.
C'DC' CDC' IAI ELO'CC'D.
CC'D CC'D. I'E'O' ) Ε'ΕΙ OEI D'C'C'
C'CID' C'CD' E'I'O E'O'I, C'D.C'
CC'D CCD ELO' EE.I' O'EI' D.C.C
C'CID' CCD' LEO EOI'S CD'C,

Iwill now examine the negatory complex fyllogifms, premiſing

however than we cannot get any new concluſions from them.
 92 On the Syllogifm.
For we have now got all the fixteen caſes in which both pre-
miſes are univerſal : and we know that there can be no fyllogifm
with a particular premiſe, except it have one of thoſe with uni-
verſal premiſes for its opponents.
Take D.D': C' or A1 + O' and A' +O, together deny E' + Iı,
that is, deny the coexistence of E' and I , that is, deny either E'
or I , that is, affert either I' or E.. This ſyllogiſm then may be
written thus,
(A + O ') (A' + 1) (either E. or I')

Now the fact is that this disjunction is ſuperfluous ; it is I'

which is always aſſerted, and E is never a neceſſary conſequence
of D.D'. For AA' gives I' as already ſhown, and AO and
O'A' are inconcluſive (and O'O, of courſe). And the rationale
of the inference is as follows : ſince X is a ſubidentical of Y, and
Y a ſuperidentical of Z, it follows that Y is ſuperidentical both of
X and Z ; conſequently, Y not filling the univerſe (our ſuppo-
ſition throughout) it follows that there are things which are nei-
ther Xs nor Zs, namely, all which are not Ys. Again, in
CC: C', which the ſame reaſoning ſhows to be only C.C.I',
none either of X or of Z is in Y, therefore every inſtance in Y
is both x and z. And thus it will appear that in every negatory
complex concluſion the whole middle term, or the whole of its
contrary, makes the ſubject matter of the ſtrengthened particular
fyllogifm which is all that can be collected.
Our conclufion is that no negatory complex fyllogifm is of
any more logical effect than the ſtrengthened particular derived
from it. Thus we may say that, ſo far as the extent and cha-
racter of the inference is concerned, the former is the latter.
I will now paſs to the general rules of the complete ſyſtem of
fyllogifms ;-
The reader muſt take pains to remember two rules of forma-
tion, perfect contraries of each other, for the dependence of the
accents (or prepositions) on the ſign (affirmative or negative cha-
racter) of the firſt premiſe. I expreſs them in the briefeft way
poffible.
Direct Rule. Affirmation (in the firſt premiſe) makes the ſecond
premise agree with both the other propoſitions, or iſolates no-
thing: negation makes the ſecond premiſe differ from both the
 On the Syllogifm. 93

others, or iſolates the ſecond premiſe. Inverſe rule. Affirmation

isolates the first premise, makes the firſt premiſe differ from both
the others in prepofition : negation iſolates the conclufion, makes
the conclufion differ from both the others. Theſe rules might
be expreſſed ſo as to make their contrariety more complete.
direct like
Thus in theinverſe rule,affirmative commencement ſhows unlike
prepoſitions in the two premiſes, and the conclufion differing
agreeing
with the firſt premiſe in prepofition : but negative commence-
from

ment ſhows unlike

like prepoſitions in the two premiſes, and the con-
clufion agreeing with the firſt premiſe in prepofition.
differing from
The ſubjects of the following rules are,
1. The eight affirmatory complex fyllogifms.
2. The eight univerſal ſimple fyllogifms.
3. The eight ſtrengthened particular ſimple ſyllogifms.
4 The fixteen particular fimple ſyllogiſms .
Omit the negatory complex fyllogifms, as fully contained in
the third of this enumeration, and the complex fyllogifms which
contain the unaccented D or C, as carrying a momentary accent
for the rule, to be expunged when the formation is completed.
Confider Dı, D, D', A1, A', I1, I' , as of the affirmative ſigns, and
C1, C, C', Ει , Ε', Οι, O', as negative.
Rule 1. In the complex ſyllogiſm all parts are complex ; in
the univerſal ſimple ſyllogiſm all parts are univerſal ; in the
ſtrengthened particular only the conclufion is particular ; in the
particular only a premiſe is univerfal.
Rule 2. Premiſes of like ſign have an affirmative conclufion ;
of unlike ſign, a negative.
Rule 3. The complex, the univerſal, the particulars which
begin with a particular, follow the direct rule ; the ſtrengthened
particulars, and the particulars which begin with a univerſal (all
that commence with a univerſal, and conclude with a particular)
follow the inverſe rule. [Or thus ; all which begin and end
alike, follow the direct rule ; all which begin and end differently,
the inverſe. ]
The complex fyllogiſms and univerſals are easily remembered
94 On the Syllogifm.
by rule : the particulars almoſt as easily. The following fub-
rules may be noted, as far as theſe laſt are concerned.
Sub-rule 1. First and second premises. A and O in the firſt
premiſe demand unlike prepoſitions in the two premiſes : E and
I demand like prepoſitions. Thus AO must be inconcluſive :
AO' must be concluſive. But EO must be concluſive : and
EO' must be inconclufive.
Sub-rule 2. First premiſe and concluſion. A univerſal in the firſt
premiſe demands an unlike prepoſition in the concluſion : a par-
ticular firſt premiſe, a like prepoſition in the conclufion.
Sub-rule 3. Second premiſe and conclufion. Every ſecond pre-
miſe demands its own prepoſition in a concluſion of like ſign :
and the other prepoſition in a concluſion of unlike ſign.
As far as the four ſpecies are concerned, every fyllogifm
formed according to the three rules is valid ; and every one
not fo formed is invalid. The following remarks are partly
recapitulatory, partly new.
Remark 1. Every complex fyllogiſm gives one univerſal ſyllo-
giſm * and two particular ones, its concomitants : and the con-
comitants are formed by changing one of the premiſes of the
univerſal and the conclufion, into their particular concomitant
propoſitions (page 63. )
Remark 2. Every ſyllogiſm has its contranominal, which aſſerts
of the contraries in the ſame manner as the firſt does of the di-
rect terms : and contranominals have all their accents different,
as in O'AO' and OAO. (page 62.)
Remark 3. Every ſyllogiſm has two opponents, made by inter-
changing the contradictories of one premiſe and ofthe conclufion,
and altering the accent of the remaining premiſe, if inconverti-
ble ( A or O) (page 88.)
Remark 4. Every complex ſyllogiſm has two ſuch opponents
formed in the fame way, the Ds being the inconvertibles, the Cs
the convertibles. Thus (:) meaning denial of, the opponents of
CD'C are C₁:C1:D' and :CDC. The firſt of theſe is
(E + I') ( I. or E') ( O' or A₁)
containing the valid fyllogifms E.E'A., ELO', I'E'O' ; being
* Syllogifm, not preceded by complex, means ſimple fyllogifm .
 On the Syllogifm. 95

EE'A, and its concomitants. And :CDC gives E'A.E' (the

contranominal of EA'E.) and its concomitants. And the fame
ofthe reft.
Remark 5. Each univerſal fyllogifm has two weakened forms,
made by weakening one premiſe and the conclufion. When the
first premiſe is weakened, it is without change of prepofition :
but when the ſecond, with change. Thus the weakened forms
of EA'E, are OA'O and EIO .
Remark 6. Each particular fyllogifm has two ſtrengthened
forms, one of which is a univerſal, the other only a ſtrengthened
particular. Thus the ſtrengthened forms of O'O are EAE.
and E'A'O..
Remark 7. In every fyllogifm except the ſtrengthened particu-
lar, the middle term is univerſal in one premiſe, and particular in
the other : and its contrary is therefore the fame. But in the
ſtrengthened particular, the middle term is univerſal in both
premiſes, or particular in both. This affords a complete crite-
rion of fyllogifm, as will be noticed hereafter : in fact, the com-
pleteness of this ſyſtem crowds us with relations, from many of
which general rules might be deduced, though they need only
appear here by caſual remark.
In O'A,O', A'OO , LAI, EOI , O'EI', ALL , LEO ,
EIO' , the middle term enters univerſally in the univerſal, and
particularly in the particular. In all the others it enters particu-
larly in the univerſal, and univerſally in the particular. In the
firſt ſet, the convertible premiſes are all fubs, the inconvertibles
are fubs in the ſecond premiſe, andſupers in the firſt. In the
ſecond ſet, theſe rules are inverted.
Remark 8. Of the twelve poſſible pairs of premiſes AA, AE,
ΑΙ, ΑΟ , ΕΑ, ΕΕ, ΕΙ, ΕΟ, ΙA, IE, OA , OE, which can give
a conclufion, each one will, in two ways, which two ways are
inverted in their accents . Thus EO appears in E'O'I and
EOI' . The two premiſe-letters and one accent dictate all the
reft : thus I'A can belong to nothing but I'A'I' . When the
ſyſtem is well learnt, it will be found unneceſſary to write more
than I'A, for the ſymbol of I'A'I'. I now ſpeak only of funda-
mental fyllogifms : the ſtrengthened ſyllogiſm A.A'I' might be
fignified by A'A'.
Remark 9. The ſyllogiſms of the three firſt claſſes are all really
 96 On the Syllogifm.
ſpecimens of one, thoſe of the fourth of two, among them, with
the eight variations XYZ , XYZ, XYz, xYz, XyZ, xyZ, Xyz,
xyz. The rules for conducting theſe changes are
Change of fubject is change of both accent and letter.
Change of predicate is change of letter.
Change of both is change of accent.

thus to paſs from E'EA' to A.EE, we note in XY change of

ſubject, in YZ change of neither, in XZ change of ſubject :
therefore XYZ is the ſet of terms into which XYZ must be
changed : and the E'E.A' fyllogifm of either ſet is the AEE
fyllogifm of the other.
The 24 fyllogifms, which are 24 with reference to the order
XY, YZ, XZ, are only 12 if the order ZY, YX, ZX, be
allowed . Thus AI'I' of the firſt is the I'A'I' of the ſecond.
Theſe ſyllogiſms are eſſentially the fame in the mode of inference
they afford. To change a fyllogifm into another of the fame mode
of inference, invert the premiſes and change the prepofition of
all the inconvertibles . Thus A'OO and O'AO' are of the ſame
inference. The pairs which in this point of view are identical are
AAA = A'A'A' E'A, E' = A'E'E' E.A'E = AEEE'EA' = EE'A,
O'AO' A'ΟΟΙΙ = A'I IL I'A'I' = AI'I' LEO = ΕΙΟ '
AO'O' = OA'Ο Ε'Ο'Ι = OE'I EOI' = O'E,I' E'Ι'Ο. = Ι'E'O '

The ninth remark admits ofconfiderable extenſion. The 'Some'

of a logical propoſition may have a much more definite character
in ſome caſes than in others. It may be a ſelected, or at least a
diftinguishableſome, which want nothing but a nominal distinction
to make the particular propofition easily and uſefully univerſal.
Whether it can be done more or leſs eaſily, and more or leſs uſe-
fully, is no queſtion of formal logic. Ifit be ſuppoſed done, the
particular is converted into a univerſal. In ' fome Xs are Ys,'
if we make a name for every ✗ which is Y, ſay M, we have
then ' Every M is Y'. This propoſition may be purely identi-
cal, or it may not. If we call every X which is Y by the name
M merely becauſe it is Y, then our univerſal is only ' Every X
which is Yis Y' . But if the name M be conferred from any
other circumſtance, which distinguiſhes the Xs that are Ys from
other Xs, then the change from the particular to the univerſal by
 On the Syllogifm. 97

means of the new reſtriction impoſed by the new name, is the

expreſſion of new knowledge.
The quantities in the conclufion are of two kinds. There are
thoſe which are brought in with the terms, and which continue
in the concluſion ſuch as they were introduced in the premiſes :
and there are thoſe which depend on the union of the premiſes,
and which are what they are only in virtue ofthe joint exiſtence
of the premiſes. For example, in IAI. we have fome Xs are
Ys, but every Y is Z, therefore ſome Xs are Zs' : if we aſk,
what Xs are Zs, the answer is, thoſe which are Ys, and no others,
ſo far as this concluſion affirms . But when we look at O'AO'
or ' fome Ys are not Xs, and every Y is Z ; therefore fome Zs
are not Xs : and if we then aſk what Zs are not Xs ; the answer
is, that this quantity does not enter with Z, but depends upon the
other premiſe, namely, upon the number of Ys which are not Xs .
In a particular fyllogifm, let us call the quantity of the ſubject
in the conclufion intrinfic or extrinfic according as it is that of
the premiſe which introduces that ſubject, or of the other premiſe.
Examination will ſhow that in every particular ſyllogiſm which
concludes in I or I', in which both terms are particular, the
quantities of the terms are, of the one intrinfic, of the other ex-
trinfic : but that where the conclufion is in On or O', either the
quantity ofthe ſubject is intrinsic and that of the contrary of the
predicate extrinfic, or vice verſä.
When the quantity of a particular term in the conclufion is
intrinfic, the invention of a name will convert the syllogiſm into
a univerſal. Thus IAA or XY + Y )Z = XZ, if M be taken
to repreſent all thoſe Xs which are Ys, and nothing elſe, becomes
M) Y + Y)Z = M)Y, of the form AAA. Again, O'A.O ' or
Y: X + Y)Z = Z : X, thrown into the form x:y + z)y = x:z, be-
comes m.y + z)y= m.z, of the form EAE , when the xs which
are ys are diftinguiſhed from the rest of the univerſe by the name
m.
There is nothing either illegitimate or uncommon in diftin-
guiſhing by a peculiar name certain fome (or even uncertainfome,
if certainly always theſameſome) of another name. Again, fince
we know that every univerſal ſyllogiſm is reducible to the form
AAA by uſe of contraries, we have now reaſon to know that
there is no fundamental inference, of the kind treated in this chap-
ter, which is any other than that in AAA , or, the contained
H
98 On the Syllogifm.
ofthe contained is contained. And there is no better exerciſe than
learning to read off each of the ſyllogiſms, univerſal and particu-
lar, into this one form, by perception, and without uſe of rules.
Take as an inſtance X: Y + y.z =XZ : what is the container,
what is the contained, and what is the middle container of one
and contained of the other. It is a parcel ofXs which are con-
tained in y, all y in Z, and therefore that parcel of Xs in Z.
This general principle ſuggeſts a notation for all the complex,
univerſal, and fundamental particular, ſyllogiſms. If we abbre-
viate X)Y + Y)Z = X)Z into XYZ), and if we denote by
XYZ, without ), that it is only a parcel of Xs (all or fome,
defined or undefined, but always the ſame), we have the fol-
lowing,
For AAA read XYZ) or zyx) | For A'A'A' read xyz) or ZYX)
-

O'AO' -
XYZ -

ΟΑ'ΟΙ -

Xyz
-

AO'O' - Zyx -
A'OO - zYX

For E'AE' read xYZ) or zyX) | For EA'E, read Xyz) or ZYx)
-IAI -
XYZ -
I'A'I' -

xyz
Ε'Ο'Ι -

ZyX -
EOI' -
zYx

For AEE read XYz) or Zyx) | For A'E'E' read xyZ) or zYX)
- Ο'ΕΙ' -
xYz -
ΟΕ'Ι -

XyZ
-

AI'I' -

zyx
-

A'LI -

ZYX

For E'EA' read xYz) or ZyX) | For EE'A read XyZ) or zYx)
- LEO -
XYz I'E'O' - xyZ
- ΕΙΟ . -
zyX -
ΕΠΙΟ' -

ZYx

Here, uſing P,Q,R, as general terms, PQR) denotes that all

Ps are Qs, and all Qs are Rs, whence all Ps are Rs : while
PQR only denotes that there is a parcel of Ps among the Qs,
and all Qs are among the Rs, whence that parcel of Ps is among
the Rs .
The rules for the connection of theſe ſyſtems are not compli-
cated, confidering the extent of the caſes they are to include.
Let the letters A,E, &c. be called proponents ; X,Y,Z, nominals:
and by the order of the nominals we always mean that X is firſt,
&c. both in XYZ, and ZYX. The nominals being direct
(X,Y,Z) and contrary (x,y,z), remember that,first,
 On the Syllogifm. 99

(first first and fecond

An affirmative Secondproponentdenotes that the second and third
third' third and firſt
nominals agree (are both direct or both contrary).
(first (first and ſecond
Anegative fecond proponent denotes that the second and third
third third and firſt
nominals differ (are one direct, one contrary).
Thus EIO muſt give Xyz or XYZ or zyX or ZYx
IEO must give XYz or xyZ or zYX or Zyx
Secondly, whether the middle term be Y or y depends only on
the accent of the middle proponent : a fub-accent gives Y, a
fuper-accent gives y. In the univerſal ſyllogiſm however, either
gives either.
Thirdly, the XYZ ſyllogiſms are the particulars which begin
with a particular : and the ZYX fyllogifms are the particulars
which begin with a univerſal.
For example, required OEI . Seeing the particular Oi, at the
beginning, take the order XYZ, ſeeing the ſuperaccent in E'
make it XyZ. Seeing the negative Or, let the exifting diſagree-
ment of the firſt and ſecond nominals continue : and the ſame of
the ſecond and third from the negative E.. Conſequently XyZ
is the fyllogifm expreſſed in nominals. Or the rationale of
the inference in OE'I is that a parcel of Xs are among the Zs
becauſe among the ys which are all among the Zs.
Again, required the nominal mode of expreſſing E'I'O.. See-
ing the univerſal E at the beginning, write down ZYX ; for the
ſuperaccent in I', write down ZyX ; for the negative in E',
continue yX ; for the affirmative in I', write zy : hence zyX is
the nominal form of E'I'Οι.
Required the proponent mode ofexpreſſing xYz. Here xY,
Yz, ſhow us that the premiſes are negative forms, and the direc-
tion of the order x, y, z, that the firſt premiſe is particular.
Then OE are the premiſes, and I the conclufion. And Y tells
us that the middle proponent has a ſubaccent. Whence OEI
is, ſo far as it goes, the proponent expreffion. And, by the laws
of form, the other accents must be as in O'E.I' , ſince the fyllo-
giſm follows the direct rule (page 93) .
 100
On the Syllogifm.
Required the proponent mode of expreſſing ZYx. Here we
note in ſucceſſion-univerſal commencement-firſt premiſe ne-
gative-ſecond, affirmative-middle accent fub. This gives ELLO
ofthe inverſe rule, or ELO .
Required the proponent notation for the univerſal xYZ) or
zyX). We fee at once EAE, or E'AE'.
The concomitants of a univerſal are found by changing the
firſt nominal into the contrary, in each of the forms, and throw-
ing away the fign of univerſality ()) . Thus the concomitants
of XyZ) or zYx) are xyZ and ZYx.
The weakened forms of a univerſal are found by merely
throwing away the ſymbol of univerſality ( )) from the two
forms of the univerſal. Thus the weakened forms of XYZ)
which is alſo zyx) are XYZ and zyx .
But we have not yet reached the climax of ſymbolic fimplicity
in the mere repreſentation of fyllogifms. An algebraiſt would
ſay that the ſtructure of the inference, as now conſidered, does
not depend upon the names ; but only upon their reference to
the names in the fundamental form XYZ) . He would there-
fore propoſe a ſimple ſymbol to repreſent letting alone, and
another to repreſent changing into the contrary. Theſe, with a
ſign of complete univerſality, and another of inverſion of order,
are all that he would find neceſſary. Let o and I ſignify letting
alone and changing into the contrary : let the terminal parentheſis
denote complete univerſality, as before, and let inverſion of order
be denoted by a negative ſign prefixed. Thus XYZ or LAI ,
would be denoted by ooo ; Zyx or A.O'O' by- 011 ; AEE
or XYz) by 001 ) or its equivalent- 011. Thus - 011 tells us
that ſome of the Zs are ys, all the ys are xs, whence ſome ofthe
Zs are xs. To write its proponent form, obſerve that - inſtructs
us to write a univerſal firſt ; II to make it affirmative ; I in the
middle to fuperaccent the middle propofition ; of to make the
ſecond premiſe negative. We have then A.O'O' or X) Y +
Z : Y = Z : X which is Zy + y)x =Zx, as aſſerted.
All that relates to univerſals in the preceding, applies to the
complex fyllogifms. Let a couple ofparentheſes imply a complex
fyllogifm : thus DaDaDa may be (XYZ) or (000). Then in (010)
or (XyZ), we are to fee that X is a ſubidentical ofy, and y of Z,
whence X is the ſame of Z. But Xy and yZ warn us to write
 101
On the Syllogifm.
contraries for the firſt and ſecond premiſes and y to fuperaccent
the middle letter : whence CC'D, is the fyllogifm expreſſed by
the names XYZ. The equivalent forms-(101 ) and (zYx) ex-
preſs itby ſaying that z is a ſubidentical of Y and Y of x, whence
z is a fubidentical of x.
I now look at the ſtrengthened particular fyllogifms. All in-
ference which is fundamental, that is, which will come from
nothing weaker than the premiſes given, has been reduced to the
one eaſy caſe of ' the contained of the contained is contained.'
The ſtrengthened particular, the type of which is A'AI , obeying
the inverſe rule of formation, and written at more length in Y)X
+ Y)Z = XZ, may be ſtated thus ' all names are common as to
what they contain in common.' If we denote this ſtrengthened
fyllogifm by XYZI , a ſymbol intended to imply ſomething be-
tween XYZ and XYZ) in the amounts ofquantity introduced,
we ſhall find that the eight ſtrengthened fyllogifms must be re-
preſented by
Α'ΑΙ ΧΥΖΙ AA'I' = xyzl
A'EO = XYz ! AE'O' = xyZI
E'A'O₁= Xyzl EAO' xYZI
Ε'Ε'Ι = XyZI EEI' = xYzl

The rules of connexion are preciſely thoſe for the particular

fyllogifms : and inverſion is abſolutely ineffective. Thus XYZI
=ZYXI .
A few words will ſerve to diſpoſe of the mixed complex fyllo-
giſms in which a complex premiſe is combined with a ſimple one,
univerſal or particular. Firſt, when a complex and a univerſal
are premiſed, and ſigns and accents are as in the direct rule (page
92), the concluſion is as it would be if the A were heightened
into D, or E into C. Thus ED' gives C1, the fame as CD'.
For E is Cor C1, and both CD' and CD' give Cı, but with
different quantities. But if the premiſes be conſtructed on the
inverſe rule, there is no more inference than can be obtained
when the complex premiſe is lowered into a univerſal : or we
have only a ſtrengthened particular. Thus in DE' or (A +
O')E', A.E' gives the ſtrengthened particular AE'O', and O'E'
is inconcluſive. And when the complex premiſe is combined
with a particular, we have only what would follow if the com-
plex premiſe were lowered into a univerſal. Thus Dil', or
102
On the Syllogifm.
(A + O' )I' can only give AI'I' ; and D'I' or (A' + O.)I' gives no
conclufion, for A'I' is inconcluſive.
The claſſification of opponent forms may be thus treated.
We know that opponent forms of AEE, for instance, be it A
E.E. or A'E'E', muſt be IEO and AII . Now whether A.EE
ſhall have IEO or ELO', whether A'II or II , depends upon
the introduction of a new and arbitrary notion of the order to be
adopted. Our firſt ſyllogiſm being deſcribed by XY, YZ, XZ,
the opponent which ends in the contradiction of the firſt premiſe
is in XZ, YZ, XY ; which, keeping Z middle, is either to be
deſcribed with reference to XZ, ZY, XY, or to YZ , ZX, YX .
Now in adopting the firſt of theſe three orders, there is nothing
which compels us therefore to prefer the ſecond to the third, or
vice verfâ.
The effect of the change of order which conſiſts in the inter-
change of Z and X is as follows. The premiſes change places ;
A and O with altered accents, altered alſo in the conclufion, E
and I with unaltered accents. Thus AI'I' becomes I'A'I' ;
E'O'I,becomes OEI.. Accordingly, it is matter of new ar-
rangement whether for instance, IEO or ELO' ſhall be called
the opponent of AEE.; and I prefer to give the name to both.
The conſequence is, the following diſtribution of opponents ;-
AA AO ОА
AE ΕΑ ΑΙ ΙΑ ΕΙ ΙΕ .
EE EO ОЕ

The three ſets repreſent letters combined in repreſentation of pre-

miſes : the first two containing fix fyllogifms each, the third
twelve. The third must be divided into two sets of fix each, in
one of which the ſubaccents are in greater number, in the other
the ſuperaccents. There are then four ſets in all. Pick any
two out of a ſet, which only differ in change of order : theſe two
have the fame opponent forms, namely, the other four of the
ſet. For instance, ALI and II , in which ſubaccents predo-
minate. Take ΑΕ, ΕΑ, ΕΙ, IE, and complete fyllogifms in
ſuch manner as to make ſubaccents predominate : giving A.EE ,
Ε.Α'Ε. , ΕΙΟ' IEO.. The laſt four are the opponents ofthe
firſt two.
In the ſet of ſtrengthened particulars the opponent forms will
be found to be univerſals weakened in the conclufion without
 On the Syllogifm. 103
being weakened in the premiſes. Thus AA'I' has A'E'O' for
one of its opponents : but A'E' may produce the univerſal con-
cluſion E' as well as its weaker form O'.
Some readers, particularly thoſe who have a tincture of algebra,
are more helped by ſymbolic notation than by language : with
others it is the converſe. To ſuit the latter, obſerve that the
language of page 78 may easily be adapted to ſimple ſyllogifms.
ThusA being fſubaffirmation, I may be ſome ſubaffirmation, O'
maybe ſome ſupernegation ; and ſo on. Thus instead of E'I'O, we
may ſay that ' fupernegation of ſome ſuperaffirmation gives ſome
ſubnegation.' Practice in this language would make the phraſe
ſuggeſt ſomething more than the notation it is derived from.
The phrafe refers to Z : there is a term partially ſuperaffirmed
of Z , namely Y ; and a complete ſubnegative of Y, namely X.
The partial ſubaffirmation declares ſome things neither Y nor Z ;
the complete ſupernegation declares that whatever is not Y is X.
Conſequently there are ſome Xs which are not Zs : or X is a
partial ſubnegative of Z. This ſubject will be reſumed.

In what precedes are two views of the deduction of all the

varieties of fyllogifm. The firſt, taking the complex fyllogifm as
the ſource, connects the ſtrengthened fyllogifms and the parti-
cular ones with the univerſals, and thus in fact reduces every
thing to the conſtituents of DDD or DDD . The ſecond pro-
ceeds from AAA , A'AI , AI'I', and II ,and forms the claſſes
of univerſal, ſtrengthened, and particular, fyllogifms by ſubſti-
tuting contraries in every way in which it can be done. Theſe
two ſyſtems have cloſe connexion, but not ſo cloſe as might
perhaps be thought : for IAI is not one of thoſe which are
connected with AAA in the formation of a complex fyllo-
gifm.
The two new views which I now proceed to give are alſo
cloſely connected, and different from the former ones, in which
we held it equally admifſſible to refer one of the concluding terms
to the middle, as in X)Y, or the middle to one of the concluding
terms, as in Y)X. But now I ask whether it be not poſſible fo
104 On the Syllogifm.
to conſtruct the ſyſtem, that we may firſt lay down the middle
term and its contrary, as conftituting the univerſe of the ſyllo-
giſm, and then complete the premiſes and their conclufion, by
properly laying down the concluding terms in their places. We
may fucceed, if, in the firſt inſtance, we confider none but con-
vertible propoſitions. And this we can do; for univerſal ex-
cluſion and particular incluſion comprehend all aſſertion. Thus
univerſal incluſion is only univerſal excluſion from the contrary,
and particular exclufion is only particular incluſion in the con-
trary.
Setting out then with the middle term and its contrary, and
restricting ourſelves to E and I, let E ſignify (univerſal) exclu-
ſion from the middle term, and e from its contrary ; let I ſignify
(particular) inclufion in the middle term, and i in its contrary.
Chooſing a pair of concluding terms, we reject II, Ii, and ii on
grounds already demonftrated, and very easily ſeen in this view,
and proceed to confider Ee, EE and ee, EI and ei, Ei and el.
Ee. From this a univerſal conclufion must follow. If one
term be completely excluded from the middle and the other from
its contrary, the terms are completely excluded each from the
other. The fundamental forms are,
Ε.Α'Ει, Χ.Υ + Ζ.y = X.Z ; A.EE , X.y + Z.Y = X.Z
and by use of XZ, Xz, xZ, xz, we thus bring out the eight uni-
verſal fyllogifms.
EE and ee. From theſe a particular inclufion must follow.
Excluſion of both terms from a third, gives partial incluſion of
their contraries in each other : for all that third term belongs
to the contraries of the other two . The fundamental forms
are,

EE.I' , X.Y + Z.Y = xz ; AA'I' X.y + Z.y=xz

from which, as before, the eight ſtrengthened ſyllogiſms are de-
duced.
El and ei. From theſe a particular incluſion must follow.
The exclufion of one term from a third, and the incluſion of
part of a ſecond term in that third, tell us that part of the par-
ticularized term is in the contrary of the univerſalized term.
The fundamental forms are,
 On the Syllogifm. 105
ELO', X.Y + ZY =Zx ; AO'O', X.y + Zy = Zx
IEO , XY + Z.Y = Xz ; OAO , Xy + Z.y= Xz
from which the fixteen particular fyllogifms are deduced.
Ei and el . From theſe no conclufion can be drawn. All that
is ſignified is that one concluding term is wholly excluded from
a third, and the ſecond partially excluded (or included in the
contrary).
It thus appears that a ſyllogiſm with one particular premiſe is
valid when the premiſes reduced to convertible forms, ſhow the
middle term in both or the contrary of it in both ; otherwise,
invalid. Alſo, that the concluſion in its convertible form, takes
directly from the particular premiſe and contrariwiſe from the
univerfal.
It alſo appears that a ſyllogiſm with both premiſes univerſal is
always valid ; with a univerſal conclufion when the premiſes
(made convertible) ſhow one the middle term and the other its
contrary ; with a particular conclufion when both ſhow the mid-
dle term or both its contrary. And the convertible form of the
concluſion takes directly from both in the firſt caſe, and contra-
riwiſe from both in the ſecond.
The other view which I here propoſe is really a different
mode of looking at that juſt given. By the time we have made
every name carry its contrary, as a matter of courſe, we become
prepared to take the following view of the nature of a propofi-
tion. A name by itſelf is a ſound or a ſymbol : its relation to
things (be they objects or ideas) is twofold. There may be in
rerum natura that to which the name applies, or there may not.
I do not here ſpeak of how many things there may be to which
a name applies : it is not eſſential to know whether they be more
or fewer, either abſolutely or relatively. The introduction of
contraries may be made the expulfion of quantity. With refer-
ence to application, then, let a name be calledpoſſible or impof-
fible according as the thing to which it applies can be found or
not.

A name may be compounded ofothers ; the compound name

being that of everything to which all the components apply.
Thus wild animal is the name of all things to which both the
names wild and animal apply. To call this compound name
106 On the Syllogifm.
impoſſible is to ſay that there is not ſuch a thing as a wild animal :
to call it poſſible is to ſay that there is ſuch a thing.
X and Y being two names, the compound name may be re-
preſented by XY when poſſible, and by XY) when impoſſible.
This does not alter the meaning of our ſymbol XY, as hitherto
uſed : as yet it has been ' there are Xs which are Ys ' and now
it is ' XY, the name of that which is both X and Y, is the
name of ſome thing or things ; ' and theſe two are the fame in
meaning, fo far as their uſe in inference is concerned. Nor need
XY), as juſt defined, be treated as a departure from, otherwiſe
than as an extenſion of, the uſe of X)Y. In X)Y, we aſſert
that X is ſomething, namely Y : in X) we aſſert that X is nothing
whatever. The proper notation, however, for indicating that
the name X has no application, is X)u, u being the contrary of
U, which laſt includes everything in the univerſe ſpoken of; fo
that u may denote nonexistence.
The propofition ' Every X is Y' aſſerts that Xy is the name
of nothing, or X)Y= Xy). Similarly ' No X is Y ' aſſerts that
XY is the name of nothing, or X.Y= XY). But ' Some Xs
are Ys ' and ' Some Xs are not Ys ' merely aſſert the poſſibility
of the names XY and Xy.
A fyllogifm, then, is the aſſertion that from the poſſibility or
impoſſibility of the names produced by compounding X or x,
Z or z, each with Y or y, may be inferred the poſſibility or im-
poſſibility of a name compounded of X or x with Z or z . The
rules of the laſt ſyſtem are now ſo easily changed into the lan-
guage of the preſent one, that it is hardly worth while to ſtate
more than one for example. Thus, if✗ compounded with Y,
and Z compounded with y, both give impoſſible names, then X
compounded with Z gives an impoſſible name. This is XY) +
Zy) =ZX) or X.Y + Z.y = Z.X, or EA'E
The view here taken of compound names will be extended
in the next chapter.
 On the Syllogifm. 107

CHAPTER VI .

On the Syllogifm.
HEN the premiſes of a fyllogifm are true, the conclufion
WHEN
is alſo true, and when the concluſion is falſe, one or

both of the premiſes are falſe. There are two kinds of modifica-
tions which it may be useful to confider: thoſe which concern
the entrance of the propoſition into the argument ; and thoſe
which affect the connexion of the ſubject and predicate.
As to the propofition itſelf, it may be true or falſe abſolutely,
or it may have any degree of truth, credibility, or probability.
This relation will be hereafter confidered ; and, according to the
principles of Chapter IX. ſo far as the propofition is probable
it is credible, and ſo far as it is credible, it is true. But as to
other modes of looking at the ſyllogiſm, are we entitled to ſay
that every thing which can be announced as to the premiſes may
be announced in the ſame ſenſe as to the conclufion ? The an-
ſwer is, that we cannot make ſuch announcement abſolutely ; but
of the premiſes as derived from that concluſion we can make it.
In what manner ſoever two premiſes are applicable, their conclu-
ſion as from thoſe premiſes is alſo applicable : becauſe the conclu-
ſion is in the premiſes. For inſtance, in the ſyllogiſm ' all men are
trees, all trees are rational, therefore all men are rational, the
premiſes are abſurd and falſe, and the conclufion taken indepen-
dently is rational and true : but that conclufion, as from thoſe
premiſes, is as abſurd as the premiſes themſelves. Again, in ' all
pirates are convicted, all convicts are puniſhed, therefore all
pirates are puniſhed, the premiſes are desirable, and ſo is the
conclufion with thoſe premiſes. But the conclufion is not de-
firable in itſelf: as that pirates ſhould be punished with or with-
out trial. Neither may we say ' ✗ ought to be Y and Y ought
to be Z, therefore X ought to be Z ' except in this manner, that
we affirm X ought to be Z in a particular way. We may not
even ſay that when ' X ought to be Y, and Y is Z ' it follows
that ' X ought to be Z,' for it may be that Y ought not to be Z.
Thus a royaliſt, in 1655, would say that the hundred excluded
108 On the Syllogifm.
members of Cromwell's parliament ought to be allowed to take
their ſeats, and alſo that all who took any feats in that parliament
were rebels ; but he would not infer that the hundred members
ought to be rebels. There is nothing which, being the property
of the premiſes, is neceſſarily the independent property of the
conclufion, except abſolute truth. It ſhould be noted that in
common language and writing, the uſual meaning of conclufions
is that they are ſtated as of their premiſes and to ſtand or fall
with them, even as to truth. Though a concluſion may be true
when its premiſes are falſe, the proponent does not mean, for the
moſt part, to claim more than his premiſes will give, nor that
any thing ſhould ſtand longer than the premiſes ſtand.
Next, we are not to argue from what we may ſay of a propo-
ſition to what we may ſay of the inſtances it contains, except as
to what concerns the truth of thoſe inſtances, or elſe to what
concerns the inſtances as parts of a whole. If I ſay ' Every X
is Y' I affert, no doubt, of each X independently of the reſt :
that is, the truth of ' Every X is Y' involves the truth of ' this
X is Y. But if, to take ſomething elſe, I maintain ' Every X
is Y' to be a desirable rule, I do not therefore aſſert ' this X is
Y' to be a defirable caſe, except upon an implied neceſſity that
there ſhould be a rule. And if I say that ' every X is Y ' is
unintelligible, I do not say that ' this X is Y' is unintelligible ;
and so on. Thus, where there muſt be a rule, as in law, ' every
man's houſe is his caſtle' is defirable, becauſe there is but one
alternative ' no man's houſe, &c.' But the propoſition, by itſelf,
may not be defirable as to the inſtance of a generally reputed
thief or receiver.
There is one caſe, however, in which a term cannot be ap-
plied to the general propoſition, unleſs it can be applied in a
higher degree to the inſtances. The propofition ' Every X is Y '
cannot be announced as of any degree ofprobability, unleſs each
inſtance has a much higher degree of probability. If μ , ν, ρ, &c.
be the probabilities of the ſeveral inſtances, ſuppoſed independent,
that ofthe propoſition (Chapter IX.) is μνρ ... which product muſt
be leſs than that of any one of the fractions of which it is formed.

I now come to the confideration of circumſtances which mo-

dify the internal ſtructure of the premiſes themſelves. And firſt
of conditions.
 On the Syllogifm. 109
A conditional propoſition is only a grammatical variation of
the ordinary one ; as in ' Ifit be X, then it is Y.' The common
form of this, ' Every X is Y,' is called categorical, or predicative.
Ofthe two forms, categorical and conditional, either may always
be reduced to the other ; as follows,
' Every X is Y' or ' If X, then it is Y'
' No X is Y' or ' If X, then it is not Y'
The particular propoſitions might be given conditionally in
various ways, but the transformation is not ſo common. Thus
' ſome Xs are Ys' might be ' if X, then it may be Y' or ' if X,
then Y muſt not therefore be denied of it,' &c.
Of the two common ſubject-matters of names, ideas and
propoſitions, it is most common to apply the categorical form to
the firſt, and the conditional form to the ſecond: in truth we
might call the conditional form a grammatical convenience for
the expreffion of dependence of propoſitions on one another,
and of names which require complicated forms of expreſſion.
Thus in pages 2 and 3, the conditional forms, containing if, are
more ſimple than the correſponding categorical forms.
A condition may be either neceſſary, orfufficient, or both. A
neceſſary condition is that without which the thing cannot be ; a
ſufficient condition is one with which the thing must be. In
pages 73, 74, I have fufficiently pointed out the completeness of
the connexion between the conditional and the categorical forms.
In any one caſe the ſufficient muſt contain all that is neceſſary,
and may contain more.
After what is ſaid in page 23, it is not neceſſary to dwell on
the reduction of a conditional* ſyllogiſm to a categorical one.
The premiſes contain the concluſion : whatever gives us the
premiſes, gives us the conclufion. But I think that the reduc-
tion of conditional to categorical forms, though juſt, and, for in-
ference, complete, is not the repreſentation of the whole of what
paſſes in our minds.
As an example of what I mean, look forward to the nume-
rical ſyſtem of Chapter VIII. Precedent to all propoſitions,
* Wallis, as far as I know, was the firſt who aſſerted that all fyllogifms
are, or can be made, categorical. He did this in the ſecond theſis attached
to his logic, headed Syllogifmi Hypothetici, aliique Compofiti, referendi funt
omnes ad Ariftotelicos Categoricorum Modos.
110
On the Syllogifm.
there are the numerical conditions which preſcribe the limits of
the univerſe under confideration. Say there are 250 inſtances in
that univerſe : this is the firſt condition . Of theſe 100 are Xs and
200 are Ys ; giving a ſecond and third condition. If we take a
propoſition, as 20XY, and ask whether it be ſpurious or not, we
have reference to the three conditions understood. But this is not
neceſſary : for it would be poſſible categorically to expreſs theſe
conditions by ' 20Xs out of 100 in a univerſe of 250 inſtances
containing 200Ys are to be found among thoſe 200 Ys ? It is
of courſe the rule of brevity not to drag about theſe conditions
with every propoſition which is employed, but rather to ſtate
them once for all. There is however ſomething more. The
conditions are a reſtriction upon the arguments intended to be
introduced, and a restriction throughout. The attachment of
them to each individual propoſition does not expreſs this : if they
be ſeen in twenty conſecutive propoſitions, there is no more than
a preſumption that they are to be ſeen in the twenty-firſt. It is
better that the limits allowed ſhould be marked out by one boun-
dary than that the ſeveral arguments ſhould each have a deſcrip-
tion of the boundary to itſelf.
Juſt as a univerſe of names is defined by ſpecifying one or
more names to conſtitute collectively thefummum genus, or uni-
verſe, ſo one of propoſitions may be defined by ſtating propoſi-
tions which are to be true, or which are not to be contradicted, as
the caſe may be. Theſe propoſitions may be conditions preced-
ing all, or fome only, of the premiſes which are uſed in argu-
ment ; or ſome may precede ſome, and others others. In
analyſing arguments, it would be found that many propoſitions
which enter as premiſes, enter each with a condition understood,
and well understood, to be granted. Whatever the conditions
may be, ſo long as the conſequent propoſitions act logically toge-
ther to produce the final reſult, then that ſame reſult depends at
laſt only on the conditions, and muſt be affirmed when the con-
ditions, and their connexion with their conſequents, are affirmed.
But then it must be understood that the reſult alſo ſtands upon
the conditions, and may fall with them. Let us now examine
the common fyllogifm, and ſee whether there be any preceding
conditions, on which the reſult depends.
On looking into any writer on logic, we ſhall ſee that existence
 On the Syllogifm. III

is claimed for the fignifications of all the names. Never, in the

ſtatement of a propofition, do we find any room left for the alter-
native, ſuppoſe there should be no such things. Existence as ob-
jects, or exiſtence as ideas, is tacitly claimed for the terms of
every fyllogifm . The existence of an idea we must grant when-
ever it is diſtinctly apprehended, and (therefore) not felf-contra-
dictory : we cannot for inſtance admit the notion of a lamp
which is both metal and not metal ; but, as an idea, we are at
liberty to figure to ourſelves ſuch a lamp as that with which
Aladdin made his fortune. An attempt at a felf-contradicting
idea is no idea ; we have not that apprehenfion of it in which an
idea conſiſts : but in no other way can we say that the attempt
to produce an idea fails. It may then be more convenient here
to dwell on objective definition of terms, as more easily con-
ceived with relation to exiſtence and non-exiſtence. Accordingly,
let us take the propoſitions X)Y and X.Y, of the character of
which the particulars muſt partake, as to the point before us.
By the meaning ofy, in relation to Y, it follows that every thing
is either Y or y : if we say that Ydoes not exiſt, then every thing
is y. If then X exiſt, and Y do not, the propofition X)Y, or
X.y is falſe, and X)y, or X.Y is true. If neither X nor Y
exiſt, I will not ſo far imitate ſome of the queſtions of the ſchools
as to attempt to ſettle what nonexisting things agree or diſagree.
IfY exiſt, but not X, then y)x is certainly true, but not thence
X)Y, for when x is, as here, the whole univerſe, the proof of
y)x=X)Y fails to preſent intelligible ideas, that is, fails to be a
proof. But Y)x or Y. X is true.
If all my readers were mathematicians, I might pursue theſe
extreme caſes, as having intereſt on account of their analogy
with the extreme caſes which the entrance of zero and of infinite
magnitude oblige him to confider. But as thoſe who are not
mathematicians would not be intereſted in the analogy, and thoſe
who are can pursue the ſubject for themselves, I will go on to
ſay that the preceding order is not the natural one. We cannot,
to uſeful purpoſe, laying down the truth of the propoſition,first,
then proceed to enquire how the non-existence of one or both
terms affects the propoſition. The existence of the terms muſt
be firſt ſettled, and then the truth or falſehood of the propoſition.
The affirmative propoſition requires the exiſtence of both terms :
112
On the Syllogifm.
the negative propoſition, of one ; being neceſſarily true if the
other term do not exiſt, and depending upon the matter, as
uſual, if it do exiſt .
Let us make the existence of the terms to be preceding con-
ditions of the propoſitions. The ſyllogiſm A.AA is then as
follows,
If X and Y both exiſt, Every X is Y
If Z alſo exift Every Y is Z
Therefore If X, Y, Z all exiſt Every X is Z.

As to the concluding terms, X and Z, they remain, as it were,

to tell their own ſtory. Whatever conditions accompany their
introduction unto the premiſes, theſe ſame conditions may be
conceived to accompany them in the conclufion. But the middle
term diſappears : and, not ſhowing itſelf in the concluſion, the
conditions which accompany it must be expreſsly preſerved.
The conclufion then is ' every X is Z, if Y exiſt ' which may be
thrown into theform of a dilemma, ' Either every X is Z, or Y
does not exiſt' .
But taking X and Z to exiſt, let us conſider the following ſyl-
logiſm, as it appears to be,
Every X is (Y, if Y exiſt)
Every ( Y, if Y exiſt) is Z
Therefore Every X is Z.
Ifthis be not a valid fyllogifm, what expreſſed law of the ordi-
nary treatiſes does it break ? The middle term, a curious one, is
ſtrictly middle : but there is no rule for excluding middle terms
of a certain degree of fingularity. That it does break, and very
obviously, an implied rule, I grant. And as to this work, the
rule laid down in Chapter III. is broken in its ſecond condition
(page 50). The two uſes of the word is do not amount to one
ſuch uſe as is made in the conclufion. That X is (conditionally)
Y which is (on the ſame condition) Z, gives that X is (on the
ſame condition) Z. Accordingly, the abſolute concluſion is only
true upon ſuch conditions as give the middle term abſolute ex-
iſtence.
But it must be particularly noted that it is enough if this ex
 On the Syllogifm. 113
iſtence be given to the middle term by the fulfilment of the
conditions which precede the entrance of one of the concluding
terms. The condition of the act of inference is, that the com-
pariſon muſt be really made, if the terms to be compared with
the middle term really exiſt, or, which is the fame, if the condi-
tions under which they are to enter be ſatisfied. The other terms
being ready, there muſt then be a real middle term : and there
will be, if the mere entrance of one of the concluding terms be
proof of the exiſtence of a middle term ; while, if the other terms
cannot be brought in, from nonexistence, there is no occaſion to
inquire about a middle term, for it is otherwise known that the
compariſon cannot be completed. I will take two concrete
inſtances, in the firſt of which one of the concluding terms, if
exifting, is held to furniſh a middle term as real as itſelf, and in
the ſecond of which no ſuch ſuppoſition occurs. Of course I have
nothing here to do with the truth of the premiſes.
Philip Francis, (if the author ofJunius), was an accuſer whoſe
filence was fimultaneous with a government appointment : an
accuſer &c. reflects diſgrace upon the government (if they knew
that their nominee was the accuſer) : therefore Francis (if &c. )
reflects diſgrace upon government (if &c.).
Homer (if there were ſuch a perſon) was a perfect poet (if
ever there were one) : a perfect poet (if &c. ) is faultleſs in
morals : therefore Homer (if &c.) was faultleſs in morals.
The firſt inference is good, even though we grant that our
only poſſible mode of knowing of the existence of an accuſer &c.
is by eſtabliſhing that Francis was Junius : it is even good against
one who ſhould affert that the accuſer &c . is a contradiction in
terms in every actual and imaginable caſe except that ofJunius.
In the ſecond cafe, we put it that the man Homer (if he ever
exiſted ; fome critics having contended for the contrary) was a
perfect poet, if ever there were one. There may never have
been one ; and then Homer (exiſtent or nonexiſtent) was not a
perfect poet. There is no condition here, which being fulfilled,
is held to amount to an aſſertion that the middle term must have
exiſted : but the condition of the existence of the middle term is
independent. Accordingly, the ſecond inference is not good : it
ſhould be Homer (if &c.) was a perfect poet, if ever there were
one : that is, or elſe there never was a perfect poet.
I
114 On the Syllogifm.
Theſe points refer to the matter of a ſyllogiſm, and not to the
form ; or rather, perhaps, hold a kind of intermediate relation .
There is another proceſs which is often neceſſary, in the
formation of the premiſes of a ſyllogiſm, involving a transforma-
tion which is neither done by fyllogifm, nor immediately reducible
to it. It is the ſubſtitution, in a compound phrase, of the name
ofthe genus for that of the ſpecies, when the uſe of the name
is particular. For example, ' man is animal, therefore the head
of a man is the head of an animal ' is inference, but not fyllo-
giſm. And it is not mere ſubſtitution of identity, as would be ' the
head of a man is the head of a rational animal' but a ſubſtitution
of a larger term in a particular ſenſe.
Perhaps ſome readers may think they can reduce the above to
a fyllogifm. If man and head were connected in a manner which
could be made ſubject and predicate, ſomething ofthe fort might
be done, but in appearance only. For example, ' Every man is
an animal, therefore he who kills a man kills an animal.' It
may be ſaid that this is equivalent to a ſtatement that in ' Every
man is an animal ; fome one kills a man ; therefore ſome one
kills an animal,' the firſt premiſe, and the ſecond premiſe condi-
tionally, involve the conclufion as conditionally. This I admit :
but the laſt is not a fyllogifm : and involves the very difficulty in
queſtion. Every man is an animal ; ſome one is the killer of
a man ' : here is no middle term . To bring the firſt premiſe
into ' Every killer of a man is the killer of an animal' is just the
thing wanted. By the principles of chapter III, undoubtedly
the copula is might in certain inferences be combined with the
copula kills, or with any verb. But ſo ſimple a caſe as the pre-
ceding is not the whole difficulty. If any one ſhould think he
can fyllogize as to the inſtances I have yet given, let him try the
following. Certain men, upon the report of certain other men
،

to a third ſet of men, put a fourth ſet of men at variance with a

fifth ſet of men.' Now every man is an animal : and therefore
Certain animals, upon the report of certain other animals, &c.'
Let the firſt deſcription be turned into the ſecond, by any num-
ber of fyllogifms, and by help of ' Every man is an animal.'
The truth is, that in the formation of premiſes, as well as in
their uſe, there is a poftulate which is conſtantly applied, and there-
fore of courſe conſtantly demanded. And it ſhould be demanded
openly. It contains the dictum de omni et nullo (fee the next chap
 On the Syllogifm. 115
ter), and it is as follows. For every term uſed univerfally less
may be ſubſtituted, and for every term uſed particularly, more.
The ſpecies maytake the place ofthe genus, when all the genus
is ſpoken of: the genus may take the place of the ſpecies when
ſome of the ſpecies is mentioned, or the genus, uſed particularly,
may take the place ofthe ſpecies uſed univerſally. Not only in
ſyllogiſms, but in all the ramifications of the deſcription of a com-
plex term. Thus for ' men who are not Europeans ' may be
ſubſtituted ' animals who are not Engliſh.' If this poftulate be
applied to the unſtrengthened forms of the Ariftotelian Syllogiſm,
(page 17) it will be ſeen that all which contain A are immediate
applications of it, and all the others easily derived.
I now paſs to the confideration of the invention of names, and
of the distinctions which are made to exiſt for the want of it.
Any one may invent a name, that is, may choose a found or
ſymbol which is to apply to any claſs of ideas or of objects. The
claſs ſhould, no doubt, be well defined : but ſmall caution is here
neceſſary, for invented words are generally much more definite
thanthoſe which have undergone public uſage. They come from
the coiner's hand as ſharp at the edge as a new halfpenny : and in
proceſs of time we look in vain for any edge at all. The right
of invention being unlimited, and the actual ſtock having been
got together without any uniform rule of formation, there can
be no reason why we should admit any distinction which can be ab-
rogated by the invention of a name, ſo far as inference is con-
cerned. I do not dispute that the modes of ſupplying the want
ofnames may be of importance in many points of view : what
I deny is, that they create any peculiar modes of inference.
The invention of names muſt either be by actually pointing
out objects named, or by deſcription in terms of other names.
With the former mode of invention, as ' let this, that, &c (ſhow-
ing them) be called X' we can have nothing to do. As to the
latter, we may make a ſymbolic deſcription of the proceſs by join-
ing together the names to be uſed, with a ſymbol indicative of
the mode of using them, in extenſion of the ſyſtem in page 106.
Thus, P, Q, R, being certain names, if we wish to give a name
to everything which is all three, we may join them thus, PQR :
if we wish to give a name to every thing which is either of the
three (one or more ofthem) we may write P,Q,R : if we want
to fignify any thing that is either both P and Q, or R, we have
116
On the Syllogifm.
PQ,R. The contrary of PQR is p,q,r; that of P,Q,R is pqr ;
that of PQ, R is (p,q)r : in contraries, conjunction and disjunc-
tion change places. This notation would enable us to expreſs
any complication of the preceding conditions : thus, to name that
which is one and one only of the three, we have Pqr, Qrp,
Rpq ; for that which is two and two only, PQr, QRp, RPq .
Thus, XY includes the inſtances common to X and Y ; but
X,Y includes all X and all Y : accordingly X,Y is a wider term
than XY, except when X and Y are identical. As in page 106,
XY, the term, ſuppoſed to exiſt, is XY, the propofition of chapter
IV ; if we wish to distinguiſh, we may make X-Y the term, and
XY the propoſition, the hyphen having its common grammatical
uſe. Thus, X-Y P-Q tells us the fame as XYP-Q, both mean-
ing, for inference, no more than that there exiſt objects or ideas
to which the four names are applicable. But the firſt tells it
thus, ſome XYs are PQs ; and the ſecond thus, ſome things are
Xs, Ys, and PQs .
With reſpect to this and other caſes of notation, repulſive
as they may appear, the reader who refuſes them is in one of two
circumſtances. Either he wants to give his aſſent or difſſent to
what is ſaid of the form by means of the matter, which is eaſing
the difficulty by avoiding it, and ſtepping out of logic : or elſe he
defires to have it in a ſhape in which he may get that moſt futile
of all acquifitions, called a general idea,* which is truly, to uſe
the contrary adjective term as colloquially, nothing particular, a
whole without parts.
If the difficulty of abſtract aſſertion be to be got over, the
eaſieſt way is by firſt conquering that of abſtract expreſſion, to
the extent of becoming able to make a little uſe of it.
Suppoſe we ask for the alternative of the following ſuppoſition,
Both X, and either P, or Q and one ofthe two R or S.' This
is no impoſſible complication : for instance, ' He was rich, and
if not abſolutely mad, was weakneſs itſelf ſubjected either to bad
advice or to moſt unfavourable circumſtances.' The repreſen-
tation ofthe complex term is X { P, Q(R,S)} ; of the contrary,
* " Je vous avoue, dit ...., que j'ai cru en deviner quelque choſe, et que
je n'ai pas entendu le reſte. L'abbé de ....
.... a ce diſcours, fit réflexion que
c'était ainſi que lui-même avait toujours lu, et que la plupart des hommes
ne liſaient guere autrement."
 On the Syllogifm. 117
x, p(q,rs) or x,pq,prs. If not the above, he was either not rich,
or both not mad and not very weak, or neither mad nor badly
adviſed, nor unfavourably circumstanced.
When a name thus formed, whether conjunctively or disjunc-
tively, enters a ſimple inference, it gives riſe to what have been
called the copulative fyllogifm, the disjunctive fyllogifm, and the
dilemma. The two laſt are not well diftinguiſhed by their defi-
nitions as given : the disjunctive ſyllogiſm ſeems to be that in
which names are confidered disjunctively, the dilemma that in
which propoſitions are ſo uſed. But a propoſition entering as part
of a propoſition, enters merely as a name, the predicates being
uſually only true or falſe, or ſome equivalent terms. Apropofi-
tion may only enter for its matter, or it may enter in ſuch a way
that its truth is the matter : in this laſt caſe it is only as a name
that it is the ſubject of inference. Thus, ' It is true that he was
fired at' is ' the aſſertion (that he was fired at) is a true aſſertion.'
I believe the beſt way would be to apply the term disjunctive
argument ſo as to include the dilemma, marking by the latter
word (as a term rather of rhetoric than of logic) every argument
in which the disjunctive propoſition is meant to be a difficulty
for the opponent on every cafe, or horn, of it.
Whatever has right to the name P, and alſo to the name Q,
has right to the compound name PQ. This is an abſolute
identity, for by the name PQ we ſignify nothing but what has
right to both names. According X)P + X)Q=X)PQis not a
ſyllogiſm, nor even an inference, but only the aſſertion of our
right to uſe at our pleaſure either one of two ways ofſaying the
ſame thing inſtead of the other. But can we not effect the re-
duction fyllogiftically ? Let Y be identical with PQ ; we have
then PQ)Y and Y)PQ, and alſo Y)P and Y)Q . Add to theſe
X)P and X)Q, and we have all the propofitions aſſerted. But
we cannot deduce from them alone X)Y, the reſult wanted, by
any fyllogiftic combination of the fix. Nor muſt it be thought
ſurpriſing that we cannot, by a train of argument, arrive at de-
monſtration of it being allowable to give to anything which has
right to two names, a third name invented expreſsly to fignify
that which has fuch right. We might as well attempt to fyllo-
gize into the reſult, that a person who fells the meat he has killed
is a butcher.
118 On the Syllogifm.
I lay ſtreſs upon this, to an extent which may for a moment
appear like diligently grinding nothing in a mill which might be
better employed, for two reasons. First, the young mathema-
tician is very apt to try, in algebra, to make one principle deduce
another by mere force of ſymbols : and the above attempt may
ſhow him what he is liable to: Secondly, I am inclined to fup-
poſe that the diſtinction drawn between the claſſes of fyllogifms
to which I preſently come, and the ordinary categorical ones, is
due to what must be deſcribed in my language as a want of per-
ception of the abſolute, less than inferential (fo to ſpeak) identity
of X)P + X)Q and X)PQ. But all other propoſitions of the
kind, however ſimple, may be made deductions. For instance,
' if X be both P and Q, and if P be R, and Qbe S, then X is
both Qand S ' is thus deduced : X)P + P) R = X) R, and X)Q
+ Q) S =X)S, and X)R + X)S is X)RS. Even P)R + Q)R=
P,Q)R is deducible ; being P)R + Q)R = r)p + r)q = r)pq=
P,Q)R. Thus it is ſeen that, as foon as the conjunctive poſtu-
late is laid down, the identity of the corresponding disjunctive
poftulate with it may be ſhown. Next, if X muſt be either P
or Q, or X) P, Q, and if P be always R, and Q be always S, then
X)R, S may be deduced from the preceding.
Firſt, that X) P and Y)Qgive XY)PQcan be deduced ; evi-
dent as it may be, it is a ſucceſſion of applications. XY)X +
X)P gives XY)P, and XY)Y + Y)Qgives XY)Q, and XY)P
+ XY)Qis XY)PQby the postulate. Next, X)P,Q is pq)x,
and P)R is r)p, and Q)S is s)q, whence, as juſt proved rs)pq.
Now, rs)pq + pq)x=rs)x, which is X)R,S. It will be a good
exerciſe for the reader to tranflate this proof into ordinary lan-
guage.
I may now proceed to extend this idea and notation relative
to propoſitions of complex terms. The complexity conſiſts in
the terms being conjunctively or disjunctively formed from other
terms, as in PQ, that to which both the names P and Qbelong
conjunctively ; and as in P,Qthat to which one (or both) of
the names P and Qbelong disjunctively. The contrary of PQ
is p,q; that of P,Q is pq. Not both is either not one or not the
other, or not either. Not either P nor 2(which we might denote
by : P,Q or .P,Q) is logically ' not P and not Q'or pq : and
this is then the contrary of P,Q.
 On the Syllogifm. 119
The disjunctive name is of two very different characters, ac-
cording as it appears in the univerſal or particular form : ſo very
different that it has really different names in the two cafes,
copulative and disjunctive. This diſtinction I here throw away :
oppoſing disjunctive, (having one or more of the names) to
conjunctive, (having all the names). The disjunctive particle or
has the fame meaning with the diſtributive copulative and, when
uſed in a univerſal. Thus, ' Every thing which is Por Qis
R or S ' means ' Every P and every Q is Ror S. But PQ is
always both P and Qin one.' Accordingly
Conjunctive | PQR uſes and collectively.
Disjunctive P,Q,R in a univerſal uſes and diſtributively,
P,Q,R in a particular uſes or disjunctively, in
the common ſenſe of that word.

' Either Por Q is true,' is an ambiguous phrafe, which is

P,Q)T or T) P,Q according to the context.
The manner in which the component of a name enters, whe-
ther conjunctively or disjunctively, is to paſs as it were for a part
ofthe quality of the name itſelf. Thus the contrary of P (con-
junctive, as indicated by the abſence of the comma) is ,p (dis-
junctive, as indicated by the comma). To teſt this aſſertion about
the mode of making contraries, let us aſk what is that of ' one
only of the two Por Q?" We know it of courſe to be ' both or
neither. The name propoſed is Pq, Op and its contrary is
(p,Q)(q,P), that is, one of the two p,Q, and one of the two q,P.
It is then either pq, pP, qQ, or PQ: the ſecond and third can-
not exiſt, therefore it is pq, PQ, as already ſeen. I need hardly
have remarked that (P, Q)( R,S) is PR, PS, OR, QS.
Obſerve that though X)PQ gives X)P, and that XPQ gives
XP, we may not say that XY)P gives X)P, nor that X)P,Q
gives X)P. But any disjunctive element may be rejected from
a univerſal term, and any conjunctive element from a par-
ticular one. Thus P)QR gives P)Q and P,Q)R gives P)R.
Alſo P.Q,R gives P.Qand PQ:R,S gives P:R. All theſe rules
are really one, namely that PQ is of the fame extent at least as
PQR. This will appear from our rules of tranſpoſition preſently
given.
120
On the Syllogifm.
Letchange from one member of the propofition to the other
be called tranſpoſition. I proceed to inquire how many tranfpo-
ſitions the various forms will bear, and what they are. Itwill
however be neceſſaryto complete our forms by the recognition, as
apropofition, ofthe ſimple afſertion ofexistence or non-existence.
By XU we mean that there are in the universe things to which
the name X applies, and we speak only offuch things under the
name. Accordingly X)U and XU do not differ in meaning.
Byu, the contrary of U, we can only denote non-existence ;
thus X.U or X)u throws the name X out of confideration.
Thus Y)X= U) X ,y ; Y.X = YX)u, &c. To fignify, for in-
ſtance, that X and Y are complements (contraries or fubcon-
traries, page 75) we have U)X,Y, which our rules will tranf-
poſe into xy)u, or x.y.
Having to confider ſubject and predicate, conjunctive and dif-
junctive, affirmative and negative, univerſal and particular, we
must think of fixteen different forms. Thus the four forms of
the univerſal affirmative are

XY) PQ ; X, Y)PQ; XY)P,Q ; X,Y)P,Q

It will be beſt here to neglect the contranominal converſes of
A and O equally with the ſimple converſes of E and I : thus
XY)PQ may be read as identical with p,q)x,y. There is alſo
one obvious tranſpoſition which we must not merely neglect but
throw out ; ſince it does not give a reſult identical with its prede-
ceſſor. I mean the tranſpoſition of M)PQ into MP)Q: the
fecond follows from the first but not the firſt from the ſecond.
Alſo the correſponding change of M.P,Q into Mp.Q, for the
ſame reaſon ,
This being premiſed, the following are the rules ;-
Direct tranſpoſition is the change from one member to the
other without alteration of name or junction : contrary, with
alteration of both .
The convertibles (E,I) allow direct tranſpoſition ofconjunctive
elements either way, from ſubject to predicate, or from predicate
to ſubject : and theſe are the only direct tranſpoſitions. Thus
X.YZ = XY , Z, and X-YZ = XY-Z .
The inconvertibles ( A,O) allow contrary tranſpoſition of con-
junctive elements from ſubject to predicate, and of disjunctive
 121
On the Syllogifm.
elements from predicate to ſubject : beſt remembered by allow-
ing SP to ſtand for conjunctive and PS for disjunctive. And theſe
are the only contrary tranſpoſitions. Thus XY) M =X)M,y
and M)X,Y = My)X .
An element that can be rejected cannot be tranſpoſed, and
vice versa. Thus X,Y)M gives X)M, and Y cannot be tranf-
pofed.
The following table exhibits the varieties of the forms A and
E, equivalents being written under one another, and converfions,
contranominal or ſimple, oppoſite.
XY) P,Q pq)x,y XY.PQ PQ.XY
Xp )Q,y Yq)P,x XP.QY QY.XP
Xq) P,y Yp)Q,x XQ.PY PY.XQ
X)P, Q,y pqY)x X.PQY PQY.X
Y)P, Q,x pqX)y Y.PQX PQX.Y
p)Q,x,y XYq)P P.QXY QXY.P
q)P,x,y XYp)Q Q.PXY PXY.Q
XYpq)u U)P, Q,x,y XYPQ.U U.XYPQ

XY) PQ p,q)x,y XY.P,Q P, Q.XY

X) PQ,y [p,q]Y)x X. [ P,Q] Y [ P, Q] Y X
Y) PQ,x [p,q]X)y Y.[P,Q]X [P,Q]X.Y
XY[p,q] )u U)[x,y],PQ XY[ PQ].U U.XY[ P, Q]

X,Y)P,Q pq)xy X, Y.PQ PQ.X, Y

[X,Y]p)Q q)xy,P [X,Y]P.QQ.[ X, Y] P
[X,Y]q)P p)xy,Q [X,Y]Q.P P.[X,Y]Q
[X,Y]pq)u U)xy,P,Q | [XY] PQ.UU.[X,Y]PQ

X, Y) PQ p,q)xy X,Y.P, Q P,Q.X,Y

[x,y][p,q])u U)xy,PQ [X,Y][P,Q].UU.[X,Y][P,Q]
If for ) we write (:) in the left hand diviſions, and eraſe the (. )
and uſe the hyphens ofpage 115, on the right, we have the tranf-
poſitions of O and I. And if we write p and q for P and Qon
the left, and change the form X)Y into X.y, we thereby change
the forms of A into thoſe of E. If more than two elements
were used, the tranſpoſitions would now be perfectly eaſy.
It appears that there are no leſs than fixteen A forms into
 122
On the Syllogifm.
which XY) P,Qmay be varied : the reaſon is that both fubject
and predicate are tranſpoſibly conſtructed. But XY)PQ ſhows
only a tranſpoſible ſubject ; X,Y)P,Q only a tranſpoſible predi-
cate : and theſe have only four forms each. Lastly, X,Y)PQ,
having neither tranſpoſible, has only two forms. By tranſpoſi-
bly constructed, I mean capable of having the elements ſeparated
by tranſpoſition. The whole term is always tranſpoſible : that is,
the complete ſubject, or the complete predicate, may be looked
on as conjunctive or disjunctive, at pleaſure. Thus in X)Y, if
we conſider this as XU)Y,u, we may make this yU)x,u or y)x.
So that the ordinary contranominal converfion may be confidered
as a caſe of the more general rule. Juſt as, in arithmetic, a num-
ber, 5, may be made to obey the laws of a + b as 0 + 5 , or of ab
as I x 5.
Syllogifms of complex terms might be widely varied, even if we
choſe to confider only each firſt caſe of the preceding table as
fundamental. Thus

XY) P,Q+ VW)P,Q=(x,y)-(v,w) AA'I'

would give fixty-four varieties of premiſes. I now proceed to

ſhow that the ordinary disjunctive and dilemmatic forms are
really common ſyllogiſms with complex terms, reducible to ordi-
nary fyllogifms by invention of names .
Example 1. Every S is either P, Q, R ; no Pis S ; no Qis
S ; therefore every S is R. Let S repreſent ' the true propofi-
tion' (fingular), and let P, Q, R be names of propoſitions, and
this then repreſents a very common form, which would be ex-
preſſed thus ' either A is B, or C is D, or E is F ; but A is not
B, C is not D ; therefore E is F.' I ſay that, where the necef-
ſary names exiſt, the final ſtep ofthis could not be diftinguiſhed
from a common fyllogifm ; which accordingly it becomes by in-
vention of names .
We have S)P,Q,R, whence Spq)R. But S.P and S.Qor S)p
and S)q give S)pq, with which S)S combined gives S)pqS. And
S)pqS + pqS)R= S) R. Let M be the name ofwhat is S and not P
and not Q, and the thing required is done. Here then is a fyllogifm
of the ordinary kind, to one premiſe of which we are led by a
uſe of the conjunctive postulate (page 116) : the neceffity for which
is the distinction between the claſs we are confidering and others.
It happens here that two of the terms of our final fyllogifm are
 On the Syllogifm. 123
identical : for Spq is of no greater extent than S. But the uſe
made of S)S is perfectly legitimate.
Example 2. ' If A be B, E is F ; and if C be D, E is F ; but
either A is B or C is D ; therefore E is F.' This can be re-
duced to
P)R + Q)R + S)P,Q=S)R
which is immediately made a common fyllogifin by changing
P)R + Q) R into P,Q) R.
Example 3. From P follows Q; and from R follows S ; but
Qand S cannot both be true ; therefore P and R cannot both
be true.' This may be reduced to
P)Q+ R)S + T.QS = T.PR
or PR)QS + T.QS =T.PR
Example 4. ' Every X is either P, Q, or R ; but every P is
M, every Q is M, every R is M ; therefore every X is M.'
This is a common form of the dilemma ; it is obviously reduci-
ble to P, Q,R)M + X)P, Q,R = X) M .
Example 5. ' Every X is either P or Q, and every Q is X. '
This is wholly inconcluſive, and leads to an identical reſult, as
follows ; X)P, Q gives Xp) , which with Q)X gives Xp)X,
a neceſſary propofition.
Example 6. If we throw X)R into the form X)R,R, we have
Xr)R, or ' Every X which is not R is R,' a contradiction in
terms. But it evidently implies that there can be no Xs which
are not Rs ; and thus alſo we return to X)R. Take ' every X
is either P, Q, or R ; every P is M ; every Q is M ; and every
Mis R.' Here X)P,Q,R = Xr)P,Q, which with P,Q)M gives
Xr)M, which with M)R gives Xr)R or X) R.
Example 7. ' Every X is either P or Q, and only one. ' This
gives two propoſitions, X)P,Q + X.PQ. Now X)XP,XQ is
identical with X)P,Q, and this may be looked on as an extreme
cafe of
X) P,Q + X )Y = X)PY, QY
but X.PQ gives XP)q and XQ)p, from which we can obtain
X)XP,XQ + XP )q + XQ)p = X)p,q
Hence X)P,Q+ X)p,q = X )[P,Q,][p,q.]
=X)Pp, Pq,Qp,Qq = X) Pq,Qp
fince Pp and Qq are ſubject to X.Pp and X.Qq. All this being
124 On the Syllogifm.
worked out in ſyllogiſtic detail, ſhows us that the tranſition from
Every X is Por Q, and no X is both' to ' Every X is either
P and not Q, or Q and not P' is capable of being made ſyllo-
giſtically. The ſtudent of logic may thus acquire the idea,
which ſo ſoon becomes familiar to the ſtudent of mathematics,
of perfectly ſelf-evident propoſitions which are deducible from
one another, as diftinguiſhed from thoſe which are not.
Example 8. ' Every X is one only of the two, Por Q ;
every Y is both P and Q, except when P is M, and then it is
neither ; therefore no X is Y.' Here is a caſe in which it is the
fact of the exception and not its nature which determines the
inference : M may be anything. This ought to appear in our
reduction : and it does appear in this way. From X)P,Q it is
obvious that X)P,Q,R,S, and fyllogiſtically demonftrable from
X)P,Q, and Xrs)X. Now in the ſecond premiſe we have
Y)PQm,pqM, or [p,q,M][P,Q,m])y
or pQ,Pq,PM,QM,pm,qm)y
from which, by rejection, follows pQ,Pq)y. And the firſt pre-
miſe is X) Pq,Qp. Whence X)y or X.Y.
It is not neceſſary to multiply examples : I will conclude this
part of the ſubject by pointing out that the ordinary propoſitions
X)Y, & c. are, with reference to their inſtances, disjunctively
compoſed : the difference between the univerſal and particular
lying in the latter being indefinite in the number of its inſtances.
Thus, if there be three Xs and four Ys, the four propoſitions
are, applying the name to each inſtance, as ſeen written at length in
X ,X,X) Y, Y, Y, Y ; X,X,X.Y, Y,Y, Y ; (X,X,X)( Y,Y, Y,Y) ;
and (X,X,X): Y,Y,Y,Y.
The propofition in page 25, is a caſe of the preceding method.
I leave the reader to ſhow it, and alſo that the hypotheſis is
flightly overſtated.
I now come to the forites, the heap or chain of fyllogifms, in
which the concluſion of the firſt is a premiſe of the ſecond, and
fo on. Take a ſet of terms, P, Q, R, S, &c. and let the order
of reference be PQ, QR, RS, &c. Then AAAA &c. is a
forites, and the only one uſually confidered : thus,
P)Q + Q) R + R)S + S)T = P)T
 On the Syllogifm. 125
The first two links give P)R, which with the third gives P)S,
which with the fourth gives P)T. Thus we have links, inter-
mediate concluſions, and afinal conclufion.
A great number of different ſorites may be formed, under the
following conditions,
The firſt particular propoſition which occurs, be it link or
conclufion, prevents any future link from being particular : for
all the concluſions thence become particular.
Examine the caſes of ſyllogiſm which proceed by the firſt
rule of accentuation (page 92), that is, which have beginning and
ending both univerſal, or both particular : theſe only can occur
in a forites, except at the end, or in the place where a particular
propofition firſt enters. It will be found that the conclufion,
when the argument goes on, must come after ſomething con-
nected with that which comes after it by the firſt rule of ac-
centuation : except at the place where a particular concluſion
comes in for the first time. For instance, E.E' gives A1, which,
ſtill keeping conclufions univerſal, must be followed by A, or E.,
which follow E' by the firſt rule. Again, take OE', which gives
I.; this must be followed either by A. or E1, which follow E' by
the ſame rule : and ſo on. Accordingly,
Any chain of univerſals, in which affirmation is followed by a
like prepofition, and negation by a different one, as A.AEA'
E'A.EE', &c. may be part ofthe chain of a forites. And the
chain muſt be either of this kind wholly, or once only broken in
one of two ways : either by the direct entrance of a particular
propoſition, or by a breach of the rule. In a chain ofthis kind,
unbroken, the conclufions are affirmative or negative, according
as an even or odd number of negatives goes to the formation of
them . All the concluſions have the ſame accent as the firſt link.
Let a particular premiſe be introduced, as in A.EE'I' &c.
The accent of the particular introduced muſt be the fame as
or contrary to that of the firſt link, according as the preceding
number of negatives is odd or even. For the accent of the firſt
link remains as long as the concluſion is univerſal, and a fyllo-
giſm with the ſecond premiſe particular follows the ſecond
rule. Thus, inſerting the intermediate concluſions; the above is
Α.Ε.Ε.)Ε (A.)Ι'(I'). And after (I') must come A' or E', ſo
that the first rule ſtill continues. But the accent of the conclu-
ſions changes.
126
On the Syllogifm.
Now let the rule of accentuation be broken. The accent of
the concluſion ſtill requires the firſt rule to be reſumed. Thus,
EE'(rule unbroken) gives A1, and E.E. (rule broken) gives I' ,
andA requires A or E, to follow E', while I' requires A' or
E' to follow E.. This one breach of rule only changes the con-
clufion from univerſal to particular. The accent of the conclu-
ſion changes as before.
The links of a forites, then, are either a chain of univerſals
following the firſt rule of accentuation, or ſuch a chain with one
breach of the rule, or ſuch a chain with one particular inſerted,
of the ſame or contrary accent to the firſt link, according as the
preceding negatives are odd or even, and made the commence-
ment of the reſumption ofthe rule (if broken). In all the cafes
the concluſion is affirmative or negative according as the preced-
ing negatives are even or odd in number : the unbroken chain
has a univerſal conclufion with the accent of the firſt link, and
the broken one a particular with the contrary accent.
Α'Ε'Ε Α'Ε' ΕΑ'Α.Ε.Ε'Α, ΑΕΑ'Ο Α'Ε'
E'A'A'E' ΕΟ'Ι'Ο ' ' ΕΕΙ'Ι'Ο '

Here are examples of the three kinds. The chain is in the

firſt row, the intermediate and final concluſions in the ſecond.
Thus the ſecond example preſents the ſyllogiſms E.Α'Ε., Ε.Α.Ο',
Ο'Ε,Ι', Ι'Ε'Ο', O'A ' ; and at length is
P.Q + R)Q + R)S + S.T + t.u + U(V = V :P
The ſorites uſually confidered are only AAA .... and
A'A'A' ..... To theſe might be added without abandoning
the Ariftotelian fyllogifm, ſuch as AEA'A'A' ...., AEAAA
....
But it would not be very easyto follow the chain in thought
without introducing the intermediate concluſions, and thus de-
ſtroying the ſpecific character of the proceſs.
And juſt as the ordinary univerſal ſyllogiſm can be reduced to
AAA , ſo the univerſal forites can always be reduced to a chain
of A. Thus A'E'EA'E' or

Q)P + q.r + R.S + T) S + t.u = p.u

is u)T + T)S + S)r + r) Q + Q)P = u)P
 127

CHAPTER VII .

On the Aristotelian Syllogifm.

ROM the time of Aristotle until now, the formal inference
F has been a matter of ſtudy. In the writings of the great
philoſopher, and in a ſomewhat ſcattered manner, are found the
materials out of which was conſtructed the ſyſtem of fyllogifm
now and always prevalent : and two distinct principles of exclu-
ſion appear to be acted on. Perhaps it would be more correct
to say that the followers collected two distinct principles of ex-
cluſion from the writings of the maſter, by help of the afſumption
that everything not uſed by the teacher was forbidden to the
learner. I cannot find that Aristotle either limits his reader in
this manner, or that he anywhere implies that he has exhausted
all poſſible modes of ſyllogizing. But whether theſe exclufions
are to be attributed to the followers alone, or whether those who
have more knowledge of his writings than myſelf can fix them
upon the leader, this much is certain, that they were adopted,
and have in all time dictated the limits of the fyllogifm. Of all
men, Ariftotle is the one of whom his followers have worſhipped
his defects as well as his excellencies : which is what he himself
never did to any man living or dead ; indeed, he has been accuſed
of the contrary fault.
The firſt of theſe exclufions is connected with the celebrated
dictum de omni et nullo, namely, that what is diſtributively affirmed
or denied of all, is diſtributively affirmed or denied of every ſome
which that all contains. It is there ſaid that in every fyllogifm
the middle term must be univerſal in one of the premiſes, in order
that we may be ſure that the affirmation or denial in the other
premiſe may be made of ſome or all of the things about which
affirmation or denial has been made in the firſt. This law, as
we ſhall fee, is only a particular caſe of the truth : it is enough
that the two premiſes together affirm or deny of more than all
the inſtances of the middle term. If there be a hundred boxes,
into which a hundred and one articles of two different kinds are
 128
On the Aristotelian Syllogifm .
to be put, not more than one of each kind into any one box,
ſome one box, if not more, will have two articles, one of each
kind, put into it. The common doctrine has it, that an article
of one particular kind muſt be put into every box, and then ſome
one or more of another kind into one or more of the boxes, be-
fore it may be affirmed that one or more of different kinds are
found together. This excluſion is a ſimple miſtake, the mere
ſubſtitution of the aſſertion that none but a certain law of infe-
rence can exiſt, for the determination that no othershall exiſt.
Any one is at liberty to limit the inferences he will uſe, in any
manner he pleaſes : but he may err if he declare his own arbi-
trary boundary to be a natural limit impoſed by the laws of
thought.
The other excluſion may involve, on the ſame terms, an error
ofthe ſame kind ; or may equally be the expreſſion of arbitrary
will: but there is what is more reaſonably matter of opinion about
it. Ariftotle will have no contrary terms : not-man, he ſays, is not
the name of anything. He afterwards calls it an indefinite or
aorist name, becauſe, as he aſſerts, it is both the name of exiſting
and non-exiſting things. If he had here made the distinction
between ideal and objective, he would have ſeen that man and
not-man equally belong to both (objectively) exiſting and non-
exiſting things : man, for example, belongs as a name to Achilles
and the ſeven champions of Chriſtendom, whether they ever ex-
iſted in objective reality or not : and not-man belongs, in either
cafe, to their horſes. I think, however, that the exclufion was
probably dictated by the want of a definite notion of the extent
of the field of argument, which I have called the universe of the
propoſitions. Adopt ſuch a definite notion, and, as ſufficiently
ſhown, there is no more reaſon to attach the mere idea of ne-
gation to the contrary, than to the direct term.
The exclufion of contraries throws out the propoſitions E'
and I', or x.y and xy, which cannot be expreſſed without either
contraries, as in x.y=x)Y = y)X, and xy =x:Y =y:X, or refe-
rence to things not named by X and Y, as in ' Every thing is
either X or Y' and ' Some things are neither Xs nor Ys,' the
moſt natural readings of ' No not-Xs are not-Ys, and ' Some not-
Xs are not-Ys.' There remain then fix modes of connexion of
X and Y, namely X)Y and Y)X, X: Y and Y:X, and XY( =
 On the Ariftotelian Syllogifm. 129

YX) and X.Y( =Y.X). Theſe fix are made eight ; for in the
common ſyſtem, XY and YX are confidered as diftinct in form,
and alſo X.Y and Y.X. But theſe eight are only treated as
four : for reference to order is not made in the ſimple propofi-
tion. Thus X)Y and Y)X are both denoted by A, XY and
YX by I, X.Y and Y.X by E, and X: Y and Y:X by O. But
the ſtandard of order which is neglected as to the propoſition by
itſelf, is adopted in the ſyllogiſm in the following manner.
The predicate of the conclufion is called the major term, and
the ſubject of the concluſion the minor term. This language is
faſhioned upon the idea of an affirmative propofition, in which
major and minor have reference to magnitude. In ' every X is Z '
Z is a name which entirely contains X and is therefore at least
as great as X, greater than or equal to X. Here is, before it
was introduced into mathematics, the idea now ſo familiar to the
mathematician, of allowing his language to include the extreme
limit of its meaning. When the ſame terms are applied to
negative propoſitions, the notion of magnitudinal incluſion is
loft ; and major and minor, being ſtill retained, muſt be pre-
ſumed to refer to real or ſuppoſed importance. The premiſes
are called major and minor, according as they contain the major
or minor term ofthe conclufion : and the major premiſe is always
written firſt. Accordingly, Z and X being the major and minor
terms, there are four poſſible arrangements, which are called the
four figures. Aristotle gives three, and tradition has it that
Galen ſupplied the fourth in number and order.
1. YZ 2. ZY 3. YZ 4. ZY
XY XY YX YX

XZ XZ XZ XZ

To me, the moſt ſimple arrangement is that which takes up

what was left off with, as in the fourth figure : and ' X is in Y,
Y is in Z, therefore X is in Z ' is more natural than ' Y is in Z,
X is in Y, therefore X is in Z.'
It is now plain, that whenever one only of the three propofi-
tions is convertible, there are two distinct ways in which the
ſyllogiſm may be written : when two only, four : and when all
three (if there were ſuch a thing), eight.
K
130 On the Ariftotelian Syllogifm.
The ſyſtem rejects all concluſions which may be made
ſtronger : thus when X. Z follows, it does not allow X : Z to
make a distinct form. But when X)Z is the concluſion, it does
not reject ZX, for, not confidering ZX as identical with XZ, it
does not confider X)Z as a ſtrengthened form of ZX. But it
does not reject ſyllogiſms in which as ſtrong a conclufion can be
deduced from a weaker premiſe : accordingly, we muſt ſearch
for Ariftotelian forms among the ſtrengthened fyllogifms of
chapter V, as well as among the fundamental ones. Now,
taking all the forms which ſhow neither E' or I', let us write
down the ſymbols of them, and the number of caſes we may
expect from each. Moreover, ſince transformation of order
makes no difference here, I put the ſyllogiſms together as in
page 96, into twelve pairs .
Fundamental AAA , A'A'A' , 1 ; O'AO', A'O01, 1 ;
ΑΟ'Ο', ΟιΑ'Ο1, 1 ; E'A,E', A'E'E', rejected ; IAI, ALI, 4;
Ε'ΟΙ , ΟΙΕ'Ιι, rejected ; EA'Ε., Α.Ε.Ε. , 4 ; I'A'I', AI'I', re-
jected ; EOI , O'E.I', rejected ; E'EA', EE'A., rejected ;
IEO , ELO', 4 ; Ε'Ι'Οι, Ι'Ε'O', rejected.
Weakened AA , 1.
Strengthened A'AI , 1 ; AA'I' , rejected ; A'EO , ΕΑΟ',
2 ; ΑΕ'Ο', Ε'Α'Oı, rejected ; E'E'I., rejected ; E.E.I' , rejected.
There are then fifteen fundamental, one weakened, and three
ſtrengthened, forms of fyllogifm in the received ſyſtem. I now
put them down, with their derivations, forms of expreſſion in
full, ordinary ſymbols, figures into which they fall, and the magic
words by which they have been denoted for many centuries,
words which I take to be more full of meaning than any that
ever were made.

Fundamental.

AAA A'A'A' Y)Z + X)Y = X)Z AAA I Barbara

O'AO' A'OOY: Z + Y)X = X : Z OAO III Bokardo
AO'O' OLA'OZ)Y + X :Y = X : Z AOO II Baroko
LAI ALI Y)Z + XY = XZ AII I Darii
-

Y)Z + YX = XZ AII III Datifi

-
-
ZY + Y ) X = XZ IAI IV Dimaris
YZ + Y)X = XZ IAI III Diſamis
 On the Ariftotelian Syllogifm. 131
Fundamental.
ΕΑΕ Α.Ε.Ε. Y.Z + X ) Y = X.Z EAE I Celarent
Z.Y + X) Y = X.Z EAE II Cefare
Z) Y + Y.X = X.Z AEE IV Camenes
-

Z) Y + X.Y = X.Z AEE II Camestres

ELO' LEO Y.Z + XY = X : ZEIΟ I Ferio
Z.Y + XY = X :Z EIO II Festino
Y.Z + YX = X :Z EIO III Ferifon
Z.Y + YX = X :Z EIO IV Frefifon
Weakened.

AAL A'A'I Z)Y + Y)X = XZ AAI IV Bramantip

Strengthened.
A'AL A'ALY)Z + Y)X = XZ AAI III Darapti
A'EO EAO' Y.Z + Y)X = X : Z EAO III Felapton
-

Z.Y + Y) X = X :Z EAO IV Fesapo

The words which represent the different moods (as they are
called) are usually collected under their figures in the following
lines.
Barbara, Celarent, Darii, Ferioque prioris.
Cefare, Cameſtres, Feſtino, Baroko, fecundæ.
Tertia Darapti, Diſamis, Datiſi, Felapton,
Bokardo, Feriſon habet. Quarta infuper addit
Bramantip, Camenes, Dimaris, Feſapo, Freſiſon.

The vowels of the different words give the ſymbol of the

fyllogifm ; thus A,A,A, are ſeen in Barbara. The conſonants
in the firſt figure have no ſpecial meaning : but in the other
figures every confonant except T and N (which are only eu-
phonic) has its meaning as follows ;-every mood of every figure
can (with two exceptions) in one way or another, be reduced to
a mood ofthe firſt figure : and the letters ſhow the way ofdoing
it. The initial tells to which mood the reduction brings us :
thus Cefare is reduced to Celarent, and alſo Cameſtres ; Feſtino
is reduced to Ferio, and ſo on. The two exceptions are denoted
by the letter K (as in Baroko and Bokardo) ; we ſhall preſently
notice them further. And S means that the preceding premiſe is to
be ſimply converted. P, that what was called converfionperacci
132 On the Ariftotelian Syllogifm .
dens is to be made, ZX for X)Z, or X)Z for ZX : accordingly,
P only occurs in the weakened or ſtrengthened fyllogifms. M
means that the premiſes are to be tranſpoſed. Thus the meaning of
the word Diſamis is nothing less than what follows. There is a
ſyllogiſm in which the middle term is the ſubject of both pre-
miſes, and when reduced to the firſt figure it becomes Darii :
the major premiſe, which must be converted in reduction, is a
particular affirmative : the minor premiſe, which must become
the major one in reduction, is a univerſal affirmative : and the
conclufion, which must be converted in reduction, is a particular
affirmative.' Thus,

YZ + Y)X = XZ Diſamis
becomes Y)X + ZY = ZX Darii
The moods Baroko and Bokardo do not admit of reduction to
the firſt figure, by any fair uſe of the phrase: but the logicians
were determined they ſhould do ſo, and they accordingly hit
upon the following plan, which they called reduction per impoſſi-
bile. AOO and OAO being the opponent forms (pages 88,
and 102) of AAA, the two moods in queſtion were connected
with Barbara (whence their letter B) by ſhowing that the latter
would make the denial of their conclufions force one premiſe to
contradict the other. Thus, Baroko, or if Z)Y and X :Y then
X: Z was proved in the firſt figure as follows. If under theſe
premiſes, X:Z be not true, then X)Z is true ; but Z)Y is true :
and Z) Y + X)Z, by Barbara, gives X)Y. But X:Y : there-
fore, if Baroko be not a legitimate form, X)Y and X :Y are both
true at once, which is abſurd. Had contraries been uſed,
Z)Y + X : Y = X : Z would have been thrown into the firſt figure
as y)z + Xy = Xz, Darii, or y.Z + Xy = X : Z, Ferio. And
Y: Z + Y) X = X : Z, Bokardo, is ſeen reduced to the firſt figure
in Y)X + zY = zX, Darii.
Aristotle did not use the fourth figure, confidering it, as is
faid, to be only an inverſion of the firſt. The introduction of it
among the figures is attributed to Galen, and it does not often
appear in ordinary works of logic before the beginning of the laſt
century. If the order ofthe premiſes be inverted, ſo as to make
the firſt figure appear, the major and minor terms will appear
wrongly placed in the conclufion. The words uſed for theſe
 On the Ariftotelian Syllogifm. 133
indirect moods of the firſt figure were usually the fifth and fol-
lowing ones in
Barbara, Celarent, Darii, Ferio, Baralip-ton
Celantes, Dabitis, Fapeſmo, Friſeſom-orum
the final ſyllables in Italics being only euphonic (Friſeſmo-orum
would have been more correct). Some uſed the words Farefmo
and Firefmo.
In calling the moods of the fourth figure by the name of in-
direct moods of the firſt figure, notice was taken of the circum-
ſtance that a tranſpoſition of the premiſes would give the ar-
rangement of the firſt figure, in every thing but the proper
arrangement of major and minor terms, which is inverted. A
little confideration will ſhow the reader that the earlier Ariftote-
lians were wiſer than the later ones in this matter. Confider the
fourth and firſt figures as coincident, and the arbitrary notion of
arrangement by major and minor vaniſhes. It was not till this
mere matter of diſcipline was made an article of faith that the
fourth figure had any ground of feceffion from the firſt.
It might seem as if the union of the firſt and fourth figures
would demand that of the ſecond and third : the firſt pair con-
taining all the moods in which the middle term occupies different
places in the two premiſes, the ſecond pair thoſe in which it has
the fame place in both. If this were done, each of the two main
fubdivifions must be itſelf ſubdivided into two . And this would
perhaps have been the more ſkilful mode of diviſion.
The distinction of figures has been condemned by many, and
particularly by Kant. Whether attacked or defended, it is eſſen-
tial that the true grounds of the ſide taken ſhould be more ex-
plicitly ſtated than is often done. The root ofthe diſtinction of
figure is undoubtedly the distinction between the two forms XY
and YX, X.Y and Y. X. It would be equally abſurd, either to
deny the identity of XY and YX, confidered as material of
inference, or to deny their difference in many other points of
view. In this work I am concerned only with what can be
inferred, and to what extent of quantity, and accordingly the dif-
tinction is to me immaterial. But if I had not merely to ſtudy
the way of uſing premiſes, but alſo that of arriving at them, it
might very well happen that the aſpects under which the ſame
134 On the Ariftotelian Syllogifm .
inference is ſeen in different figures would give it very different
ſhades of character. A ſimple inſtance will ſhow that though
the compariſon, and its extent, are all that can be attended to in
forming the conclufion, theſe points ofmeaning are not the only
ones. A perſon who wished to contest the old uſe of the word
green, as applied to unripe fruit, would say that ' ſome green
fruits are ripe,' if he wanted ſpecially to ſhow the miſapplication
of the word. But if he rather wanted to ſhow the badneſs of
the method of denying ripeneſs, he would say ' ſome ripe fruits
are green.' The propoſitions are endleſs in which, X and Y
being the terms, it is at one time X which is brought to Y for
compariſon, and at another Y to X. The ſubject of a propofi-
tion is always the object of examination ; whether the form be
X) Y, X.Y , XY, or X : Y, we examine and report upon the Xs.
If we arrange the four figures ſeparately, we shall better fee
their ſeveral peculiarities.

First Figure.
Barbara Y)Z + X)Y = X)Z Celarent Y.Z + X) Y = X.Z
Darii Y)Z + XY = XZ Ferio Y.Z + XY = X : Z

What is here declared, is in every caſe the dictum de omni et

nullo in its ſimpleſt form, in a manner which juſtifies the prefe-
rence given to this figure. The middle term being completely
contained in, or completely excluded from, the major term ; ſuch
inclufion or exclufion then follows of all ſuch part of the minor
term as is declared in the ſecond premiſe to be in the middle
term . The inference then is in this ſentence What is true of
the whole middle term, is true of its part.' And it is obvious
that in this figure the major premiſe must be univerſal, the minor
premiſe affirmative. The four forms are all found among the
conclufions. I think that the inverſion of the premiſes which
the ſyſtem of chapter V. employs will be found to give the forms
which are moſt easily tranſlated into language independent of the
middle term . The fentence ' All (or ſome) ofthe Xs are what
must be Zs, therefore all (or ſome) of the Xs are Zs' includes
Barbara and Darii: and ' All (or ſome) of the Xs are what can-
not be Zs, and therefore cannot be Zs,' contains Celarent and
Ferio.
 On the Ariftotelian Syllogifm. 135
Second Figure.
Cefare Z.Y + X ) Y = X.Z Camestres Z)Y + X.Y = X.Z
Festino Z.Y + XY =X :Z Baroko Z) Y + X : Y = X :Z

In this figure (in which only negatives can be proved) the ap-
pearance of the dictum is not ſo direct. The terms of the
concluſion are both objects of examination, and one is wholly
included, and the whole or part of the other excluded (Cefare,
Cameſtres, and Baroko) or one is wholly excluded, and the whole
or part of the other included (Cefare, Cameſtres, and Feſtino).
Or rather, to justify the distinction, we should say that the whole
included
of the major term is and the whole of the minor
excluded
excluded Cameſtres
in which the whole of the minor
included which gives Cefare
is therefore excluded from the major ; or elſe the whole of the
included excluded
major is excluded and part of the minor included which gives
Baroko in which that part of the minor is excluded from the
Feſtino
major. And it is evident enough why the premiſes must be of
different ſigns .
In the firſt figure, though all the forms be eſſentially one,
(page 98,) the reduction of either to the form Barbara requires
either the explicit uſe of contraries, or invention of a name fub-
identical to X. Accordingly, no mood of that figure is reducible
to any other by the uſually admitted reductions. But this cannot
be ſaid of any of the other figures. In the one before us, Cefare
and Cameſtres are identical, even without changing the figure.
That which is Cefare when X is major and Z minor, is Camef-
tres when X is minor and Z major. In the firſt figure, the
fame attempt made on Celarent or Darii, removes them into
another figure.
Third Figure.
Darapti Y)Z + Y)X = XZ | Felapton Y.Z + Y)X = X : Z
Diſamis YZ + Y)X = XZ Bokardo Y:Z + Y)X =X :Z
Datifi Y)Z + YX = XZ Ferifon Y.Z + YX = X :Z

The firſt and ſecond figures contain a pair of univerſals each,

136 On the Ariftotelian Syllogifm .
with one particular derived from each,by a legitimateweakening of
one premiſe and the concluſion at the ſame time : but in no in-
ſtance is the quantity of the middle term weakened. And all
the fyllogifms in theſe two figures are fundamental (page 77) .
In the cafe now before us, both the leading fyllogifms are not
fundamental, but ſtrengthened, and capable of being weakened in
two different ways. The middle term is here examined in both
premiſes : if it be wholly included in, or excluded from, one of
the concluding terms, and wholly or partly included in, or ex-
cluded from, the other (but not ſo that there ſhall be exclu-
fion from both) we have it that the whole or part mentioned in
one caſe is included in, or excluded from, that which the whole
is included in, or excluded from, in the other. There can be
none but particular concluſions.

Fourth Figure.
Camenes Z)Y + Y.X = X.Z
Bramantip Z)Y+Y)X=XZ |
Fefapo Z.Y+ Y) X = X : Z
Freſiſon Z.Y + YX = X : Z
We have now one univerſal fyllogiſm in a form which does not
admit of being weakened in this figure, and two ſtrengthened
fyllogifms, each of which has one weakened form, one of them,
Bramantip, admitting a ſtronger concluſion in another figure.
Every concluſion except A appears. The mode of inference of
the three firſt ſyllogiſms has been deſcribed in the other figures .
In Feſapo and Frefſiſon, the perfect excluſion of the major term
from the middle, accompanied by the total or partial incluſion of
the middle in the minor, ſecures the exclufion from the major,
of as much of the minor as it has in common with the middle .
I shall now proceed to the rules uſually given, and to fome
remarks on the degree in which they apply to the more general
ſyſtem in chapter V. Aldrich gives them as follows-
Diſtribuas medium : nec quartus terminus adfit :
Utraque nec præmiſſa negans, nec particularis :
Sectetur partem conclufio deteriorem ;
Et non diſtribuat, nifi cum præmiſſa, negetve.
 On the Ariftotelian Syllogifm. 137
Theſe rules, I need hardly ſay, are perfectly correct, when the
contraries of the terms are excluded, and alſo all notion of quan-
tity except all, or the indefinite ſome. Taking them in the natu-
ral order, which verſification has a little disturbed, we have ; -
1. There are to be but three terms, of which it is understood
two only appear in the conclufion, the excluded or middle term
appearing in both of the premises. This is true in my ſyſtem,
when by terms are understood alſo contraries of terms. I ſhould
ſuppoſe that there can be no objection to the admiffion of con-
traries, unleſs there be one to the conception of a contrary. Any
one may, with Aristotle, object to the word not-man, as not the
name of anything : on the grounds which immediately induced
him to call it an aoriſt, or indefinite, name. But it can hardly
be affirmed that any one admitting not-man as a name, ſhould
thereupon refuſe to recogniſe the identity of ' horſe is not man,'
with ' horſe is not-man .'
2. The middle term is to be distributed in one or the other of
the premiſes. By diſtributed is here meant univerfally spoken of.
I do not use this term in the preſent work, becauſe I do not fee
why, in any deducible meaning of the word distributed, it can be
applied to univerſal as diftinguiſhed from particular. In uſing a
name, it seems to me that we always diſtribute : that is, ſcatter
as it were, the general name over the inſtances to which it is to
apply. When I ſay ſome horſes are animals, I diſtribute certain
horſes among the animals ; and when all, all. Leaving the word,
the principle is one which clearly must be true whenever we are
restricted in quantity to all or ſome (indefinite), and when con-
traries are not admitted. In the former caſe we have, in one
form or another, to make m + n greater than η (chapter VIII. )
when we cannot know what relation either m or n has to n, unleſs
one of them, or both, be equal to n. We have no alternative
then, but to require that m or n ſhall be n. The cafes in which
there is apparently no dependence on η will be diſcuſſed in
the next chapter.
But when contraries are introduced, this rule is not univerſally
true. The exception is ſeen in
AA'I ' or X)Y + Z)Y =xz .
If all the Xs be Ys, and alſo all the Zs, it follows that there are
things which are neither Xs nor Zs, namely, all which are not
 138 On the Ariftotelian Syllogifm.
Ys. It is here, as elsewhere, implied that the middle term is
not the univerſe of the propofition.
When we come, then, to uſe contraries, the ſimple rule of the
middle term is no longer univerſally true. What other rule are
we to put in its place ? We know, of course, that every fyllo-
giſm can be reduced to an Ariftotelian ſyllogiſm, and even to one
or other of two among them, AAA or IAI, or to the firſt of
theſe, if we be at liberty to uſe invention of names (page 97) .
Again, each term, or its contrary, is mentioned univerſally in
every propofition : ſo that there is certainly one way in which
every pair of premiſes may be made to exhibit a middle term
univerſally uſed in one of them. The rule to be ſubſtituted for
the distribuas medium is, that all pairs of univerſals are con-
cluſive, but a univerſal and a particular require that the middle
term ſhould alſo be a univerſal and a particular, that is, univerſal
in one and particular in the other. Thus, in X)Y + Z )Y, as it
ſtands, the middle is particular in both ; tranſpoſe into y)x +y)z
and the middle is now univerſal in both, by which we see the
Ariftotelian concluſion. Again, in X) Y + ZY, which is of the
ſame kind, the tranſpoſition gives y)x + Z: y, which is faulty,
becauſe, though there be a particular premiſe, there is not any-
where a particular middle term. The caſes in which the middle
is of the ſame name in both places (univerſal in four, particular
in four), are the ſtrengthened ſyllogiſms only. There is nothing
to be ſurpriſed at in its thus appearing that the particularity of
the middle term is just as much a teſt of a good ſyllogiſm as its
univerſality : of every name and its contrary, one enters univer-
fally, and one particularly, in every propoſition which contains
it ; and the ſyſtem in chapter V. is as much concerned with con-
trary as with direct terms. It is thence viſible beforehand, to the
mathematician at leaſt, that any teſt must be defective, unleſs
univerſal and particular enter into it in the ſame manner.
The above contains a complete canon of validity, as ſoon as
the law of the three terms is understood, which is only a law of
definition. We may ſtate it as follows : Two premiſes conclude
-when both are univerſal, always ; when one only is univerſal,
ſo often as it happens that the middle term (be it Y or y) is once
only univerſal ; when neither is univerſal, never. By this rule
alone the thirty-two concluſive caſes can be diftinguiſhed from
the thirty-two inconclufive ones.
 On the Ariftotelian Syllogifm. 139
3. When both premiſes are negative, there is no Ariftotelian
fyllogifm. In the ſyſtem completed by contraries, there are eight
ſuch fyllogifms , as many in fact, as there are with premiſes both
affirmative. But a pair of negative premiſes never conclude with
both terms of the premiſes, but with the contrary of one or
both : and this must be ſubſtituted, as a rule of conclufion, for
the one juſt named.
4. Both premiſes muſt not be particular. This rule, which
relates wholly to quantity, muſt be preſerved in every ſyſtem
which admits no definite ratio, except that of one to one, or
all (pages 56, 57). I cannot learn that any writer on logic
ever propounded even the very fimple caſe of ' Moſt Ys are
Xs, moſt Ys are Zs, therefore ſome Xs are Zs,' as a legiti-
mate inference. And this, though it is certain that the quanti-
tative prefix moſt (plurimi) has before now excited diſcuſſion as
to whether it belonged to a univerſal or a particular.
5. By ſectetur partem conclufio deteriorem it is understood that
the negative is called weaker or lower (deterior) than the affir-
mative, and the particular than the univerſal ; and that the con-
cluſion is to be as weak as negative, or as particular, if there be
a premiſe which is negative or particular. This rule must be
preſerved, when contraries are introduced, ſo far as relates to
particulars . But ſo far as negatives are concerned, the rule muſt
be that one negative premiſe gives a negative conclufion, and two
an affirmative one.
7. The last line, et non distribuat, nifi cum premiſſa, negetve,
ſpoils the ſymmetry to procure a verſe. The concluſion is not
to be negative without a negative premiſe : that is, affirmative
premiſes give an affirmative conclufion. Alſo, no term is to be
diſtributively, (i. e. univerſally) taken in the concluſion, unless it
were ſo taken in its premiſe. A breach of this rule would be
equivalent to drawing a conclufion about what was not (or about
more than was) introduced into the premiſes.
When contraries are introduced, the diſtinction between pofi-
tive and negative is made to appear, what it really is, one of
language, or rather one of choice of names. But the distinction
of form is not aboliſhed, but is exactly what it was before. We
cannot lay down any rules for the formation of the conclufion
unleſs, in our eight ſtandard forms, we preſerve the mode of
140 On the Ariftotelian Syllogifm.
writing which belongs to the fundamental derivation of the
forms (page 61 ). Thus, the order being XY, A' is x)y and not
Y)X, and O' is x: y and not Y: X. This method of writing
being restored, when neceſſary, in pages 89 and 91 , it follows
immediately that the rule of accentuation in the notation gives
the rule by which we determine whether the conclufion takes
the terms from the premiſes, or prefers contraries. According as
the prepofition of the concluſion agrees with or differs from that of
a premiſe, ſo does the conclufion take a term from that premiſe,
or its contrary. Thus, AAA takes both terms from the pre-
miſes, but A.A'I' takes a contrary from the firſt premiſe only.
This laſt we fee if we write the ſyllogiſm as X)Y +y)z =xz .
Accordingly, we have-
Syllogisms taking both concluding terms directfrom the premises.
Univerſals which begin with A ; particulars which begin with I :
eight in number ; being all which iſolate no accent.
Taking the first term only from the premise. Univerſals begin-
ning with E ; particulars beginning with O : eight in number ;
being all which iſolate the middle accent.
Taking the second term only from the premise. Strengthened
forms and particulars which begin with A : eight in number,
being all which iſolate the firſt accent.
Taking neither term from the premises. Strengthened forms
and particulars which begin with E : eight in number, being all
which iſolate the third accent.
This is a new mode of ſtating the law of accentuation (pages
92-3) which I have preferred to place here, for fear of overload-
ing chapter V. with rules. I have not ſtated one half of thoſe
which ſuggeſted themselves. This multiplicity of relations is a
preſumption of the completeness of the ſyſtem.
In the Ariftotelian ſyſtem, there is multiplication of the fame
modes of inference, under the varieties of figure. In that which
I propoſe, there is a reduplication of most of the eſſential caſes ;
for whatever cafe is found, the fame is alſo found with X and Z
interchanged, and alſo the order of the premiſes. Again, whatever
cafe is found, it is found contranominally ; or with all the accents
(or prepofitions) altered. There are other ways (and many of
them) in which the ſyſtem is only in one half a duplicate of what
it is in the other. If all theſe modes of dividing the ſyſtem into
 On the Ariftotelian Syllogifm. 141

two correlative parts divided it into the ſame two parts, there can
be no queſtion that one alone of thoſe parts ſhould have been
preſented as the object of confideration. But this does not hap-
pen in any inſtance : ſo that it is impoſſible to diſpenſe with the
whole of the thirty-two caſes. The Ariftotelian caſes do not
form or include any half whatever of this ſyſtem.

CHAPTER VIII .

On the numerically definite Syllogifm.

N the last chapter I confidered no other quantity in names
I except all and ſome: the latter meaning ،
one or more, it
may be all.' To this extent of quantity we are limited in moſt
kinds of reaſoning, by want of knowledge of the definite extent
of our propofitions : and the few phrases (page 58), as ' moſt,'
' a good many,' &c. by which we endeavour to eſtabliſh differ-
ences of extent in ordinary converſation, have been hitherto held
inadmiſſible into logic. In this ſcience it ſeems to have been
always intended that the baſes on which its forms are conſtructed
ſhall be nothing but the ſuppoſition of the moſt imperfect and
inaccurate knowledge. Though in geometry we are permitted
to aſſume as the object of reaſoning the ideal ſtraight line, the
' length without breadth' of Euclid, which has no objective pro-
totype, and though we fee the advantage of reaſoning upon ideas,
and allowing the eſſential inaccuracies of material application to
produce no effect except in material application,-yet in the con-
fideration of the pure forms of thought, the learner has always
been denied the advantage of ſtudying the more perfect ſyſtem of
which his inferences are the imperfect imitation.
The ordinary univerſal propoſitions are of a certain approach
to definite character, both of them with reſpect to their ſubjects,
and the negative one with reſpect to its predicate alſo. In X)Y
for example, what is known is as much known of any one X as
of any other. Perfect definiteneſs would conſiſt in a more exact
degree of deſcription, and would require a higher degree of know-
ledge. But in this chapter I ſpeak only of numerical definite
 142 On the numerically
neſs, of the ſuppoſition that we know how many things we are
talking about. We may be well content to examine what we
ſhould do if we were a ſtep or two higher in the ſcale of creation,
ifby ſo doing we can manage to add ſomething to our methods
of inference in the highest to which we have as yet attained.
A numerically definite propoſition is of this kind. Suppoſe
the whole number of Xs and Ys to be known : ſay there are
100 Xs and 200 Ys in exiſtence. Then an affirmative propo-
ſition of the fort in queſtion is ſeen in ' 45 Xs (or more *) are
each of them one of 70 Ys' : and a negative propofition in
'45 Xs (or more) are no one of them to be found among
70 Ys.'
But it must be particularly noticed that in ſpeaking of a num-
ber of Xs, as 45 Xs, I do not mean certain 45 Xs which can
be diftinguiſhed from all the reſt, ſo that of any X it is poſſible
to be known whether it belong to the 45 of the propofition, or to
the remaining 55. This degree ofdefiniteneſs is one ſtep higher
than that which I here propoſe to confider, and which is deſcribed
by ' there are 45 Xs which are contained among 70 Ys, it not
being known which Xs are the 45 Xs, nor which Ys are the
70 Ys :' or elſe by ' there are 45 Xs which are not any of them
identical with any one of 70 Ys, the preciſe Xs and Ys in
queſtion being unknown.'
It cannot of courſe be diſputed that if any thing ſhould necef-
ſarily follow from any 45 Xs being found among any 70 Ys, it
will not the leſs follow from our knowing which are the Xs and
which are the Ys. But this laſt ſuppoſition only brings us to
really univerſal propoſitions. If, there being 100 Xs, 45 of
them can be ſpecifically ſeparated from the reſt, ſo as to be
known, the proceſs of ſeparation is equivalent to putting them

* Theſe words (or more) ſhow that the word definite has reference only
to the lower boundary. Of courſe nothing can be ſhown in right of " 45
or more, perhaps " except what is true in right of the 45. It is defirable
that as the premiſes, ſo ſhould be the concluſion, of a ſyllogiſm : this would
not be the caſe if we uſed premiſes definite both ways. For example, there
being 100 Ys in exiſtence, it will preſently appear that ' Exactly 55 Ys are
Xs and exactly 60 Ys are Zs, though it enable us to ſay that '15 Xs are
Zs' does not allow us to ſay ' Exactly 15 Xs are Zs, but only '15 Xs (or
more) are Zs . '
 definite Syllogifm. 143

under a ſeparate name, ſubidentical to X, and the reſt, which are

equally diftinguiſhable, under another name, alſo ſubidentical to
X, and contrary of the firſt name, when the univerſe is X.
Whether the name be long or ſhort, does not matter, nor
whether it carry the ſeparating distinction in its etymology or
not. To ſeparate in any way inſtance from inſtance by lan-
guage, is to name.
If then 45 definite Xs were known to be contained among
70 definite Ys, and if theſe Xs were each named M, and thoſe
Ys each N, and if the rest of the Xs and Ys were named P
and Q, we should have the following propoſitions,
M)X, P)X, N) Y, Q) Y, M) N, M.P, N.Q,
and all inferences. Moreover, in each cafe, we should have the
total number of inſtances which are contained under each name ;
the numbers carrying with them evidence that every X is either
Mor P, and every Y either N or Q. Subſtitute M.N for
M)N and we have the correſponding negative propoſition.
But if 45 unſeparated and inſeparable Xs be ſuppoſed known
each to be among 70 fimilarly ſituated Ys, there is no immediate
method of making any other propoſition out of the terms X and
Y except its converſe, that 45 of theſe 70 Ys are 45 Xs, and (if
the whole number of Ys be known, ſay 200) that there are 45
Xs which are not any one among 200-70, or 130 Ys. This is
then a ſimple propoſition, which becomes of a highly complex
character, when the Xs and Ys named in it are taken as defi-
nitely ſeparable from the reſt. I ſhall call it the ſimple numerical
propofition.
The distinction may be easily illuſtrated by example. " All
the planets but one' is a particular propoſition ; it is ' fome
planets :' there is no one planet of right included in it. But
6

' all the planets except Neptune' is a univerſal propofition : a-

planet-not-Neptune' is a name of Mercury, of Venus, &c.;
and of every planet it can be ſtated whether it be in the name
or not. That which is true inferentially of ' all the planets but
one' left particular, is true of all the planets but Neptune :'
but that which is true of the latter is not neceſſarily true of the
former.
Taking X, Y, Z as the terms of the ſyllogiſm, & the number
144 On the numerically
ofXs in exiſtence, η the number of Ys, and & the number of
Zs, and v the number of inſtances in the univerſe, there are of
courſe ſixteen poffible caſes of knowledge, more or leſs, of theſe
primary quantities, from all unknown to all known. Of theſe
fixteen caſes, it will be requifite to confider two only. First,
when the extent n of the middle term is known, and all the
reſt unknown ; ſecondly, when all are known. The algebraical
formulæ of the latter caſe will enable us to point out how the
ſuppoſition of any leſs degree of knowledge would affect our
power of inference.
I propoſe the following notation. Let mXY denote either of
the equivalent propoſitions, that m Xs are to be found among the
Ys, or that m Ys are to be found among the Xs. Let mX : nY
denote either of the equivalent propoſitions, that there are m Xs
which are not any one among n Ys, or n Ys which are not any
one among m Xs .
The ſymbol 10X is the algebraical ſymbol for ten equal Xs
added together, ✗ being a magnitude : it is then a collective
ſymbol. In this work, X being a name, it implies every one
out of ten inſtances of that name, diſtributively, but not collec-
tively. This diſtinction is very material, not only in this chap-
ter, but throughout every part of logic. Every X is Y ' is
diſtributively true, when, by ' Every X' we mean each one X :
ſo that the propoſition is ' The firſt X is Y, and the ſecond X
is Y, and the third X is Y, &c.' In this caſe the ſubject is X,
and the word every belongs to the quantity of the propoſition.
But every X is Y' is collectively true, when we do not mean
that any one X is a Y, nor that any number of Xs are Ys, but
that all the Xs make a Y. In this caſe the propoſition is fin-
gular : there is but one inſtance of the ſubject mentioned, that
ſubject being, not X, but the collection ' all the Xs.' Thus ' the
ten men are members of a committee' is diſtributive : ' the ten
men are a committee' is collective.
If, in ſuch a propoſition as 10XY, we were to ſuppoſe the
10 Xs ſpecifically ſeparated from the reſt, being certain affignable
ten individuals from among all the Xs, then 10X becomes a name
for each of the ten, as much as X, and may be confidered as a
univerſal term. And now IOXY and { 10X} ) Y mean the fame
things.
 definite Syllogifm. 145
Let n be known, and n only ofthe four, υ, ξ, η, ζ. The only
collections of premiſes which it is neceſſary to confider are
mXY + nYZ
mXY + nZ : SY
mX : rY + nZ : SY

Without ſome knowledge of the number of ys, of which by

ſuppoſition we have none, it would be uſeleſs to attempt to draw
an inference from a pair in which Y and y enter together, par-
tially quantified, as in mXY + nZ : ry. And nZy merely amounts
to nZ : nY .
The above three are all we need conſider : and even of theſe
the third is incapable of inference, ſince both premiſes are nega-
tive, and moreover, not reducible to a poſitive form by uſe of
contraries, the only way in which negative premiſes really acquire
a conclufion in chapter V.
Let us firſt confider the premiſes mXY + nYZ. They tell
us that among the n Ys we find m Xs and n Zs : accordingly,
neither m nor n exceeds n. If m and n together fall ſhort of n,
nothing can be inferred : Y is extenſive enough (that is, there
are inſtances enough of Y) to hold the m Xs and the n Zs with-
out any coincidence of an X with a Z. As to other Xs or Zs,
we do not know whether they exiſt ; or, if they exiſt, we do not
know that any one of them is a Y. But if m and n together
exceed n, it is impoſſible that m Xs and n Zs can find place
among η Ys, except by putting either two Xs or two Zs, or an
X and a Z, with one of the Ys. Now as by the nature of the
ſuppoſitions, there cannot be two Xs, nor two Zs, to one Y, we
must have the inference IXZ as often as there are units in the
exceſs of m + n over η . That is,
mXY + nYZ = (m + n-n) XZ
Next, let us take mXY + nZ : sY. There may be two in-
ferences, perfectly distinct from each other, the connexion of
which can only be explained in the more general ſyſtem to which
we ſhall preſently come. Firſt, let m and s together exceed n.
Then m + s-n of the Ys have the common property of being
Xs, and of being clear of the n Zs. Accordingly, we have
mXY + nZ : sY =(m + s-n)X : nZ
L
146 On the numerically
Next, let n + s be greater than η. Take the s Ys among which
no one of the n Zs is found. Becauſe n + s is greater than η, η
is greater than n-s, the number of Ys left. Accordingly,
n-(n-s) of the nZs cannot be any Ys, and therefore cannot
be any of the mXs which are Ys. Hence we have
mXY + nZ : sY =mX : (n + s-n)Z
In the appendix to this chapter (at the end of the work) will
be ſeen the manner in which all the Ariftotelian ſyllogiſms can
be brought under the firſt caſe, and the firſt * inference of the
ſecond cafe. No Ariftotelian fyllogifm can be deduced from the
ſecond inference except when s= n, in which caſe it agrees with
the firſt. For, when s is not n, we muſt, to make fuch a fyllo-
giſm, have m = n, and then, to make nZ: sY Ariftotelian, s not
being n, we must have all the Zs in n, or n = ζ. We thus get
Y)X + SY: Z, the premiſes of Bokardo. But the concluſion is
ηΧ: ( + s-n)Ζ, that of Bokardo being sX : Z. And this will
be found to be the only Ariftotelian ſyllogiſm which has this
ſecond and numerically quantified inference, depending upon the
number of Zs exceeding the number of Ys unnamed in the par-
ticular premiſe.
I now proceed to ſuppoſe that all the quantities are taken into
account. Some preliminary confiderations will be useful, as
follows .
Let two propoſitions be called identical, when, either of them
being true, the other must be true alſo : ſo that nothing can be
inferred from the one, which does not equally follow from the
other. Such propoſitions are X.Y and Y.X, ſuch are X)Y
and y)x, and ſo on. Again, two propoſitions may be identical
relatively to a third : thus, P being true, Q and R may either
follow from the other ; accordingly, as long as it is understood
that P is true, Q and R may, relatively to that ſuppoſition, be
treated as identical.
The word identical, as applied to propoſitions, is here made to
mean more than uſual, but not with more licenſe than when the
word is applied to names. Thus, man and rational animal are

* I was not in poſſeſſion of the ſecond inference till I had written what is
in page 157 .
 definite Syllogifm. 147
not identical names, quà names, for they neither ſpell nor found
alike : the identity understood is that of meaning ; where one
applies, there ſhall the other apply alſo. Similarly, as to propo-
ſitions (of which ſubject, predicate, and copula are the material
parts, juſt as ſpelling and found are thoſe of names), identity does
not confift in ſameneſs ofparts, nor in reducibility to ſameneſs, but
in fimultaneous truth or falſehood, ſo that what either is, be it true
or falſe, the other is alſo, in every cafe. Thus two propofitions,
one of which fignifies that an end has been gained, and the other
that the fole and ſufficient means of gaining it have been uſed,
are identical.
All the theory of names, their application or non-application,
may be applied to propoſitions, their truth or falsehood. To fay
that a propoſition is true in a certain caſe, is to ſay that a certain
name applies to a certain cafe: to say that it is falſe, is to ſay
that a certain name does not apply, but that its contrary does.
That contrary is what logicians uſually call contradictory : and
the name is not ſimply true or falſe, but the adjective attached to
the propofition. The conditions under which we are to ſpeak
limit us to a number of caſes which conſtitute what we may now
call, not the univerſe of the names in the propoſitions, but the
univerſe of the truth or falſehood of the propofitions. Thus we
ſhall ſuppoſe ourſelves now to be ſpeaking, not of all inſtances to
which the name U applies, but of all in which the propofition U
is true, or in which the name ' true U' applies. A caſe in
which a propofition P is true may be marked P, one in which it
is falſe, p. We may now apply the names ſubidentical, &c. and
the ſymbols, together with all the fyllogifms, complex and ſimple ;
but on each a remark may be neceſſary.
Subidentical, identical, and fuperidentical. If P be a propo-
ſition ſubidentical of Q, that is, if every caſe in which P is true be
one in which Q is true, but ſo that Q is ſometimes true when P
is not, the propofition Q is usually mentioned as effential to P,
and as a neceſſary confequence of it. Whenever P is true, Q is
true ; Q neceſſarily follows from P ; if Q be falſe, P cannot be
true ; Q is eſſential to P ; are all mere ſynonymes. Accord-
ingly ' neceffary confequent' and ' fuperidentical or identical' are
ſynonymous terms : that is (page 68), neceſſary confequent and
superaffirmative. Identity of courſe conſiſts in each propofition
 148 On the numerically
being true when the other is true. I think that, according to
general notions, it would be held more juſt to ſay that a propo-
ſition contains its neceſſary conſequence than that it is contained :
but a moment's confideration will ſhow that the latter analogy is
at least as found. If the ſecond be true whenever the firſt is
true, it may be true in other cafes alſo : ſo that we only ſay the
ſecond contains the firſt, and it may be more.
Subcontrary, contrary, and ſupercontrary. It is uſual to call
' No X is Y' and ' Every X is Y' by the name of contraries,
and to ſay that ' contraries may be both falſe, but cannot be both
true.' This is a technical uſe of the word : in common lan-
guage we ſhould ſay that either a propofition or its contrary muſt
be true ; ' have you any thing to ſay to the contrary' generally
means what a logician would expreſs by putting the word con-
tradictory in the place of contrary. I am compelled to uſe the
words contrary and contradictory as ſynonymous : at which com-
pulfion I am well pleaſed, never having ſeen any good reaſon
why, in the ſcience which confiders the relations of dicta, the
contraria ſhould be any thing but the contra dicta. The proper
word for contrary, commonly uſed to expreſs the relation of X)Y
and X.Y, is fubcontrary. Here are two propoſitions P and Q
which cannot both be true, but may both be falſe : here is a pair
which can never be aſſerted of the fame inſtance, and of which,
in many inſtances, neither can apply. In the ſame manner, the
propoſitions XY and X :Y, uſually called fubcontrary (for no
reaſon that I can find except that they are written under the ſo
called contraries in a ſcheme or diagram very common in books
of logic) ſhould be called fupercontrary : they are never both
falſe, and may be both true. This is a complete inverſion of
the uſual propoſitions : an inverſion which ſeems to me to be
imperatively required, if only my uſe offub andsuper in Chapter
IV. be allowed .

In applying theſe names to propoſitions, it must be remem-

bered that we make the ſame ſort of aſcent which we make in
paffing from ſpecific to univerſal arithmetic, in uſing a ſymbol
to ſtand for any number at pleaſure. For instance ;-Perhaps
it may be thought that XY and X :Y may ſometimes be only
contraries, and not ſupercontraries, becauſe there may be names
which make one only true and not both. But this is not correct :
 definite Syllogifm. 149
for we are confidering the propofition itſelfas an instance among
propofitions, not the propofition as fubdiviſible into inſtances, in
which name is compared with name. In ſpeaking of propo-
fitions, it is change from uſe of one name to uſe of another, or
from uſe of one number to uſe of another, which is change of
inſtance : not change from one inſtance of name to another.
And juſt as in a univerſe of names, every name introduced is
ſuppoſed to belong, or not to belong, to every inſtance in that
univerſe : ſo in a univerſe of propoſitions, I ſuppoſe every propo-
ſition, or its contrary, to apply (whether it be or be not known
which applies) in every inſtance. We have never confidered
ſuch a thing as the univerſe U, in which there are caſes in which
neither X nor x applies : we ſuppoſe there is always a power of
declaring that the name X muſt either belong or not belong
to each inſtance. In like manner, all the propoſitions in each
univerſe now conſidered, are ſuppoſed to be connected with all
the names in queſtion : ſo that X, Y, being two of them in their
order of reference, A,or O is true in each caſe, and A' or O',
E or I , and E' or I' . We might, if we pleaſed, enter upon a
wider ſyſtem. For though we cannot imagine of any object of
thought, but that it is either X or not X, be X what name it
may, yet we can imagine of propoſitions that they may be wholly
inapplicable, as being neither true nor falſe. The firſt aſſertion
is all the more true, that it could hardly be exemplified without
exciting laughter : as I ſhould do if I reminded the reader that a
book is either a cornfield or not a cornfield. We have never
confidered names under more predicaments than two ; never,
for inſtance, as if we were to ſuppoſe three names X1, X2, X3, of
which everything must be one or the other, and nothing can be
more than one. But we ſhould be led to extend our ſyſtem if
we confidered propoſitions under three points of view, as true,
falſe, or inapplicable. We may confine ourſelves to ſingle alter-
natives either by introducing not-true (including both falſe and
inapplicable) as the recognized contrary of true : or elſe by con-
fining our reſults to univerſes in which there is always applica-
bility, ſo that true or falſe holds in every caſe. The latter hypo-
theſis
Thiswilldigreffion
beſt ſuit my preſent purpoſe.
is ſomewhat out of place here, but I have
preferred to retain the matter of it until I had occafion to use it.
150 On the numerically
I now proceed to affert that the ſimple numerical propofition
has no occaſion for a numerically definite predicate. Let us
confider firſt an affirmative propofition, ſay Of 10 Xs, each
is to be found among ſome 15 Ys.' Of course it is ſuppoſed
there are 15 or more Ys in exiſtence. With this let us compare
' 10 Xs are to be found among the Ys.' Theſe two propoſitions
are identical : if 10 Xs be among 15 Ys, there are 10 Xs among
the Ys : and if 10 Xs be among the Ys they are certainly 10
Ys ; put on 5 more Ys at pleaſure, and they can be faid to be
among 15 Ys in juſt as many ways as we can chooſe 5 more Ys
to make up the 15. Note, that ifthe 10 Xs were among certain
ſpecified 15 Ys, then, though the firſt propoſition would give the
ſecond, the ſecond would not neceſſarily give the firſt. But we
are now ſuppoſing that numerical ſelection is only numerically
definite : definite as to the number, not as to the inſtances which
make up that number. When therefore we say ' 10 Xs are
among 15 Ys' we ſay neither more nor leſs than when we ſay
10 Xs are among the Ys.' It is in fact ' 10 of the Xs are 10
of the Ys ' and the converſe 10 of the Ys are 10 of the Xs' is
the fame propofition.
Now let us take a negative propofition, ' 10 of the Xs are not
to be found, any one of them, among ſome 15 Ys,' abbreviated
into ' 10 Xs are not in 15 Ys.' If there be 25 Ys in exiſtence
this propofition must be true ; mean X and Y what they may.
It is as true as that the X which is one Y is not any other Y.
Say there are 25 or more Ys : take any 10 Xs you chooſe, and
put them down on any 10 Ys you chooſe. Then certainly
there are 15 Ys left, no one of which is any of thoſe 10 Xs.
Again, if there be 25 Xs in exiſtence, ſtill the propoſition muſt
be true. For if the 15 Ys were all there are, and they were all
Xs, there ſtill remain 10 Xs which are not any one in the 15 Ys.
Accordingly, the propoſition ' m Xs are all clear of n Ys,' when-
ever either the whole number of Xs, or the whole number of
Ys, exceeds m + n, ſays no more than is conveyed in our perma-
nent underſtanding that no object of thought can be more than
one X or one Y. But let it be otherwise ; let neither Xs nor
Ys be as many as m + n in number. Say there are 20 Xs and
23 Ys and let 10 Xs be clear of 15 Ys. There must now be
at least 15 + 10-20, or 5 Ys which are no Xs at all, and at
leaſt 15 + 10-23, or 2 Xs which are no Ys at all. First, it is
 definite Syllogifm. 151
plain that there are no 10 Xs among thoſe Ys which are clear of
15 Ys : for there are but 23 Ys in all. Therefore, 2 at least
of theſe 10 Xs must be Xs which are not Ys : which with 8
Xs that may be Ys, will be clear of the remaining 15 Ys.
Therefore 2 Xs at least are not Ys. Again, there are no 15
Ys among thoſe Xs which are dear of 10 Xs, for there are but
20 Xs in all. Five Ys which are not Xs muſt exiſt, which
with 10 that may be Xs, will be clear of the remaining 10 Xs .
Accordingly, if the whole number of Xs be , and the whole
number of Ys be n, the propofition ' there are m Xs which are
no one to be found among n Ys' is eſſentially true of every caſe of
that univerſe, whenever m + n is leſs than either & or η. But
when m + n is greater than both & and n, there are two propo-
ſitions, neceſſarily involved, which are not eſſentials of all caſes
ofthat univerſe : namely, that there are m + n - Ys which are
not any Xs, and m + n - n Xs which are not any Ys .
But, it may be aſked, if n ſhould be less than 4, and m + n
greater than n, but ſtill leſs than 4, may we not affirm that
m + n - n Xs are not Ys ? Undoubtedly we may, but then we
do not affirm ſo much as already belongs to every caſe of the
univerſe. For if & be greater than w, no more than η Xs can be
Ys, and there are left -n Xs which cannot be Ys : and ξη
is, in the caſe ſuppoſed, more than m + n- n.
Let v be the number of inſtances in the univerſe, & and η being
the number of Xs and of Ys. The following uſes ofthe notation
will be readily ſeen to expreſs preceding reſults, or others imme-
diately deducible.
¼ greater than n (ξη)Χ: ηΥ or ( ξη)Xy
η greater than ( - ) : X or (η - ξ) Yx
m + n greater than & and than n gives
mX : nY =(m + n - n)X : yY =(m + n )Y: EX
AX) = XY =( - n)yx = Y: ( - 4)x
OX :Y =mX : nY =mXy =my: ( - 4)x
A' Y)X = nXY =( - 4)xy = nX : (v- n)y
O'Y : X = mY: X = mYx =mx : (v- n)y
Ε. Υ.Χ = Xy = nYx =ξy : ( - 4)x = nx : (v- n)y
IXY =mXY =mX : (v - n)y =mY : ( - 4)x
E' x.y =( - 4)xY = (v - n)yX =( - )Y : X = ( -n)Χ : nY
I' xy =mxy =mx : nY =my : EX
152 On the numerically
I now examine the modes of contradicting mXY and mX : nY.
As to the firſt, it is obvious that (m always meaning that m are,
but that more may be) either m or more Xs are Ys, or elſe
-m + 1 or more Xs are not Ys. The contradiction then is
either of the equivalents

( m + 1)X: ηnY and (n - m + 1 ) : X

It will be fatisfactory to evolve the contradiction of mX : nY

by a method which will again demonftrate the caſes in which no
contradiction exiſts ; or in which the propoſition is always true.
Let us put the two names in the leaſt favourable pofition for
making mX : nY true. Let p then be the number of Xs which
are not Ys, all the reſt being Ys. Take the p Xs which are not
Ys (p must not be ſo great as m, for then the propofition is
made good by the Xs which are not any Ys) and m-p from
thoſe which are Ys. All the mXs thus obtained are clear of
n-(m -p) or n - m +p Ys. Let this juſt ben : that is, let
p =m + n - n. Then -p, the number of Xs which are Ys, is
-(m + n - n) or + n - m - n. Let but one more X be Y,
and the propofition begins to be contradicted : for now m + n-
ท -1 Xs are not Ys, we must take up n + 1 - n ofthoſe which are
Ys to make m Xs, and there only remain η- ( n + 1 - n) or n - I
Ys clear of the m Xs. And it is plain that if we cannot do it by
uſing firſt all the Xs which are not Ys at all, ſtill leſs can it be
done by using thoſe which are. Accordingly the contradiction of
mX : nY is

( + n -m - n + 1)XY

Then, in order to have a propoſition which can be contra-

dicted, m + n muſt be greater than 4, or equal to + 1 at leaſt,
for otherwise + n - m - n + I would be greater than n, or more
Ys than n muſt be Xs, which is abſurd: and ſimilarly m + n
muſt be greater than n. Otherwise, all contradiction is abſurd,
or mX : nY is always true.
Aſſuming theſe laſt conditions, however, the contradiction of
mX : nY is made easier. To be capable of contradiction, it must
amount to ( m + n - n)X : nY . Now when m + n - n Xs are not
Ys, and no more, + n - m - n Xs are Ys. One or more
 definite Syllogifm. 153
above this, or let ( + n - m -n + 1 )XY, and mX : nY cannot be
true.

Thus much for contradictory or contrary propoſitions. I

ſhall preſently confider the contranominal propoſitions .
We must guard ourſelves from preſcribing the uſe of any premiſe
which neceſſarily belongs to all caſes in the univerſe (of propo-
ſitions) . Let P be a propofition which may or may not be true,
laid down as a premiſe, and Q a propoſition which is true in
every cafe. Let R be their neceſſary conſequence, or legitimate
inference : then it is not ' whenever P and Q are true, R is
true,' but ' whenever P is true, R is true.' So far as R is a con-
ſequence of Q, ſo far it is a conſequence of every thing which
neceſſarily gives Q ; and thus it is a conſequence of the ſuppoſed
conſtitution of the univerſe from which the propoſitions are
taken. Now this conſtitution is always underſtood ; it may be
a convenience that R ſhould be.deduced by firſt deducing Q, but
it cannot be a neceſſity. And R is a conſequence of P and this
conſtitution, not of P and Q.
For example, let the univerſe of propofitions be all that can
be formed out of the ſuppoſitions of the existence of 20 Xs, and
30 Ys, and 40 Zs, in one univerſe of names. Let us join to-
gether 15XY and 10Z : 20Y. Our rules of inference will pre-
ſently ſhow us, that 5X : 10Z is the neceſſary conſequence of
theſe premiſes : but this reſult is not only true when 15XY is
true, without anything elſe, but even without that ; becauſe
5+ 10 falls ſhort of 40.
Again, we must guard ourſelves from adopting the conclufion
which follows from premiſes, when that conclufion is true in all
cafes by the conſtitution of the univerſe : it is then a fort of
Spurious * conclufion, legitimate enough as an inference, but of a
perfectly distinct character from inferences which would bear

* To this word, as here uſed, I have heard much objection ; and when I first
took it, it was unwillingly, and for want of a better. But on further confi-
deration I am well ſatisfied with it. The objection ariſes from the idea of
falſe or worthleſs being generally attached to the word. But, though itmay
be uſual for ſpurious things to be worthless, it is not neceſſary. Ifa London
maker of razors ſhould put the name of a great Sheffield houſe upon them,
thoſe razors would be ſpurious. Suppoſe them as good as thoſe of the Shef-
field maker, or better, they are ſtill ſpurious : though it may be true enough
154 On the numerically
doubt but for the premiſes, or would bear contradiction under
other premiſes. Say that in the above univerſe we join the pro-
pofitions 15XY and 30Z : 20Y. Both theſe propofitions are
capable of contradiction : the ſecond is 20Z : ηΥ (η means 30,
but the ſymbol reminds the reader that 30 is all) or 10Y : Z
( being 40). Now, by laws of inference, 15XY + 30Z : 20Υ
yields 5X : 30Z, which is always true in that univerſe.
Here is a caſe in which premiſes capable of contradiction give
a conclufion which is not.
The rule of inference is obviously as follows. We cannot
ſhow that Xs are Zs by comparison of both with a third name,
unleſs we can affign a number of inſtances of that third name,
more than filled up by Xs and Zs : that is to ſay, ſuch that the
very leaſt number of Xs and Zs which it can contain are together
more in number than there are ſeparate places to put them in.
If our premiſes, for example, ſeparate ſome 30 Ys, and dictate
that among thoſe 30 Ys there must be 20 Xs and 15 Zs, it is
clear that there must be at least 5 Zs which are Xs. For if we
put down the 20 Xs which are to go in, and try to put the
Zs into feparate places, we are ſtopped as foon as we have filled
up the 10 remaining out of the 30 Ys, and must put the other
5 Zs among the Ys which have been made Xs. Accordingly,
ſo many Xs at least must be Zs as there are units in the number
by which the Xs and Zs to be placed, together exceed the num-
ber of places for them. All the other rules of inference are
modifications of this. For example, to prove that 10 Xs are not
Zs, we muſt ſhow ſome number of inſtances (be they Ys or ys,
or part one and part the other) overfull (in the above ſenſe) of
Xs and zs, to the amount of 10 at least ; ſo that 10 Xs are zs,
or are not Zs. To prove that ſome xs are ys, we muſt ſhow a
number of inſtances in which the least numbers of xs and zs

that the chances are rather in favour of their resembling the ware of Peter
Pindar's hero. In this work, a ſpurious inference is that which paſſes for
the confequence of certain premiſes, but does not in reality follow from
thoſe premiſes any more than from an infinity of others : being true by the
conſtitution of the universe. It is made to have the mark of thoſe premiſes,
when in truth we cannot know whether thoſe premiſes be poſſible or not,
until we have firſt examined a conſtitution which virtually contains our con-
clufion.
 definite Syllogifm. 155
which it can contain, overfill it, or in which the greatest number
of Xs and Zs which it can contain underfill it, or do not fill it,
though made completely ſeparate.
In examining the fundamental laws of fyllogiftic inference, it
is not neceſſary to confider any thing but the poſitive forms.
For mX : nY, when not ſpurious (and we ſhall ſee that the
ſpurious cafes may be rejected) is ( m + n - n)X : nY, which is
(m + n - n)Xy or ( m + n - 4)xY. There are, then, but two
fundamental caſes : one in which the predicates are the fame,
one in which they are contraries. We shall accordingly have to
confider
mXY + nZY and mXY + nZy :

and it will preſently appear that not more than one, even of
theſe, is abſolutely neceſſary. In each caſe we muſt aſk, what
collective inſtances of Y or of y, or partly of one and partly of
the other, receive any dictation as to how they are to be filled
with Xs, with xs, with Zs, or with zs : and what is the least
number of each which can be allowed to every ſuch collection.
But there is yet ſomething to do, ſuggeſted by the preceding
remarks. Let us take one propoſition, a type of all we ſhall
have to confider, ſay mXY . This means that XY is true to at
leaſt m inſtances. Now, this propoſition may involve Xy, or
xY, or xy. Firſt, as to Xy. To get the leaſt number of Xs
among the ys, we must put the greatest number among the Ys.
If all the Xs will go among the Ys (or if n be greater than or
equal to ) there need be no Xs among the ys : but if not (or if
η be leſs than 4) then ξη Xs must be among the ys, in every
cafe. Accordingly
mXY gives ( ξη)Xy
where by ξ- η underſtand o, not only when is equal to n, but
when it is leſs. This reſult is ſpurious, ſince it is true or falſe,
by the mere conſtitution of the univerſe, independently of mXY.
Secondly, as to xY. Since mXY is equally mYX, the fame
reaſoning ſhows that
mXY gives ( η - ξ)xY
where - is to be understood in the ſame way. This reſult is
alſo ſpurious for a like reaſon.
 156 On the numerically
Thirdly, as to xy. Since there muſt be m Xs among the Ys,
the greateſt poſſible number of xs is n-m . If this be as great
as v- , the whole number of xs, there need be no xs among
the ys : but if n - m be leſs than v- , there must then be at least
-- (n - m) xs among the ys, or v + m - ξ - n. Conſequently
mXY = (v + m - n- ) xy .
I here put the ſign = becauſe theſe propoſitions are really equi-
valents. Treat the ſecond in the ſame way as that which de-
duced it from the firſt, and we have
(v + m- n- 4)xy =( u + v + m - n-
=mXY

If v + m be not greater than η + ξ, the equivalent does not exiſt .

We are already well acquainted with one caſe of this propofition.
Let m= : then mXY is X)Y and the equivalent becomes
(υ- η)xy, which, as v- n is the whole number of ys, is y)x.
The rule is, if two names have a certain number of inſtances
at leaſt in common, to the whole number in the univerſe add
that number of inſtances, and fee if the ſum exceed the whole
number of inſtances of both names together. If it do ſo, the
exceſs ſhows the leaſt number of inſtances which the contraries
of theſe two names must have in common. Follow this rule,
and we have
mXY =(v + m - n- 4) xy
mxY = ( + m - n)Xy .
mXy = (n + m - 4)xY
mxy = ( + n + m - v)XY
Here are exhibited the equivalent contranominal forms. The
following reſults may now be deduced.
First, theſe contranominals being formed in the ſame way,
each from the other, in any one pair, whatever we prove of the
firſt from the ſecond, we alſo prove of the ſecond from the firſt.
.
The mathematician would call them conjugate pairs. Next, ſince
all the four pairs are but verſions of the firſt, with difference of
names, whatever we prove univerſally of the firſt pair, we prove
of all. Now, taking the firſt of any pair and making it poſſible,
which is done by allowing m not to exceed the number of either
of the names mentioned, the ſecond may be poſſible or impoſſi-
 definite Syllogifm. 157
ble, according as the ſubtraction indicated can be done or not.
But whenever the ſecond is impoſſible, the firſt is fpurious. Take
mXY, and let (v + m -- n)xy be impoſſible, or v + m (and ſtill
more v) leſs than + n . Now as all the Xs and η Ys muſt find
place in the vinſtances of the univerſe, and ξ + η exceeds v, we
muſt, in every caſe of the univerſe of propoſitions, have at least
(ξ + η - υ) ΧΥ. But v + m is less than + η or 4 + n - v greater
than m : conſequently, mXY is ſpurious, a larger propofition
being always true.
As we are not to admit ſpurious propofitions among our pre-
miſes, we had better write all premiſes double, putting down each
of the forms, and making double forms of inference. The pre-
fence of the ſymbols of all neceſſary fubtractions will remind the
reader of the ſuppoſitions which must be made, to inſure a legiti-
mate fyllogifm. I now take the ſeveral forms.
m XY n ZY (m + n - n)XZ
(v + m- ξ-n) xу + (v + n- ζ-n) zy =(v + m + n -n-- )xz
The law of inference here tells us (page 154,) that m + n being
greater than n, (m + n - n)XZ, be it ſpurious or not, follows from
the upper premiſes. The lower premiſes alſo give their inference
if
(v + m - ξη) + ( v + n - ζ -n) be greater than v- η
v- n being the number of the ys. This laſt is equivalent to ſay-
ing that v + m + n is greater than + η + ζ. Firſt, remark that
one ſpurious premiſe neceſſarily gives a ſpurious conclufion. Say
that v + m is leſs than + η, or that mXY is ſpurious. Then,
ſince v + m is leſs than + n, and n does not exceed 4, it follows
that v + m + n is leſs than + η + ζ ; whence the contranominal
ofthe conclufion does not exiſt, or the concluſion is ſpurious, as
aſſerted .
Next, obſerve that the concluſion may be ſpurious, though
neither of the premiſes be ſo. For though v + m be greater than
ξ + n, and v + n than ζ + n, and therefore 2v + m + n greater than
ξ + ζ + 2n, or v + m + n + (vn) greater than η ++ , it by no
means follows that v + m + n alone is greater than n ++ ζ.. It
is alſo viſible in the mode of formation of the ſecond inference,
that to ſay v + m exceeds + n, and v + n exceeds ( + n, only gives
158 On the numerically
existence to the premiſes : to give them conclufion, the ſum of
the two exceſſes muſt itſelf exceed v-n, the whole number
of ys.
Thirdly, we must not omit to examine the poſſible caſe in
which a premiſe is partially ſpurious. For example, there are 10
Xs and 20 Ys in a univerſe of 25 inſtances ; accordingly, 10+
20-25, or 5, of the Xs must be Ys. Let one of the premiſes be
8XY : this is not then all contingent, and capable of contra-
diction ; we only learn ſomething about 3 out of the 8 Xs. And
I call this propofition partially ſpurious. But it will give no
trouble : for we must deal with the premiſes and their contra-
nominal equivalents before we can pronounce for a conclufion ;
and of two propoſitions which are contranominal equivalents of
each other, one must be partially ſpurious. To ſhow this, obſerve
that if mXY be not partially ſpurious, it is becauſe v is greater
than ξ + η ; or 2v than ξ + n + v ; or (υ- ξ) + (υ - η) than v . But
then the numbers of xs and ys together exceed the whole num-
ber of inſtances in the univerſe ; whence ſome xs must be ys, or
the contranominal equivalent of mXY is partially ſpurious.
Now, to write down the various forms of inference. There
are fixteen ways oftrying for an inference : we may combine a
propofition in XY, or xy, or XY, or Xy, with one in XZ, or xz,
or xZ, or Xz . But theſe ſixteen caſes really combine four and
four into only four distinct caſes. Thus the one we have been
confidering, really contains the combinations of XY and YZ,
XY and yz, xy and YZ, and xy and yz. It is in our power to
make either pair the principal pair, and to give the other pair as
contranominals of the firſt pair.
Thus, we may write the case of inference we have been con-
fidering, as in the firſt of the following lift, the others being ob-
tained from the firſt, by changing X into x, or Z into z, or both.
The fign + placed in the middle implies the coexistence of the
four propoſitions : and independent numeral letters are introduced
as ſeen, which will preſently be connected with the others by
equations, inſtead of being expreſſed in terms of them.
mXY nYZ = JpXZ The equations preſently given for
+
m'xy n'yz p'xz this caſe apply with certain changes
to the other cafes .
 definite Syllogifm. 159
mxY nYZ) SpxZ Here X and x are made to change
m'Xy + n'yz = pXz their former places : in the equa-
tions, & and ' muſt change places.
mXY JpXz Here Z and z change places : as
m'xy
+nYzl
+nyZJ
=

p'xZ muſt & and g' in the equations.

mxYnYz = pxz Here X and x, and alſo Z and z,
m'Xy+nyZJ} (p'XZ change places ; as muſt and ' ,
and & and g', in the equations.

In the new manner of writing the form we have already con-

fidered, being the firſt of the four, we have juſt written
m' for v + m - ξη p for m + n - ท
n' for v + η ζη •
p'for v + m + n - η - ξ - ζ

Let us write ", η', ξ' , for υ - ξ, υ - η, υ- ζ, the numbers of xs,

ys, and zs : and then, + ξ , η + n', 5 + 5', being all the fame, (for
each is v) we may write η- ξ' for ', ' - .for ' -5, and ſo on.
That is, in the difference of two, one of which is accented, we
may interchange the letters if we pleaſe. The equations of con-
nection for the firſt or ſtandard caſe, are then

m' =m + ' - n = m + n' - ξ m =m' + ξη' =m' + η - ξ'

n' = n + ζ' η =n + η - ζ n =n' + ζ -η' =n' + η - ξ'

p =m + n - n = m' + n' + n- ' - ' ] = m + n' - '

or m' + n' + ξ - -

=n + m'- '
or m' + n' + ζξ' η'

p' =m' + n' - n' = m + n -٤] = m' + n –ζ

or m + n +
+ -

= n' + m -

or m + n + ζ' - ξη

For the ſecond caſe we must write m' =m + n = m + η'- ' ,

and ſo on. I now proceed to the ſeveral diviſions into which
our uſual modes of thinking make it convenient to ſeparate the
caſes of this moſt general form.
 160 On the numerically
Firſt, when every thing is numerically definite. In this caſe,
as ſeen, every form requires an examination of the premiſes and
conclufion, as to whether they are or are not ſpurious.
Secondly, when v, the number of inſtances in the whole uni-
verſe of names, is wholly unknown. In this caſe ¿' is indefinite
when is definite, and vice verſâ ; and ſimilarly one at least of
each two, η or η', & or ζ' , is indefinite. There are then no ſpu-
rious conclufions ; or, which is the ſame thing, none which are
known to be ſuch : for the ſpuriouſneſs of a premiſe or conclu-
fion confifts in our knowing that it must be true of its two terms,
independently of all comparison of thoſe terms with a third.
Thirdly, when ξ, η, 4, are all indefinite, as well as v. In this
caſe, as here ſtated, there is no poſſibility of inference. We can-
not tell whether m + n be or be not greater than n, if we do not
know what η is, in any manner, or to any extent.
But here we introduce that degree of definiteneſs by which
we diftinguiſh the universal from the particular (or poſſible parti-
cular, ſee page 56) propoſition. If we can know that either of
the two, m and n, is the fame as n (greater neither can be) then
we know that m + n is greater than n. And at the fame time
we make Y univerſal, in one or the other of the premiſes. And
the fame if we can know that either m' or n' is n' .
The following are the forms which may all be derived from
the firſt, by using all the varieties of contrary names and contra-
nominal equivalents. If we want, for instance, to ſhow the con-
nection of the fourteenth with the firſt, we throw the firſt into
the form

(m + n' - )xy + nYZ = ( m + n + n' --4)xz

We then change x into X, and Z into z, changing at the
ſame time into ' and into ' : and thus we get
(m + n' - ')Xy + nYz = (m + n + n' -'- ')XZ
Now, for m + n'- ' write m', that is, for m write m' - n' + '
and we have
m'Xy + nYz = (m' + n - ')XZ .
which is one of the forms of the fourteenth. And (n + m' - )xz
is only the contranominal of (m' + n - ' )XZ.
 definite Syllogifm. 161

1. mXY + nYZ = (m + n - n)XZ = (m + n + n' - ξ - 4)xz

2. m'xy + nYZ = (n + m' - ') XZ =(m' + n - 2)xz
3. mXY + n'yz = (m + n' - ') XZ = (n' + m - 4)xz
4. m'xy + n'yz =( m' + n' + n -'- ')XZ =(m' + n' - n')xz
5. mxY + nYZ =( m + n - n)xZ = (m + n + n' -'- 5)Xz
6. m'Xy + nYZ =(n + m' − ¿)xZ =(m' + n - ;)Xz
7. mxY + n'yz =( m + n' - ')xZ =(n' + m - ' )Xz
8. m'Xy + n'yz = (m' + n' + η - " - )xZ = (m' + n' - ')Xz
9. mXY + nYz = ( m + n - n)Xz = (m + n + n' -- ' )xZ
10. m'xy + nYz = (n + m' - ' )Xz =(m' + n - z')xZ
11. mXY + n'yZ = (m + n' - ; )Xz = (n' + m− )xZ
12. m'xy + n'yZ = (m' + n' + η - ζ - ' )Xz = (m' + n' - ')xZ
13. mxY + nYz =( m + n - n)xz =(m + n + n' -'- ' )XZ
14. m'Xy + nYz = ( n + m' - )xz = (m' + n - ')XZ
15. mxY + n'yZ =(m + n' - )xz = (n' + m − ¿')XZ
16. m'Xy + n'yZ=(m' + n' + η - ζ- )xz = (m' + n' - n')XZ

The fyllogifms of chapter V are all particular caſes of the

above lift, obtained as follows : -

1. m =η Α'ΙΙ 9. m = ηn [A'OO
m =η, η =ζ A'A'A ' m =η , η = ζ' Α'Ε'Ε'
η =η IAI η =η IEO
n =n , m =ξ AAA n =n , m = ξ AEE
m =n , n =η A'AI m =n , n = Α'ΕΟ
m = ξ , η =ζ AA'I' m =ξη =ξ' ΑΕ'Ο'
2. m'= ‫'اع‬ Α'ΙΙ 10.m'= ' Α'ΟΟ
η ζ I'A'I' n =ξ' Ι'Ε'Ο '
m' = ', η = ζ A'A'A ' m' = ', n =
= ζ' Α'Ε'Ε '
3. m = ξ AI'I' II . m = ΑΟ'Ο'
n' =ξ' LAI n' =ζ LEO
m = =ξ , n' =ξ' AAA m = , n' = ζ AEE
4. m' =η AI'I' 12:m'n' AO'O '
m =η , n' =ξ' AAA m' =n', n'' =ζ AEE
1
n' =n' I'A'I' n' =η' Ι'Ε'O '
n' =n', m' = ' A'A'A ' n' =n', m' = ' Α'Ε'Ε'
m'= ', n' =n' AA'I' m' =n', n' =n' ΑΕ'Ο '
m' = ξ'n' =ξ' A'AL m' = ' n'' = ζ Α'ΕΟ
M
162 On the numerically
5. m = η ΕΙΟ' 13. m = n EOI'
m = η , η =ζ EA'E m = η , η = ζ' EE'A
η =η O'AO ' η =η Ο'ΕΙ'
n =n , m = ξ' E'AE' n =n m= Ε'ΕΙΑ'
m =η , η = η EAO ' m =η η =η ΕΕΙ'
‫اع‬ ‫اع‬
m = ', n = ζ Ε'Α'Ο m = ξ' η = ξ' Ε'ΕΙ,
6. m'= ΕΙΟ' 14. m' = EOI'
η ζ ΟΑ'Ο η =ζ' ΟΕΙ
m' = ξ , η = ζ EA'E m' = ξ, η = ξ' ΕΕ'Α
7. m = ΕΙΟ, 15. m = Ε'ΟΙ,
n' = ξ' O'AO' n' = Ο'ΕΙ'
m = ξ', n' = ξ' E'AE' m =ξ , η =ζ Ε'ΕΑ'
8. m' = η' ΕΙΟ, 16. m'n' Ε'ΟΙ
m' =n' , n' =ξ' E'AE' m η', η' = ζ Ε'ΕΑ '
n' =n' OLA'O n' =n' ΟΕΙ
n' =n', m' = ξ ELA'E n' =n' m
m' = EE'A
m =n', n' =η' E'A'O η,, η =η
m =n Ε'ΕΙ
m = ξη' =ξ' EAO' m' =ξη' =ζ EEI'

We have thus another mode of eſtabliſhing the completeneſs

of the ſyſtem of fyllogifm, laid down in the last chapter : that is,
of the ſyſtem in which there is only the common univerſal and
particular quantity. Theſe ſyllogiſms of numerical quantity, in
which conditions of inference belonging to every imaginable caſe
are repreſented by the general forms which numerical ſymbols
take in algebra, muſt of neceſſity be the moſt general of their
kind. And examination makes it clear that, except the preced-
ing, there can be no ſyllogiſm exiſting between X, Y, Z, and their
contraries. Many fubordinate laws of connexion might be no-
ticed between the general forms and their particular caſes. Thus,
each univerſal occurs three times, each fundamental particular
twice, and each ſtrengthened particular twice. The firſt form
in pages 158, 159, gives only affirmative, the fourth only negative,
premiſes : the ſecond and third one ofeach kind, commencing with
a negative in the ſecond, and with an affirmative in the third.
There are two remarkable ſpecies of fyllogifm (or rather,
which ought to have been remarkable) : which I ſhall now pro-
ceed to notice.
The distinction of larger and ſmaller part, when diviſion into
 definite Syllogifm. 163
two parts is made, is as much received into the common idiom
of language as the distinction of whole and part itſelf. Most of
the Xs are Ys,' is nearly as common as ' All the Xs are Ys :'
though ' feweſt of the Xs are Ys,' is only ſeen as ' most of the
Xs are not Ys.' The ſyllogiſms which can be made legitimate
by the use of this language will do equally well for any fraction,
provided we couple with it the fraction complemental to unity
(which in the caſe of one half is one half itſelf). Let a and β
ſtand for two fractions which have unity for their ſum, as and
Let XY and X :Y indicate that leſs than the fraction a
α

of the Xs are or are not Ys. Let " XY and " X : Y indicate
that more than the fraction a of the Xs are or are not Ys .
Then the following fyllogifms ariſe from the caſes with the
numbers prefixed.
1. YX + PYX = XZ 9. «YX + Y : Z = X : Z
4. ay : X + By : Z =xz 12. ay: X + ByZ =Z : X
5. Y : X + BYZ = Z : X 13 . Y : X + Y : Z = xz
8. ayX +By: Z = X : Z 16. ayX +ByZ = XZ
It will be ſeen that here are but three really distinct forms ; of
which the ſimpleſt examples are as follows,
Moſt Ys are Xs ; Moſt Ys are Zs ; therefore ſome Xs are Zs .
Moſt Ys are Xs ; Moſt Ys are not Zs ; therefore ſome Xs are
not Zs .
Moſt Ys are not Xs ; Moſt Ys are not Zs ; therefore ſome
things are neither Zs nor Xs.
It is hardly neceſſary to obſerve that in one of the premiſes
' more than' may be reduced to ' as much as :' but not in both.
Thus, if two-ſevenths exactly of the Ys be Xs, and more than
five-ſevenths of the Ys be Zs, it follows that ſome Xs are Zs .
The above fyllogifms admit a change of premiſe, as follows :
If we say that more than ths of the Ys are Xs, we thereby fay
that leſs than ths of the Ys are xs : or "YX and BY: X are the
ſame propofitions. Thus, 'moſt are' is equivalent to ' a minority
(none included) are not.' Hence we have
BY : X + Y : Z = XZ
β

and ſo on . Or we may combine the two forms, as in

«YX + Y : Z = XZ
164 On the numerically
The above are the only ſyllogiſms in which indefinite particu-
lars give conclufions, by reaſon of that approach to definiteneſs
which confifts in deſcribing what fractions of the middle term are
fpoken of, at least, or at moſt. But they are not the only fyllo-
gifms ofthe fame general ſpecies. In every caſe inference follows
when there is a certain preponderance ; and the largeneſs of the
inference depends upon the extent ofthat preponderance. Thus
in (12) there is an Xz inference when m' + n' + n exceeds ζ + ' :
ſo many units as there are in this exceſs, ſo many Xs (at leaſt)
are zs .
Now in every caſe, a pair of univerſal premiſes give in-
ference : and in every caſe there muſt be a degree of approach
to univerſality at which inference begins. The ordinary fyllo-
giſms, I ſuſpect, are, and are meant to be, not ſuch as ' Every
X is Y, every Y is Z, therefore every X is Z,' but ' generally
ſpeaking X is Y, and generally ſpeaking Y is Z, therefore gene-
rally ſpeaking X is Z.' And by ' generally ſpeaking' is meant
the affertion that an enormous majority of inſtances make the
aſſertion true. A fyllogifm of this fort is the oppoſite of the à
fortiori fyllogifm ; and might be ſaid to be true ab infirmiori. If
we have X)Y with p exceptions, and Y)Z with q exceptions ;
then, in form ( 1. ) we have m = -p ,n = n - q, and m + n - n =
E-p- q. As long, then, as the number of exceptions altogether
fall ſhort of the number of Xs, there is inference : if the total
number of exceptions be very ſmall, compared with the number
ofXs, there is the ' generally ſpeaking' kind of inference. Ex-
amine all the univerſal caſes, and it will be found that the fame
law prevails ; namely, that there is inference when the numbers
of exceptional inſtances in both premiſes together do not amount
to the number of inſtances in the univerſal term of the conclu-
fion ; and that there is exceptional univerſality (as we may call it)
in the conclufion, whenever the whole amount of exception is
very ſmall,
This compared
leads with Ithat
us to what willnumber of inſtances
call the theory of . exceptional
particular fyllogifms. We have ſeen that the eight complex
affirmatory fyllogifms, which are all à fortiori in their conclu-
fions, afford each two particular fyllogifms. We have denoted
coexistence by + ; and the coexistence of two propoſitions gives
more than either. Let us denote exceptive coexiſtence by -:
thus, P-Q means that the propoſition P is true except in the
 definite Syllogifm. 165
inſtances contained in Q. Thus, X)Y-X : Y means that every
X (with ſome exceptions) is Y. This is, of courſe, A -O1, and
only differs from I in the mode of expreffion not being ' fome
more than none at all' but ' ſome leſs than all. In the expref-
fion
(Α- )( - )( - )

we have the ſymbol of the ab infirmiori ſyllogiſm ſtated above,

ſubject to the poſſibility of nonexistence if the number of excep-
tions in the two premiſes ſhould exceed the number of inſtances
in the univerſal term of the conclufion. If we look at AO , as
a ſymbol deſcriptive of premiſes, we fee one of the inconclufive
forms ; that is, a form from which we cannot draw an inference.
But this is only becauſe our inferences are all poſitive, and imply
aſſertion offufficiency in the premiſes. There is no uſe (except
to ſhow the manner in which the parts of a ſyſtem hang together)
in declarations of infufficiency : for we know that all collections
of premiſes, whatever they may be ſufficient for, will be infuf-
ficient for an infinite number of different things. And it is
important to remember that while ſufficiency is accompaniedby
must be, inſufficiency only allows may be. From A.A.the con-
cluſion A must be true : from AO (and as far as theſe are con-
cerned) it may be falſe. Accordingly AOO and OAO may
ſerve to expreſs the two defects of (A- 1)(A - )( - )
from AAA , exiſting in the ab infirmiori fyllogifm, and poffibly
preventing conclufion altogether : juſt as A.O'O' and O'A.O'
ſhow the additional conditions by the fulfilment of which AAA
is elevated into the àfortiori fyllogifm D.D.D. It is worth
while to dwell upon the varieties of this cafe. The ab infirmiori
ſyllogiſms of the ſtrengthened particulars were previously confi-
dered.
In all the cafes yet treated, we have had, more or less, the
power of giving inſtances in common language, without recourſe
to numerical relation expreſſed in unusual terms. This of courſe,
is always the cafe in the fyllogifms of chapter V.; and we
have given one common inſtance (though never met with in books
of logic) from each ſet of ab infirmiori fyllogifms. But there are
ſtill caſes of the ſame ſort to be confidered. Though in our de-
finite relation (page 56) of all, we usually (in books of logic at
166 On the numerically
leaſt) make the relation exiſt, for each propoſition, between the
terms of the propoſition itſelf, yet it may be aſked whether we
cannot ſometimes infer ſuch a ſpecies of univerſal as this, ' for
every Z there is an X which is Y ;' Z being one of the names
of the ſecond premiſe. If we examine the firſt two caſes, which
will be guide enough, we shall find the following reſults from the
new ſuppoſitions now made.
1. m = ζ, n = 4, gives ¿XY + YZ = n'xz : or if for every Z
there be an X which is Y, and for every X a Z which is Y,
then, ſo many ys as there are, ſo many things which are neither
X nor Z. This ſyllogiſm has little new meaning, and no new
application : it requires =ζ, and therefore X)Y and Z)Y.
2. m' = ', gives ( xy + n YZ = nxz , or if for every Z there be
that which is neither X nor Y, and if ſome Ys be Zs, there are
as many inſtances which are neither X nor Z. This is a new and
effective form .
2. n = ' , gives m'xy + 'YZ =m'XZ, a new form.
Theſe two caſes will be preſently further confidered. Now,
obſerve that if m + n in the firſt form, or m' + n in the ſecond,
be v, that is, if the pair m and n be & and ξ' , or η and η', or gand
', we have inference of the kind required. The firſt form gives
no new fyllogifm: ſince v is more than n, Ys which are Xs, and Ys
which are Zs, to the number of v, give the form (1.) by the main
law of inference (page 154). In the ſecond form, if m' + n = v,
we diſtribute among the Ys and ys, Zs and xs to the full number
of both, ſo that wherever there are not xs (that is, wherever
there are Xs) there are Zs : or X)Z as obtained from the form .
But every way of conſtructing m'xy + nYZ =(m' + n − ¿')XZ
which gives m' + n = v, is only a caſe ofAAA . For m' cannot
exceed n' , and n cannot exceed n : and m' + n being vor η + n' ,
we must have m' = n' and u = n ; whence the aſſertion made. The
forms we are now in ſearch of, ſo far as quite new, are all con-
tained in the two new ones above noted ; and of theſe, the ſecond
is but a transformation of the firſt. The eight varieties derived
from uſe of contraries, or from the forms in page 161 , beginning
with the ſimpleft, are
XY + nZ : nY = nX : ζZ ' XY + nyz = nXZ
X : nY + nZY = nX : Z ζ'Χ : ηΥ + nY : Z = nXZ
 definite Syllogifm. 167
Y : X + nZ : nY = nxz 'Y: X + nyz =nZ : X
ζxy + nZY = nxZ + nY : Z = nZ : X

Theſe are fyllogifms, which exhibit a curious kind of antago-

niſm to the particular fyllogifms. Take the ſyllogiſm A.O'O', the
terms being M,Y, Z ; we have then M)Y + Z : Y = Z : M . Of
courſe the conclufion M : Z is not legitimate from theſe premiſes
alone : but if M have as many inſtances as Z, then M : Z is
legitimate. For if Ms, as many as there are Zs, be among the
Ys, and ſome of the Zs be not among the Ys, though all the
reſt were, there would not be enough to match all the Ms, or
ſome Ms are not Zs. Now, let M be a name given to an X
which is Y, and let ſuch Xs have as many inftances as Z, and
the above becomes the firſt of the ſyllogiſms in the last liſt.
Thus, O'O is legitimate, if the quantity of the ſubject men-
tioned in I be taken from the Zs. The ſecond ſyllogiſm is ENO',
altered into OIO in the fame manner.
The reader may find all the reſults of the above caſe in the
following rule, in which it is understood that all the ſuper-propo-
ſitions are to be written either way : thus, A' is written x)y,
or Y)X, and O' is nx : ny, or nY: X (page 62 ). Write down
any pair of particulars, followed by I if the pair be of the ſame
ſign, and O if the pair be of different figns : as in OOI or IOO .
Accent the pair in contradiction to either the direct rule (page
62) as far as the words affirmative and negative are concerned :
that is, let a negative beginning iſolate nothing, and an affir-
mative beginning iſolate the middle propofition : or elſe, ac-
cent the pair according to the inverſe rule. Thus, OI
and O'O'I' contradict the direct rule, and O'O'I and OO '
preſerve the inverſe rule. To make theſe fyllogifms good (in the
particular way in queſtion) proceed thus :-When the direct rule
is contradicted, take the quantity of the first concluding term
from the total of theſecond, if the ſecond premiſe be affirmative,
and from its contrary, if negative. When the inverſe rule is
preſerved, take the quantity of the second from the total of
the first . Thus, in O'O'I' the direct rule is contradicted :
and it ſtands m'x : n'y + n'y : ' z =p'xz. The ſecond premiſe is
negative, the total of its predicate g' inſtances, that of the con-
trary 4. Accordingly, (x : n'y + n'y: 'z=n'xz, or Y: X +
168 On the numerically
n'Z : nY = n'xz, which is one ofthe forms already obtained. Again,
O'O'L preſerves the inverſe rule, and is m'x: n'y + nZ : nY =
pXZ. The total of the firſt term is ' inſtances. Hence,
m'x: n' + ' Ζ : ηY = m'XZ , orm'Y : EX + ¿' Z : yY = m'XZ, which
is derived from one of the forms given, by interchanging X
and Z.
This claſs of fyllogifms with transpofed quantity naturally leads
to the queſtion, Is it uſed ? Do ſuch fyllogifms occur in ordi-
nary or in literary life ? If not, there is no reason for ſelecting
them from the infinite number of caſes which the numeri-
cally definite ſyſtem affords. To try this, ſuppoſe a perſon, on
reviewing his purchaſes for the day, finds, by his countercheques,
that he has certainly drawn as many cheques on his banker (and
may be more) as he has made purchaſes. But he knows that he
paid fome of his purchaſes in money, or otherwise than by
cheques. He infers then that he has drawn cheques for fome-
thing elſe except that day's purchaſes. He infers rightly enough ;
but his inference cannot be reduced to a common fyllogifm, with
the names in queſtion for terms. It is really a fyllogifm oftranf-
poſed quantity, as follows :-
For every ' memorandum of a purchaſe' a ' countercheque '
is a ' tranſaction involving the drawing of a cheque. '
Some ' purchaſes ' are not ' tranſactions involving, &c.'
Therefore ſome ' countercheques ' are not ' memoranda of pur-
chaſes .'
It may be worth while to give one inſtance of the verification
of the contradictory form. By page 152 it appears that the con-
tradiction of mXY is ( m + 1 )Xy, or (y - m + 1)xY, and that
of m'Xy is ( -m' + 1 )XY, or (η' -m' + 1)xy.
To mXY join the contrary of ( m + n - y)XZ, or ( + y - m
-n + 1 )Xz : we have then
mYX + ( + y - m - n + 1 )zX ;
the inference of which is (m ++ y - m - n + 1- )Yz, that is,
(y -n + 1 )Yz, the contrary of nYZ .
Returning to the forms in page 161, it will be obſerved that we
have no double inferences. In every caſe we have made uſe of
one form of inference : if u be known, the other is a real equi-
valent ; or elſe it is impoſſible, and as we have ſeen, then the
 definite Syllogifm. 169
firſt is ſpurious. If u be not known, then the ſecond is either
perfectly indefinite, or elſe identical with the one choſen. Ex-
amination will ſhow that in every one of the caſes cited in page
161 , the neglected form of inference is only ſaved from perfect
indefiniteneſs when we are able to apply the word all to one orother
of the terms : the number being as indefinite as before ; the rela-
tion thus obtained being definite. Take the firſt form, and make
n= n ; by the firſt inference we then get the fyllogifm LAI : by
the ſecond, we get (m + v -- ;)xz, indefinite both in number
and relation. We do not know what v, ξ, and ζare. If we
knew as much as that m + v is leſs than + ζ, we ſhould know
our inference to be ſpurious,* it being not the leſs an inference.
Now, add the condition m = : the firſt inference gives the fyl-
logiſm AAA , the ſecond inference now becomes (2-4)xz :
definite relation enters, and we have z)x, or X)Z, or Ai, as
before. And the ſame of the other forms .
The reader may perhaps ſuppoſe that I ought to have com-
menced this chapter with the complex numerical fyllogifm, in
imitation of the method which I followed in treating the ordinary
fyllogifm. But in truth there is no ſyſtem of complex fyllogifm
of perfect numerical definiteneſs both in premiſes and conclufion,
To ſhow this, let m,XY with the comma, mean that there are
exactly m Xs which are Ys, neither more nor fewer. Accord-
ingly m,XY is a ſynonyme for mXY + (y - m)xY. Now com-
bine m,XY and n,ZY, or
(mXY + n - mxY)(nZY + n - nzY)
We then have mXY + nZY =(m + n - n)XZ
(n -m)xY + (n - n)zY =(n - m - n)xz
mXY + (n - n)zY =(m - n)Xz
(n -m)xY + nZY =(n -m)xZ
* I must again remind the reader, of the diſtinction between ſpurious and
illegitimate, which exiſts in my language. The ſpurious inference follows
from the premiſes, and is perfectly good and true : but from the conſtitution
of the univerſe, it will always be true, whatever premiſes in that univerſe are
taken. The illegitimate inference is that which does not follow from the
premiſes. A conclufion not known to be ſpurious, that is, there not being
the means of knowledge, is not ſpurious : but an illegitimate conclufion can-
not be made legitimate, that is, following from the premiſes, by any further
knowledge.
170 On the numerically definite Syllogifm.
Two only of theſe have meaning : let them be the two upper
ones. We can affign then Z or z to ( m + n - y) + (m -n), or
to 2m- y of the Xs. But there are not all of the Xs here : for
m is less than y, and than 4, whence 2m is leſs than + ξ, or
2m- η leſs than §. The rest of the Xs, + y- 2m in number,
may, for aught theſe premiſes declare, be either Zs or zs.

CHAPTER IX .

On Probability .

HE moſt difficult inquiry which any one can propoſe to

T himſelf is to find out what any thing is : in all probability
we do not know what we are talking about when we aſk ſuch a
queſtion. The philofophers of the middle ages were much con-
cerned with the is, or effence, of things : they argued to their own
minds, with great juſtice, that if they could only find out what a
thing is, they ſhould find out all about it : they tried, and failed.
Their ſucceſſors, taking warning by their example, have inverted
the propoſition ; and have fatisfied themſelves that the only way
of finding what a thing is, lies in finding what we can about it ;
that modes ofrelation and connexion are all we can know of the
eſſence of any thing ; in ſhort, that the proverb 'tell me who you
are with, and I will tell you what you are,' applies as much to
the nature of things as to the characters of men. We are apt
to think that we know more of the eſſence of objects than of
ideas ; or rather, of ideas which have an objective ſource, than
of thoſe which are the conſequence of the mind's action upon
them. I doubt whether the reverſe be not the cafe : at any rate,
when we content ourſelves with inquiry into properties and rela-
tions, we have certain knowledge upon our moſt abſtract ideas.
The object of this chapter is the confideration of the degrees of
knowledge itſelf. That which we know, of which we are cer-
tain, of which we are well aſſured nothing could perfuade us to
the contrary, is the existence of our own minds, thoughts, and
perceptions, the two laſt when actually preſent. This highest
knowledge, this abſolute certainty, admits of no imagination of
the poſſibility of falſehood. We cannot, by ſtopping to confider,
 On Probability . 171
make ourſelves more ſure than we are already, that we exiſt,
think, ſee, &c. Next to this, come the things of which we can-
not but ſay at last we are as certain of them as of our own exif-
tence ; but of which, nevertheless, we are obliged to ſay that we
arrive at them by proceſs, by reflection. Theſe we call neceffary
truths (page 33). The necessity of admitting theſe things cauſes
ſome to imagine that they are merely identities, that they amount
to ſaying that when a thing is, it is : but this is not correct. TΤο
ſay that two and two make four (which must be), and that a
certain man wears a black coat (when he does ſo) both involve
the pure identity that whatever is, is ; and not one more than
the other. Nor is two and two identically four, though necef-
farily fo . Our definitions of number ariſe in the proceſs of fim-
ple counting. Throw a pebble into a baſket, and we ſay one :
throw in another, and we say two ; yet one more, and we fay
three, and ſo on. The full definitions of the ſucceſſive numbers
are ſeen in

1
(I + I) {(1 + 1) +1 } [ { ( 1 + 1 ) + 1} + 1 ], &c.

That three and one are four is definition : it is our pleaſure to

give the name four to 3+ 1. But that 3 + 1 is 2+ 2 is neither
definition nor pure identity. It is not even true that ' two and
two' is four ; that
[{(1 + 1 ) +1} +1 ] is ( I + I) + ( 1 + 1)
It is true, no doubt, that ' two and two' is four, in amount,
value, &c. but not in form, conſtruction, definition, &c.
There is no further uſe in drawing distinction between the
knowledge which we have of our own exiſtence, and that of two
and two amounting to four. This abſolute and inaſſailable feel-
ing we shall call certainty. We have lower grades of knowledge,
which we uſually call degrees ofbelief, but they are really degrees
of knowledge. A man knows at this moment that two and two
make four : did he know it yesterday ? He feels perfectly certain
that he knew it yesterday. But he may have been ſeized with a
fit yesterday, which kept him in unconsciouſneſs all day : and
thoſe about him may have been warned by the medical man not
to give him the least hint of what has taken place. He could
ſwear, as oaths are usually understood, that it was not ſo : if he
 172 On Probability.
could not ſwear to this, no man could ſwear to anything except
neceſſary truths. But he could not regard the aſſertion that it
was not ſo, as incapable of contradiction : he knows it well, but,
as long as it may poſſibly be contradicted, he cannot but ſay that
he might know it better.
It may ſeem a ſtrange thing to treat knowledge as a magnitude,
in the fame manner as length, or weight, or furface. This is
what all writers do who treat of probability, and what all their
readers have done, long before they ever ſaw abook on the ſubject .
But it is not cuſtomary to make the ſtatement ſo openly as I
now do : and I confider that ſome juſtification of it is neceſſary.
By degree of probability we really mean, or ought to mean,
degree of belief. It is true that we may, if we like, divide pro-
bability into ideal and objective, and that we must do ſo, in order
to repreſent common language. It is perfectly correct to ſay ' It
is much more likely than not, whether you know it or not, that
rain will foon follow the fall of the barometer.' We mean that
rain does foon follow much more often than not, and that there
do exiſt the means of arriving at this knowledge. The thing is ſo,
every one will ſay, and can be known. It is not remembered,
perhaps, that there is an ideal probability, a pure ſtate of the mind,
involved in this aſſertion: namely, that the things which have been
are correct repreſentatives of the things which are to be. That
up to this 21st ofJune, 1847, the above ſtatement has been true,
ever fince the barometer was uſed as a weather-glaſs, is not de-
nied by any who have examined it : that the connexion of
natural phenomena will, for ſome time to come, be what it has
been, cannot be ſettled by examination : we all have ſtrong rea-
ſon to believe it, but our knowledge is ideal, as diftinguiſhed
from objective. And it will be found that, frame what circum-
ſtances we may, we cannot invent a caſe of purely objective pro-
bability. I put ten white balls and ten black ones into an urn,
and lock the door of the room . I may feel well aſſured that,
when I unlock the room again, and draw a ball, I am justified
in ſaying it is an even chance that it will be a white one. Ifall
the metaphyſicians who ever wrote on probability were to witneſs
the trial, they would, each in his own ſenſe and manner, hold me
right in my aſſertion. But how many things there are to be
taken for granted ! Do my eyes ſtill distinguiſh colours as be-
 On Probability. 173
fore ? Some perſons never do, and eyes alter with age. Has
the black paint melted, and blackened the white balls ? Has any
one elſe poſſeſſed a key of the room, or got in at the window,
and changed the balls ? We may be very fure, as thoſe words
are commonly used, that none of theſe things have happened, and
it may turn out (and I have no doubt will do ſo, if the reader try
the circumstances) that the ten white and ten black balls will be
found, as diftinguiſhable as ever, and unchanged. But for all
that, there is much to be aſſumed in reckoning upon ſuch a
reſult, which is not ſo objective (in the ſenſe in which I have
uſed the word) as the knowledge of what the balls were when
they were put into the urn. We have to aſſume all that is re-
quiſite to make our experience of the past the means of judging
the future .
Having made this illuſtration to draw a diſtinction, I now pre-
miſe that I throw away objective probability altogether, and con-
ſider the word as meaning the ſtate of the mind with reſpect to
an aſſertion, a coming event, or any other matter on which ab-
،
folute knowledge does not exiſt. It is more probable than im-
probable ' means in this chapter ' I believe that it will happen
more than I believe that it will not happen.' Or rather ' I ought
to believe, &c . :' for it may happen that the ſtate of mind which
is, is not the ſtate of mind which ſhould be. D'Alembert be-
lieved that it was two to one that the first head which the throw
of a halfpenny was to give would occur before the third throw :
a juſter view of the mode of applying the theory would have
taught him it was three to one. But he believed it, and thought
he could ſhow reaſon for his belief: to him the probability was
two to one. But I ſhall ſay, for all that, that the probability is
three to one : meaning, that in the univerſal opinion of thoſe who
examine the ſubject, the ſtate of mind to which a perſon ought
to be able to bring himſelf is to look three times as confidently
upon the arrival as upon the non-arrival.
Probability then, refers to and implies belief, more or leſs, and
belief is but another name for imperfect knowledge, or it may be,
expreſſes the mind in a ſtate of imperfect knowledge. There is
accurate meaning in the phrafe ' to the beſt of his knowledge and
belief, the firſt word applying to the ſtate of his circumstances
with reſpect to external objects, the ſecond to the ſtate of his
174 On Probability .
mind with reſpect to the circumſtances. But we cannot make
any uſe of the distinction here : what we know is to regulate
what we believe ; nor can we make any effective uſe of what
we know, except in obtaining and defcribing what we believe, or
ought to believe. According to common idiom, belief is often
a lower degree of knowledge : but it is imperative upon us to
drop all the quantitative diſtinctions of common life, or rather
to remodel them, when we come to the conſtruction of a
ſcience of quantity.
I have faid that we treat knowledge and belief as magnitudes :
I will now put a broad illuſtration of what I mean. We know,
(ſuppoſe it known) that an urn contains nothing but two balls,
one white and one black, undiftinguishable by feeling : and we
know (ſuppoſe this alſo) that a ball is to be drawn. Disjunctively
then we know white will be drawn : black will be drawn,' one
or the other muſt be. How do we ſtand as to ' white will be
drawn,' and ' black will be drawn,' ſeparately ? Clearly in no
preponderance with reſpect to either. May we then properly
and reaſonably ſay that we divide our knowledge and belief of
the event one or the other ' into two halves, and give one half
to each. I can conceive much objection to this ſuppoſition :
but, whether they formally make it or not, I am fure writers on
probability act upon it, and are accepted by their readers.
Let us confider what magnitude is, that is to ſay, how we
know we are talking about a magnitude. We know that when-
ever we can attach a diſtinct conception of more and leſs to dif-
ferent inſtances, ſo as to ſay this has more than that, we are
talking of comparable magnitudes. We ſpeak of a quantity of
talent, or of prudence : we ſay one man has more talent than
another, and one man more prudence than another : but we never
ſay that one man has more talent than another has prudence. If
we occafionally ſay he (the fame one man) has more talent than
prudence, it is only as an abbreviation : we mean that he has not
prudence enough to guide his talent. Juſt as we might ſay (though
we do not) that there is more cart than horſe, when the horſe
cannot draw the cart : juſt as, ſpeaking very looſely, we do ſay,
the preffure of the atmosphere is not fifty inches ; meaning that it
is not enough to balance the preſſure of fifty inches of mercury
in the barometer. And thus, both up to, and beyond our means
 On Probability . 175
of meaſurement, we form to ourſelves diſtinct notions of com-
parable magnitudes, and incomparable magnitudes, as well as of
the meaning of the ſomewhat incorrect, but eaſily amended,
figures of ſpeech by which we ſometimes talk of comparing the
latter.
But the object of all quantitative ſcience is not merely magni-
tude, but the measurement of magnitude. And when are we en-
titled to ſay that we can meaſure magnitude ? As soon as we
know how, from the greater, to take off a part equal to the leſs :
a proceſs which neceſſarily involves the test of which is the
greater, and which is the leſs, and, in certain caſes, as it may
happen, of neither being the greater nor the leſs. As to ſome
magnitudes, the clear idea of meaſurement comes foon: in the
caſe of length, for example. But let us take a more difficult
one, and trace the ſteps by which we acquire and fix the idea :
ſay weight. What weight is, we need not know : the Newto-
nian, who makes it depend on the earth's attraction, and the
Ariftotelian, who referred it to an impulſe which all bodies pof-
ſeſs to ſeek their natural places, are quite at one on their notions
of the meaſurable magnitude which their ſeveral philoſophies dif-
cuſs. We know it as a magnitude before we give it a name :
any child can diſcover the more that there is in a bullet, and the
less that there is in a cork of twice its fize. Had it not been for
the ſimple contrivance of the balance, which we are well aſſured
(how, it matters not here) enables us to poiſe equal weights
against one another, that is, to detect equality and inequality,
and thence to afcertain how many times the greater contains the
leſs, we might not to this day have had much clearer ideas on the
ſubject of weight, as a magnitude, than we have on thoſe of
talent, prudence, or ſelf-denial, looked at in the ſame light. All
who are ever ſo little of geometers will remember the time when
their notions of an angle, as a magnitude, were as vague as, per-
haps more ſo than, thoſe of a moral quality : and they will alſo
remember the ſteps by which this vagueneſs became clearneſs
and preciſion .
Now a very little confideration will show us that, the moment
we begin to talk of our belief (the mind's meaſure of our know-
ledge) of propoſitions ſet before us, we recognize the relations
called more and leſs. Does the child feel that the bullet has
176 On Probability .
more ſomething than the cork one bit better than an educated
man feels that his belief in the ſtory of the death of Cæfar is
more than his belief in that of the death of Remus. Let any
one try whether he have not in his mind the means of arranging
the following ſet in order of magnitude ofbelief, including within
that term all the range which comes between certain knowledge
of the falſehood, and certain knowledge of the truth, of an affer-
tion. Let them be 1. Cæfar invaded Britain with the ſole view
ofbenefiting the natives. 2. Two and two make five. 3. Two
and two make four. 4. Cæfar invaded Britain. 5. Romulus
founded Rome. He will probably discover the gradations of
neceſſary truth, moral certainty, reaſonable preſumption, utter
incredibility, and neceſſary falſehood. Theſe are but names given
to different ſtates of the mind with reſpect to knowledge of pro-
pofitions afferted ; and I say they expreſs different ſtates of
quantity.
The only difficulty, and a ſerious one it can be made, may be
ſtated in the following question ;-Are we to confider the fort
of belief which we have of a neceſſary propofition (as two and
two make four), that is, abſolute knowledge, to which contra-
diction is glaring abſurdity-as only a ſtrengthened or augmented
ſpecimen of the fort of knowledge which we have of any con-
tingent propofition (fuch as Cæfar invaded Britain) which may
have been, or might have been, falſe, and can be contradicted
without abſurdity ? I anſwer, we can easily ſhow that the dif-
ference of the two caſes is connected with the difference be-
tween finite and infinite, not between two magnitudes of dif-
ferent kinds. The mathematician will easily apprehend this,
and will look upon the various difficulties which ſurround even
the explanation as upon things to which he is well accuſtomed,
and which he underſtands by many parallel inſtances. We can
invent circumſtances under which a contingent propofition ſhall
make any degree of approach to neceffity which we pleaſe, but
ſo that no actual attainment ſhall be arrived at. If an urn con-
tain balls, and if one ball must be drawn, then, the balls being
all white, it is neceſſary that a white ball must be drawn, as
neceſſary as that two and two being in any place, there are four
in that place : for there are no degrees of neceſſity. But let it
be that there are black balls alſo, at the rate of one to a thousand
 On Probability. 177

white ones : the drawing of a white ball is no longer neceſſary ;

but there is ſtill a ſtrong degree of aſſurance that a white ball
will be drawn. We do not readily ſee how much : becauſe the
urn has no viſible relation to our uſual caſes ofjudgment. But
let it be made to repreſent the life of a youth oftwenty : and let
the drawing of a white ball repreſent his living to come of age,
and of a black one his death in the interval. There ought to be
feven black balls to the thousand white ones to make the caſes
parallel. And yet we know that our aſſurance of his ſurvival is
generally very ſtrong : be it wiſe aſſurance or not, it exiſts, and
we act upon it. Now ſuppoſe the rate to be one black to a mil-
lion of white : the aſſurance is much increaſed, but ſtill there is
no neceſſity ; the black ball may be drawn. Take one black to
a million of million of white, or a million of million of million,
&c.: long before we have arrived at ſuch a point, we have loft
all conception of the quantitative difference between our belief in
drawing a white ball, and our belief that two and two are four.
We ſay it is almost impoſſible that one trial ſhould give a black
ball : and this very phraſe is a recognition of the ſameneſs for
which I am contending. Except on the ſuppoſition of ſuch
ſameneſs, there is no almost impoſſible, nor nearly certain. Be-
tween the impoſſible and the poſſible, the certain and the not
certain, there must be every imaginable difference, if we do not
-admit unlimited approach. For it will clearly not be contended
that, repreſenting certainty, ſay by 100, we can make an ap-
proach to it by an uncertainty counting as, ſay 90, but nothing
higher. Repreſenting the ſtate of abſolute knowledge by 100, any
one, with a little confideration, will say that the laws of thought
fix no numerical limit to our approach towards this ſtate : but
that things ſhort of certainty are capable of being brought within
any degree of nearneſs to certainty. On ſuch confiderations, I
ſhall aſſume that neceffity on the one hand, a certainty for, and
impofſſibility on the other, a certainty against, are extreme limits,
which being repreſented by quantities, may allow our knowledge
of all contingent propoſitions to be repreſented by intermediate
quantities.
It must be fully allowed, nay, imperatively inſiſted on, that
nothing in the numerical view, tending to connect neceſſary and
contingent propofitions, can at all leſſen the diſtinction between
N
 178 On Probability.
them ; nor give the latter any reſemblance to the former, except
only in the quantities by which they are indicated. Though
there be only one black ball to as many white ones as would fill
the viſible univerſe, yet between that cafe and the one of no
black balls muſt always exiſt the eſſential difference, that in the
former a black ball may be drawn, and in the latter it cannot.
But this very great distinction between the neceſſarily certain
and the contingent, is it compatible with their being repreſented
by numerical quantities as near to one another as we pleaſe ? I
answer that all who are acquainted with the relations of quantity
`` are aware that nearneſs of value is no bar to any amount ofdif-
ference of properties. A common fraction, for instance, may
be made as near as we pleaſe in value to an integer : but there
do not exiſt, even among propoſitions, more eſſential, or more
ſtriking, differences, than thoſe which exiſt between the properties
of integers and of fractions. There are crowds of theorems (I
ſhould rather ſay unlimited crowds of claſſes of theorems) which
are always true when integers are uſed, and never true when
fractions are uſed. Let any quantities be named, integer or frac-
tional, and it is eaſy to make claſſes of theorems which are true
for thoſe quantities, and not for any others, however near to them.
The reader who is not a mathematician muſt rely upon the know-
ledge of the one who is, that the difference between two quan-
tities, no matter how nearly equal, may be connected with other
differences as complete, and by practice as easily recognized, as
the difference between neceſſary and contingent truth .
I will take it then that all the grades of knowledge, from
knowledge of impoſſibility to knowledge of neceffity, are capable
of being quantitatively conceived. The next queſtion is, are
theſe quantities capable, in any cafe, of meaſurement, or of com-
pariſon with one another. At preſent, we stand as the child
ſtands with reſpect to the bullet and the cork : perceptive of
more and leſs, but without a balance by which to make compa-
rifons. To ſhow the poftulate on which our balance depends,
let us ſuppoſe an urn, which, to our knowledge, contains white,
black, red, green, and blue balls, one of each colour. It is within
our knowledge that a ball must be drawn : accordingly we have
full knowledge (and of courſe entire belief) that the reſult ' no
ball ' is impoffible, and that ' white, or black, or red, or green, or
 On Probability. 179
blue' is neceſſary. To the reſult ' white ' we accord a certain
probability, that is, a certain amount of belief. If a man tell us
that white will be drawn, we may hold him raſh, but we do not
pronounce his communication incredible : let another tell us that
' black, or red, or green, or blue' will be drawn, and we hold him
not fo raſh, and his communication more credible. We may
hold with either, if he will deſcribe his knowledge and belief as
partial, and give them their proper amounts. Now, whether we
ſhall proceed, or ſtop ſhort at this point, depends upon our ac-
ceptance or non-acceptance of the following POSTULATE : -
When any number of events are disjunctively poſſible, ſo that
one of them may happen, but not more than one, the meaſure
of our belief that one out of any fome of them will happen,
ought to be the amount ofthe meaſures of our ſeparate beliefs
in each one of thoſe ſome.
I mean that any one ſhould ſay, A, B, C, being things of
which not more than one can happen, ' my belief that one of the
three will happen is the ſum of my ſeparate beliefs in A, and in
B, and in C. ' This is the poftulate on which the balance de-
pends ; and there is a fimilar poftulate before we can uſe the
phyſical balance. The only difference (and that but apparent)
is that we are to ſpeak of weights collectively, and of events dif-
junctively. The weight ofthe (conjunctive) maſs is the ſum of the
weights of its parts : the credibility ofthe (disjunctive) event is the
ſum of the credibilities of its components. There are ſeveral
may-bes, any one of which may become a has-been : when we
ſpeak disjunctively, it is of the will-be, which cannot be faid of
more than one : the may-be of an event deſcribed as contained in
' A, B, C,' is to be repreſented as in quantity the ſum of thoſe
in ' A,' in ' B,' and in ' C.'
Is it matter of mere neceſſity that, talking of phyſical weight,
the weight of the whole is equal to the ſum of the weights of the
parts ? We have learnt to admit this poftulate, of which no
man ever doubted: but no one can ſay that it was neceſſary.
The laws of matter and mind being both what they are, the con-
nexion between phyſical collection and mental ſummation is, I
grant, neceſſary : the ſimpleſt of manual, and the ſimpleſt of
mental, operations, are and, with us, muſt be, concomitants.
But, in the first place, it is not true that the weight of the
180 On Probability .
whole is equal to the ſum of the weights of the parts, in the
manner in which the reader probably imagines it to be true. Let
the firſt part we hang on the balance be the weight which is
correctly meaſured by W. Then if we hang under it another
weight, as correctly repreſented by V, we think we are quite ſure
when we say that the collective maſs must have a weight W + V
becauſe its parts have the weights W and V. But its parts have
not the weights W and V. The
weight of V is diminiſhed by the
upward attraction of W, and is ,
W
ſay, V-M : the weight of W is
V
as much increaſed by the down-
ward attraction of V, and is W
+M. And though V- M and
W+ M added together do give V + W, yet it was not in this
way that the reader made out his neceffary truth. The univer-
fal equality of action and reaction did not exist in the thoughts of
the firſt perſon who formed a diſtinct conception of the weight
of the whole as compoſed of the ſum of that of the parts : and he
was only right by the (ſo far as he was concerned) accidental
-circumſtance, that two things of which he knew nothing, coun-
terbalanced each other's effects . Nor do we know at this mo-
ment, as of neceffity, that the propofition is correct. We have
much reaſon to think that the law of equality of action and reac-
tion is mathematically true : but, let it fail to the amount ofonly
one grain in a thouſand million of tons, and the propoſition is
not true, but only nearly true.
Again, the co-existence of thoſe laws of mind and matter
which beſt, ſo to ſpeak, fit each other, and which make the phe-
nomena of the external world, after due confideration, appear to
be almoſt what they must have been, is not, to our apprehenſions,
a neceſſary coexiſtence. We can imagine the following reſult,
though we cannot trace what the full conſequences of it would
be on the expreſſion of the laws ofthought. Conceive ſentient
beings, to whom the ſimpleſt mode of arithmetical ſucceſſion is
not 0, 1, 2, 3, &c. but 1, 10, 100, 1000, &c. their powers of
numeration being ſo conſtructed that the ſecond of theſe ſuccef-
ſions has that character of fundamental ſimplicity which we
attach to the first. Of course, their primary ſymbols would be
 On Probability . 181

ſignificative of 1, 10, 100, &c. It would be impoſſible for us to

conceive any mode by which ten or any other number could be
thus fundamentally attached to unity, in a manner ſhared by no
third number : but, I am not ſaying, ' Imagine how this could
7
be,' but, ' Imagine that it is.' There is no contradiction in the
ſuppoſition, either to itſelf, or, till we know much more of the
mind than we now do, to anything elſe. Beings ſo conſtituted
would have logarithmic brains ; and if, thus conſtituted, they
were placed among our material laws of exiſtence, the manner
in which the weight of the whole is to be inferred from thoſe of
the parts, would be a profound mystery for ages, only to be folved
in an advanced ſtage of mathematical ſcience. A recent mode
ofconſtructing mathematical tables, which generally carries with
it the name of its eminent inventor, Gauſs, would conſtitute
one of their principal neceſſities : they would have to uſe it as
their only mode (except actual experiment) of finding out that
what we repreſent by 156 and 200, together make (and this
making would be a complicated proceſs) 356.
Instead, then, of trying to eſtabliſh it as perfectly natural and
neceſſary to ſay that our belief of ' one of the two A or B, when
both cannot happen,' is, quantitatively ſpeaking, the ſum of our
belief in A, and our belief in B, I have rather endeavoured to
ſhow that the analogous caſes with which we first think of com-
paring this propofition, other kinds of compoſition, are not fo
natural and neceſſary as is ſuppoſed. There are two ways of
levelling ; by bringing up the lower, or bringing down the higher.
And I particularly wiſh in this chapter to prevent the reader
from accepting the arithmetical doctrine of probability quite fo
rapidly as is usually done. In furtherance of this object, I pro-
ceed to the following poſſible objection.
It may be ſaid, you have, by thus formally identifying proba-
bility with belief, and ſtating a poſtulate which, in expreſs terms,
has not the moſt axiomatic degree of evidence, rendered ſome-
what difficult that which in the ordinary view of fimple chances,
is very eaſy. This charge, I hope, is true : ſuch was my inten-
tion, at least. And my reaſon is, that in the ordinary view of
the ſubject, one of two things occurs : either probability is ſepa-
rated by definition from ſtate of belief, though it be known that
the two words will afterwards be confounded without any per
182
On Probability
miffion ; or elſe the poftulate is tacitly aſſumed, and the difficulty
which I ſuppoſe myſelf charged with introducing, is ſlurred over.
Take a common queſtion ;-An urn has two white balls and
five black ones : there are ſeven equally likely drawings, two
white ; therefore the chance or probability of drawing a white
ball is called two-ſevenths. But the chance of either particular
white ball is one-ſeventh. Now firſt, if any one ſhould ſay
that this is mere definition, I can, of course, allow it : but it then
remains to ſhow what connexion this defined probability has with
any ordinary acceptation of the word. But if, probability mean-
ing belief, or fentiment of probability actually exifting in the
mind, or index of the proper degree of belief, &c. &c.-the
above ſtatement be made as fundamentally evident, I ſhould then
aſk how it is known that the probability of one or the other
white ball being drawn ' is properly ſet down as the fum of the
probabilities of the ſeparate white balls. And I cannot conceive
any anſwer except that it is by an aſſumption of the poftulate.
That ſuch aſſumption will finally be knowingly made, on the
fullest conviction, by every one who ſtudies the theory, I have
no doubt whatever : nor that it has been made, no matter in
what words, nor with what clearness of avowal, by every one
who has studied that theory. And therefore I hold it defirable
that the beginner ſhould know what I have here told him .
It is indifferent, as far as the theory is concerned, what nu-
merical ſcale of belief we take. We might, if we pleaſed, copy
Fahrenheit's thermometer, ſet down knowledge of impoffi-
bility as 32 ° , perfect certainty as 212 °, and other ſtates of mind
accordingly. Thus, 122° would repreſent perfect indecifion,
belief inclining neither way, an even chance. But this would
complicate our formulæ : the uſual and preferable plan is to af-
ſume o as the index of knowledge of impoſſibility, I as that of
certainty, and intermediate fractions for the intermediate ſtates.
This mode of eſtimation makes formulæ and proceſſes ſo much
more eaſy than any other, that it must be adopted ; but there is
a ſtrong objection to it in one point of view: as follows.
When we speak of belief in common life, we always mean
that we confider the object of belief more likely than not : the
ſtate of mind in which we rather reject than admit, we call
unbelief. When the mind is quite unbalanced either way, we
 On Probability. 183
have no word to expreſs it, becauſe the ſtate is not a popular *
one. The quantitative theory calls by the name ofbelief every
admiffion of poffibility. When there is only one black ball to a
million of white ones, there is ſome belief that a black ball will
be drawn ; a much larger belief in a white one. Itwould be
advantageous in ſome reſpects that o ſhould repreſent the ſtate
of indifference, + 1 , that of knowledge of certainty, and - 1,
that of knowledge of impoſſibility. But this would complicate
formulæ too much. I conſider it therefore defirable to use the
common meaſures and formulæ, but to aſſociate them with the
onejuſt propoſed, in the following manner.
When a perſon tells us that his belief in an aſſertion is, ſay ro,
meaning that he conſiders it 3 for and 7 against, or 7 to 3 against,
we ſhould ſay in common talk that he diſbelieves, but not very
ſtrongly. In the language of this theory, we say that he both
believes and diſbelieves, the latter more ſtrongly than the former.
Let us add that his authority is against the conclufion. If he ſay
that it is in his mind an even chance, or that he has no opinion
one way or the other, let us ſay that he gives no authority either
way. If we adapt this definition to the ſuppoſition that + I and
- I repreſent the extremes of authority for and againſt, we have
the following rules. The meaſure of authority is twice the mea-
ſure of belief diminiſhed by unity, for, when poſitive, against,
when negative : the meaſure of belief is half of unity increaſed
(algebraically) by the meaſure of authority. If a repreſent the
meaſure of belief, and A that of authority, then
A= 2a- 1 , a= ( 1 + A)
It is alſo adviſable to have a term to repreſent what are uſually
called the odds. Some might think it defirable to rid the ſubject
as much as poſſible of words derived from gambling : aſtrono-
mers have done the ſame thing with the phrases of astrology, and
chemiſts with thoſe of alchemy. When it is 7 for and 3 against,

* Many minds, and almoſt all uneducated ones, can hardly retain an
intermediate ſtate. Put it to the firſt comer, what he thinks on the queſtion
whether there be volcanoes on the unſeen ſide of the moon larger than
thoſe on our fide. The odds are, that though he has never thought of the
queſtion, he has a pretty ſtiff opinion in three ſeconds.
184 On Probability.
it might be faid that the relative testimony for, is 4, and that
against, . But the brevity of the firſt phrase will inſure its con-
tinuance, let who will try to change it.
The ordinary rule is a conſequence of the notions hereinbefore
laid down, and of the particular mode of meaſurement adopted.
It is as follows ;-When all the things that can happen can be
reſolved into a number of equally probable (or credible) cafes,
fome favourable and ſome unfavourable to the event under con-
fideration, then the fraction which the favourable caſes are of all
the caſes, meaſures the probability (or credibility) of the arrival
of the event : and the fraction which the unfavourable caſes are
of all the cafes, meaſures the probability (or credibility) of the
non-arrival. There are, for instance, in an urn, 5 white, 4 black,
and 3 red balls, 12 in all. It is aſſumed that we know them to
be equally likely to be drawn ; which here means no more than
that we know nothing to the contrary. That one ball muſt be
drawn, is ſuppoſed certainly known. Accordingly, our belief in
' one or another' is repreſented by I : which is, by the poftulate,
the ſum of the ſeveral credibilities of the balls ; which laſt are all
equal. Therefore each ball has : and by the ſame poftulate,
the event ' one or other of the white balls ' or the drawing of a
white ball, has ; of a black ball ; of a red ball, 1.
Inſtances like the above, in which we invent all the caſes and
have arbitrary power over their number, are the only ones on
which we can employ à priori numerical reaſoning. They are
alſo the only ones on which we can try experiments. It is im-
portant to know whether, as a matter of fact, our belief, nume-
rically formed, will be approximately juftified by the reſults of
trial. And this juſtification is found to exiſt, in the following
way. It is a remote, but certain, concluſion from the theory,
requiring mathematical reaſoning too complicated to introduce
here, that events will, in the long run, happen in numbers pro-
portional to the objective probabilities under which the trials are
made. For instance ;-if a die be correctly formed, ſo that no
one face has more tendency than another to fall upwards, the
probability of throwing an ace is ; that of not throwing an ace
is . The theory tells us its own worthleſſneſs, if in the long
run, not-ace do not occur five times as often as ace. If 60,000
trials were made, the theory would tell us to expect about 10,000
 On Probability. 185
aces and about 50,000 not-aces. Practice confirms the theory :
not, that I know of, in the actual caſe juſt cited, but in fimilar
ones. I will ſtate an inſtance.
Throw a half-penny up, and if it give tail, repeat the throw,
and ſo on, till head arrives : and let this ſucceſſion be called afet.
The probability that a ſet ſhall confift of one throw, is ſhewn
by the theory to be ; that it ſhall have two throws,; three
throws,; and ſo on. If a very large number of ſets be tried,
we are to expect that about half will be of one throw, about a
quarter of two throws, about an eighth of three throws ; and ſo
on, as long as the number is large enough to give any proſpect
of ſomething like an average. This experiment has been tried
twice : once by the celebrated Buffon, and once by a young
pupil of mine, for his own fatisfaction ; both in 2,048 ſets. The
reſults were as follows ; the third column ſhowing the number
of each kind which the theory aſſerts to be moſt probable.
B H
Head at the first throw 1061 | 1048 | 1024
No head till the 2nd throw 494 507 512
3rd 232 248 256
4th
-

137 99 128
5th 56 71 64
6th 29 38 32
7th 25 17 16
8th 8 9 8

9th 6 5 4
10th 3 2

11th I I

12th Ο

13th -

14th I
I
15th -

16th I

&c Ο

2048 2048 2048

In Buffon's trials, there were altogether 1992 tails to 2048
heads, and in Mr. H's there were 2044 tails to 2048 heads.
Inſtances in which we can command all the cafes are to the
186 On Probability.
mind, in this theory, what acceſſible lengths are to the eye. We
can meaſure the latter by a rule, and ſo train the organ to judge
of lengths which cannot be approached, or caſes in which the
rule is not at hand.
I ſhall now refer the reader to other works on the ſubject, for
further details on the operative part, and proceed to juſt as much
as is neceſſary for the particular purpoſe of the next chapter,
namely, the application of the hypothefis of meaſure of belief to
queſtions of argument and testimony. Two theorems will be
enough : the firſt relating to independent events, the ſecond to
the probability of events which are neither wholly independent,
nor wholly confequent, either upon the other. The word event
is uſed in the wideſt poſſible ſenſe: it does not even neceſſarily
mean future event. Unleſs our knowledge, either of the cir-
cumſtances, or of the event itſelf, thereby undergo ſome altera-
tion, it is nothing to us now whether it has happened, or is to
happen.
Let there be two events, P and Q, of which the probabilities
are the fractions a and b ; and let them be wholly independent of
one another, the arrival or non-arrival of either being perfectly
independent of that of the other. The probability that both
ſhall happen is the product of a and b : and ſimilarly for more
events than two. Suppoſe, to take an inſtance, that a is ≠ and b
is . We must then conſider P as an event which has 3 ways
of failing to 4 of happening : if we would have an urn from
which the credibility of drawing a white ball ſhould be that of
the happening of P, we must put in 4 white balls and 3 not
white (ſay black) balls. Similarly to repreſent Q, we must have
an urn of 3 white and 2 black balls. Now to afcertain the
profpect of drawing white from both urns, we must count all
the caſes. A ball from the urn of 7 may be combined with one
from the urn of 5, in 7 x 5 or 35 ways. But a white ball from
the firſt urn may be combined with a white ball from the ſecond,
in 4 × 3 or 12 different ways. There are then 35 caſes in all,
12 of which are favourable : hence the probability in favour of
white from both (which is that of the two events both happen-
ing) is
12

35
ororx
7× 5 75
or ab.
 On Probability . 187
Similar reaſoning may be applied to more events than two.
This theorem has a large number of confequences, ſome of
which we may notice.
When a is the probability for, 1 - a is the probability againft.
This I ſhall always denote by a' : fimilarly b' will stand for 1 - b ;
and ſo on .
Required the probability that of a number of independent
events, P,Q, R, &c one or more ſhall happen. Let a,b,c, &c. be
the ſeveral probabilities, then that of their all failing is the pro-
duct a'b'c' ....and that of their not all failing (or of one or
more happening) is 1 -a'b'c' .... Accordingly, if there be
only two events, for ' one or both ' we have 1- (1 - a)( 1 - b )
which is a + b-ab. If the number of events be n, and all
equally probable (fo that a =b =c, &c. ) for ' one or more ' we
have 1 - a " or I- ( 1 - a)" .
It is a conſequence of this laſt that, however unlikely an event
may be, it is ſure (in the common ſenſe of the word) to happen,
if the trial can be repeated as often as we pleaſe. However
ſmall a may be, or however near to unity I- a, n may be taken
ſo great that ( 1 - a)" ſhall be as ſmall as we pleaſe, or 1- ( 1 - a)"
as near to unity as we pleaſe, or the probability that the unlikely
event will happen once or more in n times, as great as we pleaſe.
Let a = 1 : (k + 1), which means that the odds are k to I against
the event on any one trial : the following rough deductions will
ſhow what kind of reſults the formula gives, true within an in-
ſtance or two when k is confiderable. In k inſtances it is an
even chance that the event happens once or more ; in 2.3k, it is
9 to 1 ; in 4.6k, 99 to 1 ; 6.9k, 999 to 1 ; 9.2k, 9999 to 1 :
and in 23k, it is ten thouſand millions to Ⅰ . Thus, ſuppoſe at
each trial it is a hundred to one againſt ſucceſs. Then of thoſe
who try 70 efforts, as many will ſucceed once or more as will
altogether fail, in the long run. Of thoſe who try 6900 times,
only one of a thouſand will always fail. A person who will not
examine an aſſertion that comes to him with ten to one againft
it, muſt count it an even chance that he throws away one or
more truths, if he follow his plan ſeven times .
Let us now ſuppoſe that there are reaſons why the ſeveral in-
ſtances which can arrive are not equally credible. Suppoſe the
urn to contain a white, a black, and a red ball, and ourſelves to
 188 On Probability.
have reaſons to think the balls not equally probable or credible,
but that 6, 5, and 2 are the proportions of the degrees of belief
we ſhould accord to them ſeverally. If then 6x repreſent the
probability of a black ball, 5x and 2x will represent thoſe of the
other two ſeverally. By the poftulate, 13x repreſents that of one
or the other. But this is certainty ; whence x must be 13, and
6
,, and are the probabilities of the white, black, and red
balls. That is to ſay, when the ſeveral inſtances are unequally
probable, we must count each inſtance as though it occurred a
number of times proportioned to its probability, and then proceed
as in the caſe of equally probable inſtances. Thus, in the above,
instead of ſaying (as we ſhould do if the balls were equally pro-
bable) that the probability of the white ball is
I 6 6m
or
I + I + I,
we ſay it is 6 + 5 + 2 ; 6m + 5m + 2m
would do, m being any number or fraction whatsoever.
Now ſuppoſe two urns, one of all white balls, and the other
of all black ones. If we actually draw a ball, and find it white,
we know that the urn choſen to draw from must have been the
firſt : the ſecond could not have given that drawing. But fup-
poſe the firſt urn to have 99 white balls to one black, and the
ſecond one white to 1000 black. If we now draw again, and
draw a white one, not knowing from which we drew, we feel
almoſt certain, from the drawing, that we have choſen the firſt
urn. We ſtill feel almost certain that the ſecond urn would have
given a black ball. This inverſion of circumſtances, this conclu-
ſion that the circumstances under which the event did happen,
are moſt probably thoſe which would have been most likely to
bring about the event, is of the utmoſt evidence to our minds :
but the queſtion now before us is, are we to call it a ſecond poſtu-
late, or is it deducible from the other one ? It is ſo deducible,
and is not a ſecond poftulate ; but it has not been uſual to give a
very diſtinct account of the deduction. * If it could not be made,
* So well eſtabliſhed is this ſpecies of inverſion in the mind, that both
Laplace and Poiſſon, the two moſt eminent mathematical writers on the fub-
ject, of the preſent century, have in a certain caſe aſſumed that an equation
which gives the moſt probable value of x in terms of y, is therefore the one
which gives the moſt probable value ofy in terms of x. This is carrying
the principle too far.
 On Probability. 189
the following proceſs would, no doubt, be ſufficient : it has often
been held fo. Let the urns have 6 white balls to I black, and 2
white balls to 9 black. Then the probabilities of drawing a
2

white ball from the two are and rr, which are in the propor-
tion of 33 to 7. If, becauſe when we choose the firſt urn, we
have nearly five times as much chance of a white ball as the
ſecond one would give, we conclude that a known white ball
from an unknown urn is in that proportion more likely to have
come from the firſt urn ; we shall have and 45 for the proper
degrees of belief in the two urns. For if 33x be that for the
firſt urn, then 7x muſt, by the aſſumption, be that for the
ſecond : and for one or the other, we have 40x. But this is
I

certainty ; whence x must be 40

To reduce this reſult to dependence upon the firſt poſtulate,

proceed as follows. The probability that two events are con-
nected, our belief, that is, in the connexion, must be the ſame
whether the two events, or either of them, have happened, or
whether they be yet to happen : unless there be ſomething
in the happening which alters our knowledge, and puts us in
a different ſtate for forming a judgment. Suppoſe I make up
my mind, rightly or wrongly, as to how far I will believe that
a white ball, if drawn, will have been drawn from the firſt urn.
An inſtant after, I am told that the trial I anticipated has been
made, and the contingency which I ſuppoſed has occurred ; a
white ball has been drawn. I know no more than I took myſelf
to know in my hypothesis ; and cannot therefore have any
means of altering my opinion. Now, without altering the pro-
portions in the urns, change the numbers of the balls, ſo that
there may be the ſame total number in each : let them be
{66 white, 11 black} {14 white, 63 black}
Now put each ball in an urn by itſelf, 154 urns in all. This
gives to any one ball, if I choose an urn at hazard. But it
was ſo before : as to the firſt of the two urns for inſtance, was
I

the probability of choofing that urn, andI

that of choofing one
particular ball from it : and × is 14. If we then remove
all the urns with black balls, ſo that a white ball must be drawn,
the chance of its being one of the 66 is for . If without
removing the black balls, we think of the probability of a white
 190 On Probability.
ball, if drawn, being of the 66, or of the 14, the credibilities of
thoſe ſuppoſitions are as 66 to 14. If, having choſen an urn, we
find it contains a white ball, the ſame probabilities are ſtill in
that proportion.
The rules derived from ſimilar reaſoning, whether for judging
of the probabilities of precedents from an obſerved confequent,
or for judging of the probabilities of events which reſtrict each
other, are preciſely the fame, as follows. If the probability of
the obſerved event, ſuppoſed ſtill future, from the ſeveral poſſible
precedents, ſeverally ſuppoſed actually to exiſt, be a,b,c, &c : then,
when the event is known to have happened, the probabilities
that it happened from the ſeveral precedents are
a b
for the firſt, for the ſecond, &c.
a + b + c + ... a + b + c ....

Again, if there be ſeveral events, which are not all that could
have happened ; and if, by a new arrangement (or by additional
knowledge of old ones) we find that theſe ſeveral events are now
made all that can happen, without alteration of their relative cre-
dibilities : their probabilities are found by the fame rule. If a, b,
c, &c. be the probabilities ofthe ſeveral events, when not restricted
to be the only ones : then, after the restriction, the probability
of the firſt is a÷(a + b + ... ), of the ſecond, b ÷(a + b + ... )
and ſo on .
Wemay obtain a very distinct notion of this laſt theorem, as
follows . Suppoſe two events, which are among thoſe that can
happen, and let one, ſay, be twice as probable as the other. This
means, that among all the independent, and equally likely, caſes,
there are twice as many favourable to the firſt as to the ſecond.
Now, ſuppoſe by ſome alteration of ſuppoſitions, the introduction
of new knowledge, for instance, it is found, all the caſes remain-
ing as before, that all are prevented from happening except theſe
two events. This new ſtate of things does not alter the caſes in
number : accordingly, the proportion of the probabilities of the
two events is as before, two to one. But now one of them muſt
happen : or the ſum of theſe probabilities must be unity. It
follows then that one of them is , and the other . The ſame
reaſoning may be applied to more complicated cafes.
It frequently happens, when different problems are ſolved by
 On Probability . 191
the fame formula, that they may be confidered as the fame pro-
blem in two different points of view : and alſo that one and the
ſame problem may be confidered as belonging to either claſs.
For instance ;-Let there be two witneſſes, whoſe credibilities
(or the probabilities that in any given inſtance they are correct)
are a and b . As long as we do not know that they are talking
about the ſame thing, the probability that both will tell truth is
ab. But the moment we know that they both affert the fame
thing, the problem is changed : they must now be either both
right or both wrong ; before, one might have been right and
the other wrong. To take the firſt view of the problem, we
have now an obſerved event, both ſtate that the circumſtance
did happen. There are two precedents ; the event did, or did
not, happen. If it did, the probability of the obſerved event
(which is then that both are right) would be ab ; if it did not, it
would then be (1 - a)( 1-6). Accordingly, the probability that
the obſerved event did happen, will be, by the rule above, ab
divided by ab + ( 1 - a)(1 - b) .
If we take the ſecond view, we have, before the restriction,
four poffible caſes, the probabilities of which are ab, a( 1-6),
b( 1 - a) and ( 1 - a)( 1-6). After the restriction, only the firſt
and fourth are poſſible : whence the conclufion is as juſt given.
Full exemplifications of theſe methods will be found in the next
chapter.

CHAPTER X.

On probable Inference.
are two ſources of conviction, argument and testi-
THERE
mony, reason why the thing ſhould be, ſtatement that the
thing is. When the argument is neceffarily good, we call it
demonstration : when the ſtatement can be abſolutely relied on,
we call it authority. Both words are uſed in lower than their
abſolute ſenſes ; thus, very cogent arguments are often called
demonſtration, and very good evidence, authority.
I ſhall ſuppoſe all the arguments I ſpeak of to be logically
 192 On Probability.
valid ; that is, having conclufions which certainly follow from
the premiſes. If then the premiſes be all true, the conclufion is
certainly true. If a, b, c, &c. be the probabilities of the indepen-
dent premiſes, or the independent propoſitions from which pre-
miſes are deduced, then the product abc... is the probability that
the argument is every way good.
Argument being offer of proof, its failure is only failure of
proof : and the conclufion may yet be true. But teftimony is an
aſſertion of the truth of the conclufion ; and its failure can only
be failure of truth. If a propoſition of Euclid turn out to be
badly demonftrated, the enunciation need not therefore be falſe.
An argument may prove, diſprove, or neither prove nor dif-
prove : a teſtimony cannot be true, falſe, or neither true nor
falſe. This diftinction generally gains no more than a one-fided
admiffion : perſons begin to ſee it when ſome over-zealous bro-
ther writes weakly on their own ſide of a queſtion ; but they are
very apt to think, with respect to the other fide, that anſwering
the arguments is diſproving the conclufion.
Teſtimony is, for the above reaſon, more easily understood than
argument. It is the moſt effective mode of conveying know-
ledge to the uneducated. But it must not be ſuppoſed that, in
any ſtage of reaſon, argument can be the only vehicle of infor-
mation, even on ſubjects called argumentative. This point is
one of great importance.
When argument is demonſtration, it eſtabliſhes its conclufion
against all teſtimony. The idea of an infallible witneſs bearing
evidence against a demonftrated conclufion, is a contradiction.
That n conſecutive numbers have a ſum which is diviſible by n,
whenever n is odd, is demonſtrated. If a thousand of the beſt
qualified witneſſes that ever lived, both for honesty and arith-
metic, were to ſwear that they had diſcovered IOI very high
conſecutive numbers, the ſum of which is not diviſible by 101 ,
any mere beginner in mathematics would be more ſure that a
thouſand good witneſſes had lost their wits or their characters,
' than any one elſe can be of anything not admitting of demon-
ſtration .

But when argument does not amount to demonſtration, not

only is the truth or falſehood of the conclufion matter of credi-
bility, but the iſſue of the argument is not that mere truth or
 On Probability. 193
falſehood. It does not ſtand thus : ' According as this argument
is good or bad, ſo is the concluſion true or falſe, but ' According
as this argument is good or bad, ſo is the concluſion true in this
way, or not true in this way, (that is, either falſe, or true in ſome
other way). ' If we were to say ' men are trees, and trees have
reaſon, therefore men have reaſon, we have a perfectly logical
argument, falſe in the matter of both premiſes : but we cannot
deny the conclufion.
Suppoſe now that an argument is preſented to us of which we
are fatisfied that the like will prove their conclufions to be true
in the particular modes aſſerted, in nine caſes out of ten. What
are we to ſay of the truth or falſehood of the conclufion ? We
have of belief to its being true in one particular way : how
much ſhall we add for other poſſible ways ? Are we to reſt in
the conclufion as having 9 to 1 for it, or are we to allow more ?
We cannot ſay, let us confine ourſelves to the grounds we have
got, and believe or diſbelieve, not in the conclufion, but in the
conclufion as obtained in that one way.
I take it for granted that the mind must have a ſtate with
reſpect to every aſſertion preſented to it, with reaſon, or without
reaſon. Every propoſition, the terms of which convey any mean-
ing, at once, when brought forward, puts the hearer into ſome
degree of belief, or, if we uſe the common phrafe, of belief or
unbelief : including, of course, the intermediate ſtate, which is as
clearly marked upon our ſcale as any other. Men who are
accuſtomed to ſuſpend their opinion, as it is called, that is, to
throw themſelves into the intermediate ſtate when they have
no definite reaſon to think either way, are interested in this
queſtion as much as any others. If there be ſome ſtate, though
not numerically appreciable, in which their belief muſt be, there
is ſome ſtate, which they would rather know numerically than
not, in which it ought to be. In the preceding caſe, ſuppoſe it
known that 9 to 1, or fo, is granted to the conclufion from the
argument alone, and any one wishes to ſuſpend his opinion as to
I

the remaining . Is he to grant half of that to, and ſay that

1+
I
or is what he would wish to make the meaſure of his
belief, if he knew how ? The confideration of this question will
enter among others.
The manner in which he deals with the reſult of the argument
Ο
 194 On Probability.
muft depend upon testimony, uſing the word in its wideſt ſenſe.
Firſt, every man has, as juſt noticed, a teftimony in his own
mind as to every propoſition. He may ſet out with the inter-
mediate ſtate : he may have no reaſon to lean either way, and
may know it ; that is to ſay, he may have to apply an argument
off to an exifting probability of Or he may have previous
good reaſon, or bad reaſon, which makes him lean to the affer-
tion or denial ; and the meaſure of this leaning muſt then be
combined with 16. Or he may have other teftimony to combine
with that of his own previous ſtate. Any way, he cannot have
a definite opinion on the bare truth or falſehood of the conclufion
of the argument, without appeal to the previous ſtate of his own
mind at least, if not to that of others .
It is generally ſaid that we are to throw away authority, and
judge by argument alone ; that our reaſon is to be convinced,
and not biaffed by the opinion of others ; that no conclufions are
worth anything, except thoſe which a man forms for himself. All
the forms in which this frequent caution is expreſſed, I take to
be distortions ofthe very needful warning not to allow authority
more weight than is properly due to it : a warning, by the way,
which is just as much wanted with reſpect to argument as to
authority. For every mistake which has been made by taking
authorities on truſt (that is, taking bad witneſſes to prove the
goodness of aſſerted good ones), one miſtake at least has been
made by taking arguments on preponderance : that is, treating
them as proving their concluſion, as foon as they ſhow it to be
more likely than its contradiction.
To form the habit of allowing authority no more weight than
is due to it, and the fame of argument, is undoubtedly one great
object of mental cultivation : but it ought not to be forgotten
that it is another and juſt as great an object to form the habit of
allowing them no less. Suppoſe an argument of value T is pre-
ſented, and that at the ſame time we have the teſtimony of a
witneſs against the concluſion, of whom we know that he leads
us right 1000 times for each once that he misleads us. Isthere
any ſenſe in reducing this witneſs to one of no authority, or of an
even chance, upon the principle of depending on argument only?
Except the argument be demonſtration, we must be prepared to
-admit that a witneſs may be as good as an argument, or better.
 On Probability. 195
I ſhall now proceed to the ſeveral problems which this ſubject
requires, confidering firſt teftimony alone, next argument alone,
and then the two in combination .
Problem 1. There are independent teftimonies to the truth of
an affertion, of the value μ, ν, ρ, &c. (one of them being the
initial teftimony of the mind itſelfwhich is to form the judg-
ment) : required the value of the united teftimony.
Let μ' be 1 - μ, &c. as in page 187. Here is a problem of
the ſame claſs as in page 190 ; the restrictions are, that all the
teftimonies are right, or all wrong, the independent chances of
which are μνρ ... and μ'ν' ' ... Hence the probabilities are
μυρ ...
for ;
μνρ' ...
μυρ... +μ'ν'' ... μυρ... +μ' '... againft.
Obſerve, firſt, that any numbers proportional to μ, μ' &c. will
do as well : and if the products have a common denominator,
(as generally they have) the numerators only need be used. Se-
condly, the easiest way of expreffing the reſult is by faying that it
is μνρ... τo μ'ν' ' ... for, or μ'ν'ρ'... τo μνρ ... against.
For instance, let it be in my mind 99 to one againſt an affer-
tion, that is, I bear only the teftimony rod in favour of it. Let
four witneſſes, for whoſe accuracy it is 2 to 1, 3 to 1, 4 to 1, 5
to 1 , depoſe in favour of it : I want to know how it ought to
ſtand in my mind. The teftimonies for and againft, are
I 2 3. 4. 5. and 99 I II I
100' 3' 4' 5' 6'
Hence, neglecting the common denominator, it ought to be
IX2 × 3 × 4 × 5 to 99XIXIXIXI , or 120 to 99, or 40 to 33,
for the aſſertion .
Obſerve that in ſaying the witneſs gives teftimony, ſay , it is
of no confequence whether it be a queſtion of judgment, or of
veracity, or of both together. I mean that, come how it may, I
am ſatisfied that when he ſays anything, it is 2 to I he ſays what
is correct .
An eaſy rule for the more common modes of expreffion pre-
ſents itſelf thus. The combined relative teftimony is the product
of the ſeparate relative teftimonies. Thus, two witneſſes of 6
truths to one error, and of 7 truths to one error, are equivalent
 196 On Probability.
to one witneſs of 42 (or 6x7) truths to one error. Three wit-
neſſes of 8, 6, 5 truths to 7, 3, 11 errors are equivalent to one
witneſs of 80 truths to 77 errors .
Ajury of twelve equally trustworthy perſons, after conferring
together, agree to an affertion on which previously I had no
leaning. Suppofing me fully fatisfied that ſuch agreement gives
100 to 1 for their reſult, what am I to think of the deliberate
opinion of any one among them, that is, of his opinion after he
has had the advantage ofdiſcuſſion with others.
Let μ be the value of ſuch teftimony from any one ; then by
the queſtion

μ12 : ( 1 -μ)12 : : 100 : 1 , or μ : Ι -μ :: 100 : I

ſay as 1.468 to 1. That is, I think inconfiftently if I rely on
the united verdict as upon 100 to 1, unleſs I am prepared to
think it 1468 to 1000, or about 3 to 2, for each juror alone.
Ofm + n equally trustworthy jurors, a majority m are for, and
n against, a conclufion. If u be the value of the teftimony of
each, then the odds are to be taken as being μ ™ ( 1 - μ )" for, and
μ"( 1 -μ)m againft. But
μη( I1 -μ) : μ²( I1 - µ)m : : µm -n : (1 -µ)m-n
which are exactly as if the majority m- n had been all, and
unanimous. From the original formula it will appear that two
equally good teftimonies on oppoſite ſides produce no effect on
the reſult.
If then, the unanimity of the jury box in this country could
be confidered as that of deliberate conviction, we might say that
a larger jury, with the condition that the majority ſhould exceed
the minority by 12 at least, would be always as good, and often
better. But there are various confiderations which prevent the
above reſult from being applicable. The neceffity of being unan-
imous, as our law ſtands, may lower the value of the verdict. On
the other hand, a jury of 30, required to find by a majority of 12,
would generally proceed to a vote before they had put the matter
to each other with the real defire to gain opinion which the pre-
ſent practice produces : conſequently, the value of their verdict
would perhaps be lower than that of the majority only, required
to be unanimous .
 On Probability. 197

The theory thus appears to confirm the notion on which we

often act, that a given exceſs of majority over minority, is of the
ſame value whatever the numbers in the two may be. And this
might be the cafe, if the thing called deliberation in a large body,
were as well adapted to the diſcovery of truth as the fame thing
in a ſmaller one. The reader muſt remember that this teft does
not compare the one witneſs on his own judgment with a num-
ber after common deliberation ; but the firſt, after common deli-
beration with others, is compared with the whole.
But in this, and all the problems of this chapter, the distinction
must be carefully drawn between the credibility of a circumſtance
at one time and at another. For example, a witneſs enters with
10 to 1 in his favour, and owing to combination with others, the
refult comes out that it is 100 to I he is in error in the par-
ticular matter on which he gives evidence. We cannot believe
both that it is 10 to I he is right, and 100 to 1 that he is wrong.
What we believe is the latter, for the cafe in queſtion.
As another inſtance, ſuppoſe m independent witneſſes of equal
goodneſs (μ) unite in affirming that a certain ball was drawn from
a lottery of n balls : colluſion being ſuppoſed impoffible. My
knowledge of the circumstances of the affirmation here alters the
problem. If n be confiderable, it is almoſt impoſſible that the
witneſſes, by independent falſehood or error, ſhould all pitch on
the fame wrong ball. To find the bias this ought to give me to
the conclufion that they have told the truth, I muſt obſerve that
there being n - 1 balls not drawn, whichever of theſe any one
chooſes, by error, the chance of any one of the reft choofing the
fame is ÷( n - 1 ), the probability that all the m- 1 ſhall chooſe
the fame is ÷(n - 1 ) -1. Hence, the odds are as μ™ to ( 1 - μ )m
multiplied by the laſt-named expreſſion, or as (n - 1)m--1µm to
( 1 -μ) . If n be very great, the odds may be enormous for the
aſſertion, even though μ, the credibility of each witneſs, may be
ſmall. In caſes of ordinary evidence, the thing aſſerted is usually
one out of almoſt an infinite number of equally poſſible aſſertions,
and the agreement of even two witneſſes (for when m is two or
upwards, n appears in the formula) is certain conviction, if, as
aſſumed, we know the two witneſſes to be not in colluſion. If
μ= 1 ÷ n, which is as much as to ſay that the evidence of each
witneſs makes a ball no more likely to have been the one drawn,
 198 On Probability.
becauſe he ſays it, that it was on our mere knowledge that a ball
had been drawn, it turns out I to n- 1 for the truth of the affer-
tion, juſt as it was before the evidence. But let μ =( 1 + a)÷ n,
a being any fraction, however ſmall, that is, let each witneſs
make the aſſertion more probable than at first, however little :
then the odds for its truth become

(I + a)" to
(1 n- I )m- 1(n - 1-a)
which odds may be made as great as we pleaſe, by ſufficiently in-
creaſing m. That is to ſay, however little each witneſs may be
good for, in real ſupport of the aſſertion, or in making it more
probable than it is of itſelf, a ſufficient number of witneſſes, cer-
tainly independent, will give it any degree of credibility what-
ever.

The ſtudent of this ſubject is always ftruck by the frequency

of the problems in which the ſcience confirms an ordinary notion
of common life, or is confirmed by it, according to his ſtate of
mind with reſpect to the whole doctrine. It is impoffible to ſay
that we have a theory made to explain common phenomena, and
hence affording no reaſon for ſurpriſe that it does explain them.
The firſt principles are too few and two fimple, the train of
deduction ends in conclufions far too remote . I believe hundreds
of caſes might be cited in which the reſults of this theory are
found already efſtabliſhed by the common ſenſe of mankind : in
many of them, the mathematical ſciences were not powerful
enough to give the modes of calculation, when the principles of
the theory were firſt digefted.
There are problems, however, in which we cannot easily
come into poſſeſſion of data on which many will agree. The
fſimple queſtion of independent witneſſes is not one of them :
but the queſtion of collufion is. One of the difficulties is as
follows. We cannot inſtitute independent hypotheſes upon the
goodness of the witneſſes and the probability of their having con-
ferred upon their evidence. They declare, expreſſly or by im-
plication, that they have not done ſo : if they have, there is falſe-
hood in one part of their evidence ; or, which makes the difficulty
ſtill greater, there may have been general, but (as they affert or
imply) not particular conference : they may have been biaffed by
 On Probability. 199
each other, without knowing how or to what extent. The firſt
ſtep in one view of the problem is easily made, as follows .
Let u be the value of the evidence of each witneſs, m their
number, n the number of aſſertions they have power to chooſe
from, all as before. Let a be the probability that there has been
particular conference between them. There are then four caſes
to which the problem is restricted :- ( 1 ) they have conferred and
agreed to ſpeak truth ; (2) they have not conferred and all ſpeak
truth ; ( 3) they have conferred and agreed on a falſehood ; (4)
they have not conferred and have all lighted upon the fame falfe-
hood. The à priori probabilities of theſe four cafes are
m
(1 - λ)( 1 -μ)
λμ , (1 -λ)μ", λ ( 1 -μ) , :
(n- 1)m-1
and the odds that they ſpeak the truth (ſuppoſing n ſo great that
we may reject the fourth cafe) are µm to (1 -μ) . Now
comes the practical difficulty of this question ;-How are a and
μ to be connected ? Every caſe which is worth examining fup-
poſes that the greater the chance of there having been particular
conference, the leſs is the witneſs worth from that very circum-
ſtance. For it is to be remembered that we are not generally
able to give the witneſs a character wholly independent of his
evidence in the caſe before us ; in hiſtorical queſtions, for in-
ſtance, it frequently happens that we have nothing but the wit-
neſſes to try * the cafe by, and nothing but the cafe to try the
witneſſes by. A very common occurrence is this ;-that a cafe
is one in which no one would throw any doubt upon the wit-
neſſes, except for fufpicion of conference, and just as much doubt
as there is fufpicion of conference. This makes μ = 1 - λ, and
gives ( 1 - λ) : ^ +1 for the odds in favour of the affertion. On this
ſuppoſition, it follows that whenever the chances are against all
the witneſſes having conferred particularly, their number, if great
enough, ought to give any degree of credibility to the aſſertion.
* This gives riſe to two great tendencies, which very nearly divide the
world among them. Some fettle the cafe in their own minds, and then try
the witneſſes : ſome ſettle the witneſſes and then try the cafe : not a few
bring their ſecond refult back again to juſtify their firſt aſſumption. When
there are two unknown quantities with only one equation, it is eaſy for thoſe
who will affume either to find the other. But the difficulty is to find the
moſt probable value of both .
200
On Probability.
Problem 2. Let there be any number of different affertions, of
which one must be true, and only one : or of which one may be
true, and not more than one : or of which any given number
may be true, but not more : required the probability of any one
poffible cafe.
The ſolution of all theſe varieties depends on one principle,
explained in page 190 ; requiring the previous probabilities of all
the confiftent cafes to be compared. As an inſtance, ſuppoſe
four affertions, A,B,C,D, and ſuppoſe μ,ν,ρ,σ, to be the probabi-
lities from teſtimony, for each of them. If either of them have
ſeveral teftimonies, their united force must be ascertained by the
laſt problem. Firſt, let it be that one of them must be true,
and one only. The probabilities in favour of A,B,C,D, are in
the proportion of μν'ρ'σ', νμ'ρ'σ', ρμ'ν'σ', and σμ'ν'ρ' . Either of
theſe, divided by the ſum of all, represents the probability of its
cafe. Secondly, let it be that one of them only can be true, and
all may be falſe. Put on the fifth quantity μ'ν'ρ'σ' , for the caſe
in which all are falſe. For example, there are four distinct
aſſertions, not more than one of which can be true. The ſepa-
rate evidences for theſe four aſſertions give them the probabili-
ties 3 and . There is a certain aſſertion which is true if
either of the firſt three be true : required the probability of that
aſſertion. Here, neglecting the common denominator, which
is 7 × 11 × 8 × 5 in every cafe, the probabilities of the ſeveral
aſſertions, and that of all being falſe, are as 2.8.7.1, 3.5.7.1,
1.5.8.1, 4.5.8.7, and 5.8.7.1 , or as 112, 105, 40, 1120, and
280. The odds for one of the firſt three caſes against one of
the other two are 112 + 105 + 40 to 1120 + 280 or as 257 to
1400 ; or it is 1400 to 257 against the truth of the aſſertion.
Suppoſe the condition were that two of the affertions, but not
more, may be true, and that one must be true. Then the pof-
ſible caſes, meaning by an accent that the aſſertion is not true,
are AB'C'D', BA'C'D', CA'B'D', DA'B'C', ABC'D', ACB'D',
ADB'C', BCA'D', BDA'C', CDA'B'. Conſequently, the pro-
babilities of theſe caſes are in the proportion of μν'ρ'σ', υμ'ρ'σ' ,
ρμ'ν'σ' , &c. And the odds in favour of, ſay A, being true, are as
the ſum of all the terms which contain u, to the ſum of thoſe
which contain μ' .
When we wiſh to ſignify that no evidence is offered either for
 On Probability. 201

or against one of the aſſertions, we must put it down as having

the teſtimony . To put down o in the place of would be to
make an infallible witneſs declare that it is not true. Suppoſe
there are four aſſertions, one of which must be true and one
only : evidence of goodness is offered for the firſt, and none
either way for the others. Required the probability of the firſt.
The probabilities of the four aſſertions are in the proportion of
4.1.1.1, 1.3.1.1 , 1.3.1.1, and 1.3.1.1, and it is 4 to 9 for the
firſt, or 9 to 4 against it.
Problem 3. Arguments being ſuppoſed logically good, and the
probabilities of their proving their concluſions (that is, of all
their premiſes being true) being called their validities, let there
be a conclufion for which a number of arguments are preſented,
of validities a, b, c, &c. Required the probability that the con-
cluſion is proved.
This problem differs from thoſe which precede in a material
point. Teſtimonies are all true together or all falſe together :
but one of the arguments may be perfectly found, though all the
reft be prepofterous. The queſtion then is, what is the chance
that one or more of the arguments proves its conclufion. That
all ſhall fail, the probability is a'b'c' .... that all ſhall not fail, the
probability is 1 - a'b'c' .... Accordingly, if we ſuppoſe n equal
arguments, each of validity a, the probability that the conclufion
is proved is 1- ( 1 -a)" . And, as in page 187, ifthe odds against
each argument be k to 1 , then, the number of fuch arguments
being as much ask, the conclufion is rendered as likely as not.
But are we really to believe, having arguments against the
validity of each of which it is 10 to 1, that ſeven ſuch arguments
make the conclufion about as likely to be true as not. If ſuch
be the cafe, the theory, uſually ſo accordant with common notions,
is ſtrangely at variance with them. This point will require ſome
further confideration.
In this problem I confider only argument, and not teftimony,
which, nevertheless, cannot be finally excluded (fee page 194).
If the conclufion be one on which our minds are wholly un-
biaſſed to begin with, it may ſeem that we have no eſcape from
the preceding reſult. And to it we must oppoſe, for confidera-
tion at least, the common opinion of mankind that ſtrong argu-
ments are the prefumption of truth, weak arguments of falſehood.
202
On Probability .
If a controverfialiſt were to bring forward a hundred arguments,
and if his opponent were ſo far to answer them as to make it ten
to one against each, there can be no doubt that the latter would
be confidered as having fairly contradicted the former.
We must not forget that argument, in a great many caſes, in-
volves and produces the effect of teftimony, and this in an easily
explicable and perfectly justifiable manner. If I were to pick up
a bit of paper in the ſtreets, on which an argument is written,
for a conclufion on which I have no previous opinion, and by an
unknown writer, and if I could say that that argument left on
my mind the impreſſion of ten to one againſt its validity, I might
be prepared to allow it to ſtand as giving of probability, and
upon that ſuppoſition to combine it with my previous opinion, ½,
as in the next problem. But ſuppoſe it is on a queſtion of
phyſics, and Newton is the propoſer of it, and that it is his only
argument, and therefore, I conclude, his best. The caſe is now
entirely altered : poffibly the conclufion is one on which the
following argument would have great probability : ' If this con-
cluſion were true, it could be proved ; if it could be proved,
Newton could have proved it ; therefore if it were true,
Newton could have proved it : but Newton cannot prove it ;
therefore it is not true.' If the cafe be ſuch that the two pre-
miſes of this laſt argument have each 9 to 1 for it, or ; then,
though the original argument give TT for the concluſion, the mere
circumſtance of Newton bringing this argument as his beſt is o
against it. If Newton at the ſame time declare his belief in the
conclufion, we have on one fide his argument and his authority,
on the other fide the argument arising from his being reduced
to ſuch an argument.
That ſuch confiderations have weight, we know : and that
they ought to have weight, we may easily fee. It is of courſe,
dependent upon the particular conclufion what weight ſhall be
attached to the aſſertion, if this conclufion were true it could
be proved. ' The courts of law conſtantly act upon this princi-
ple. They confider (very juſtly I think) that evidence, however
good it may be, is much lowered by not being the beſt evidence
that could be brought forward. If a man be alive, and capable
of being produced with ſufficient eaſe, they will not take any
number of good witneſſes to the fact of his having been very
 On Probability . 203
recently alive. In enumerating the arguments, then, for or
againſt a propofition, thoſe must be included, if any, which ariſe
out of the nature, mode of production, or producers, of any
among them. And until this has been properly done, we are
not in a condition to apply the methods of the preſent chapter.
Problem 4. A conclufion and its contradiction being produced,
one or the other of which must be true, and arguments being
produced on both fides, required the probability that the conclu-
ſion is proved, diſproved (i. e. the contradiction proved), or left
neither proved nor diſproved.
Collect all the arguments for the conclufion, as in the laft
problem, and let a be the probability that one or more of them
prove the conclufion. Similarly, let b be the probability that one
or more of the oppoſite arguments prove the contradiction. Both
theſe cafes cannot be true, though both may be falſe. The pro-
babilities of the different caſes are thus derived. Either the
conclufion is proved, and the contradiction not proved, or the
conclufion not proved and the contradiction proved, or both are
left unproved. The probabilities for theſe caſes are as a ( 1 - b ),
b( 1 - a) and ( 1 - a)(1 - b), and the probability that the conclufion
is proved is a( 1-6) divided by the ſum of the three, and ſo on .
The fraction ( 1 - a)( 1-6) divided by this ſum may be called the
inconclufiveness of the combined arguments. The manner in
which this inconclufiveneſs is to be diftributed between the hypo-
theſis of the truth and falſehood of the conclufion must depend
upon teftimony, in the complete ſenſe of the word.
The predominance of one fide or the other, as far as argu-
ments only are concerned, depends on which is the greatest,
a( 1 - b) or b( 1 - a), or fimply on which is the greatest, a or b .
Ifthe arguments on both fides be very ſtrong, or a and b both very
near to unity, then, though a( 1 - b) and b( 1 - a) are both ſmall,
yet ( 1 - a)( 1 - b) is very ſmall compared with either. The ratio
of a ( 1 - b ) to b( 1 - a) on which the degree of predominance de-
pends, may, conſiſtently with this ſuppoſition, be anything what--
ever. But we cannot pretend that, when oppoſite ſides are thus
both nearly demonftrated, the mind can take cognizance of the
predominance which depends upon the ratio of the ſmall and
imperceptible defects from abſolute certainty. The neceſſary
conſequence is, that the arguments are evenly balanced, and are
204 On Probability .
as if they were equal : there is no ſenſible notion of predominance.
This is the ſtate to which moſt well conducted oppofitions of
argument bring a good many of their followers. They are fairly
outwitted by both fides, and unable to anſwer either, and the
conclufion to which they come is determined by their own pre-
vious impreſſions, and by the authorities to which they attach
moſt weight ; and theſe are, of course, thoſe which favour their
own previoufly adopted fide of the queſtion.
When no argument is produced on one fide of the queſtion,
the cafe is very different from the caſe of the preceding problems,
in which no teftimony is produced. Here the question is, ' Has
the conclufion been proved or not proved ; ' and when no argu-
ment is produced, we are certain it has not been proved. Ас-
cordingly, if no argument were urged for the contradiction, we
ſhould have 1 - b = 1 , or b = 0.
If, in the preceding problem, the two fides of the queſtion be
not contradictions, but fubcontradictions, of which neither need
be true, but both cannot be, the problem is ſolved in the ſame
way, for the caſes are just the fame. But we may introduce a
distinction which the former cafe would not admit. When one
muſt be true, every argument against one is of equal force for
the other ; which is not the cafe when neither need be true.
Let there, then, be arguments for the firſt concluſion and against
it, and let a and p be the probabilities that one or more of the
arguments for, prove it, or againſt, diſprove it. Let band q be
the fimilar probabilities for the ſecond conclufion. Then, there
are theſe cafes :-1 . The arguments (or ſome of them) for the
firſt are valid, against it invalid, and thoſe for the ſecond are
invalid (it matters nothing whether thoſe against the ſecond be
valid or invalid). 2. The arguments for the firſt are invalid,
thoſe for the ſecond valid, and againſt it invalid. 3. The argu-
ments against the firſt are valid, and thoſe for it invalid. 4. The
arguments against the ſecond are valid, and thoſe for it invalid.
5. All the arguments are invalid. Accordingly, the probabilities
that the firſt is proved, that it is diſproved, that the ſecond is
proved, that it is diſproved, and that neither of the two is proved
nor diſproved, are in the proportion of a( 1 -p)( 1 − b), (1 - a)p,
b( 1 - q)( 1 - a), ( 1 - b)q, and ( 1 - a)( 1 - b)( 1 - p )( 1-9).
Problem 5. Given both teftimony and argument to both fides
 On Probability . 205
of a contradiction, one ſide of which must be true, required the
probability of the truth of each ſide.
This is the most important of our cafes, as repreſenting all
ordinary controversy. Collect all the teftimonies, and let their
united force for the firſt ſide be µ, and, from the nature of this
caſe, 1 - μ for the other fide. Let a and b be the probabilities
that the firſt ſide and the ſecond ſide are proved by one or more
of the arguments in their favour. Now, obſerve that, for the
truth of either ſide, it is not eſſential that the argument for it
ſhould be valid, but only that the argument against it ſhould be
invalid. Accordingly, the probabilities of the two fides are in
the proportion of μ ( 1 − b) and ( 1-4) ( 1 - a), and the probabili-
ties of the two fides are repreſented by
μ( 1-6) (1 -μ)(1 -a)
μ( 1-6) + ( 1 - μ)(1-a) μ( 1-6) + ( 1 - μ)( 1 - a)

Firſt, let there be no teſtimony either way : we must then have

μ== I - μ ; conſequently, theſe probabilities are as I - b to
I- a . Let no argument have been offered for the ſecond ſide,
or let b = 0. Then we have I to 1 - a, for the odds, or
1÷( 2 - a) for the probability of the firſt ſide being true. It has
been uſual to ſay that if an argument be preſented of which the
probability is a, the truth of the concluſion has alſo the probabi-
lity a. Probably the above was the caſe intended as to teftimony,
&c., and the probability ſhould then have been
I
or a+
(1-a)2
I

2- a 2- a

which is always greater than a. Or, as we might expect, the

poſſibility of the conclufion being true, though the argument
ſhould be invalid, always adds ſomething to the probability of its
being true. Moreover, 1÷( 2 - a) is always greater than : or
any argument, however weak, adds ſomething to the force of
the previous probability. The ſame thing is true in every caſe.
Suppoſe a new argument to be produced for the firſt ſide, of the
force k. The effect upon the formula is to change I - a into
( 1 - a) ( 1 - k), and the odds in favour of the conclufion are in-
creaſed in the proportion of I to 1 -k. But this is to be under
206 On Probability .
ſtood ſtrictly in the ſenſe deſcribed in page 202, namely, we are
to ſuppoſe that the newly produced argument is ſingle, that is,
does not by the circumstances of its production cauſe itſelf to be
accompanied by an argument for the ſecond ſide, or against the
firſt. If this laſt ſhould happen, and the argument thus created
for the ſecond ſide have the force 1, the odds are altered in the
proportion of 1-1 to 1 - k.
From the above it appears that oppoſite arguments of the
force a and b are exactly equivalent to a teftimony the odds for
the truth of which are as 1 - b to 1 - a. Thus, ſuppoſe we have
for a concluſion witneſſes whoſe teftimonies are worth ,
; arguments for of the ſeveral forces,,, ; and arguments
againft of the forces , ㅜㅜ, . Writing numerators only, we put
down
For, 2, 2, 4, 9 ; 7, 9 , 1:
Against, 1, 1 , 3, 1 ; 4, 1, 3.
Hence it is, 2. 2. 4. 9. 7. 9. I to I. I. 3. 1. 4. 1. 3 , or
252 to 1 for the conclufion.

An argument, we ſhould infer beforehand, is better than a

teftimony of the fame force ; for the failure of the argument is
nothing against the conclufion, but the failure of the teftimony is
its overthrow. So ſays the formula alſo : the introduction of a
teftimony of the value k, not before received, alters the exiſting
odds in the proportion of k to 1 - k : but the introduction of an
argument of the ſame force alters them in the greater proportion
of I to I - k. Thus, the introduction of the teftimony of a
perſon who is as often wrong as right ( 1) alters the odds in the
proportion of I to I, or does not alter them at all : but the intro-
duction of an argument which is as likely as not to prove the
conclufion, alters them in the proportion of I to I- , or of
2 to I.

Are we not in the habit, unconsciously, of recognizing ſome

ſuch distinction ? Do we not give much more weight to argu-
ment than to teſtimony ? I ſuſpect the anſwer ſhould be in the
affirmative : that an argument of 3 to I does convince us much
more than a teftimony of 3 to 1. I ſuſpect we ſhow it, not in
numerical appreciation, of courſe, but in liftening to and allow
 On Probability. 207
ing weight to arguments, when we ſhould refuſe teftimony of
the fame character .
It may be doubted, however, whether we have much ſcope
for experiment on the lower degrees either of teftimony or argu-
ment. Perhaps it is not often we meet a witneſs, whether as
bearing teftimony of veracity to a fact, or of judgment to a con-
cluſion, whoſe evidence is as low as ; and the ſame perhaps of
an argument.
I have ſpoken, in the previous part of this chapter, of the
rejection of authority, that is, of teftimony, authority being only
high teſtimony. Let us now examine by the formula and fee
what it amounts to. Let a be the probability that the argument
proves its concluſion : and let us therefore perſiſt in ſaying that
a is the probability for the truth of the conclufion. In the for-
mula, b being=0, let u be made a÷( 1 + a), it will be found that
the probability for the conclufion, μ divided by μ + (1 - μ)(1 - a),
comes out a, as required. Conſequently, in the cafe of a fingle
argument, the total rejection, as it would be thought, of all tefti-
mony, is really equivalent to accompanying every argument by a
teſtimony leſs than , depending upon its own force. It is to
declare that, by the laws of thought, an argument of o is of its
own nature accompanied by a witneſs of 17, one of by a wit-
neſs of , and ſo on ; this is clearly not what was meant. Nor,
I ſuppoſe, can it be meant that we are arbitrarily to ſtart with the
testimony , and to reduce our own evidence, and that of all
others, to the fame. If there be any ſenſe in which the rejection
of authority is defenſible, it must be when we are required to
proceed as if we were in perfect ignorance what the value of the
authority is. We cannot ſuppoſe it to be as likely to have one value
as another. Suppoſe, for instance, that the arguments have un-
known propoſers : we cannot treat their authorities as ifthey were
juſt as likely to be excefſſively high or low as to be very near
to none at all. The more rational ſuppoſition is that the autho-
rity ſhould be more likely to be ſmall than great, as likely to be
againſt as for, and very unlikely to be exceſſively great either for
..or againft. I cannot here enter into the mode in which ſuch an
hypothefis can be expreſſed or uſed : but the reſult of the ſimpleft
formula which ſatisfies the above conditions, is as follows : -
Let r= ( 1-6 )÷( 1 - a ), b and a meaning as above ; then the
208 On Probability .
probability that the conclufion is true, which has a for the
validity of its argument, &c. is
r(r³-6r2 + 3r + 6rlogr + 2)÷(r - 1 )*
where logr means the Naperian logarithm (99-43rds of the
common logarithm will be near enough for the preſent purpoſe).
If, for instance, r= 2 , which, on the ſuppoſition of no previous
balance of teftimony, would give 2 to 1 for the conclufion, the
formula juſt written gives 636, or 636 to 364, ſomething leſs
than 2 to 1 .
In the cafe firſt diſcuſſed in page 202, it may be thought that
the weakness of a propoſed argument, from one who ſhould have
brought a better, if there had been one, may be confidered as a
testimony against the conclufion rather than an argument. Sup-
I
poſe his argument, for instance, to have only the probability .
He tells us then, that after he has done his best, it is 9 to 1
against the propofition being proved. If we are very confident
that it could be proved, if true, and that he could do it, if any
one, he comes before us as a teftimony of 9 to I against the
truth of the conclufion, or very nearly ſo. If we take, then, all
that his argument wants of demonſtration, as ſo much evidence
from him against the conclufion, this amounts to ſuppoſing that,
a being the validity of his argument, a is alſo his teftimony for
the conclufion (and 1- a that against it). If there be only argu-
ment for, and none againft, and if our minds be previously unbi-
afled, we represent this cafe by putting a for u in the formula,
and the odds for the conclufion are then as a to ( 1 - a)². On
this ſuppoſition, which I incline to think well worthy of attention,
we ſhould not confider an unoppoſed argument from an acute
reaſoner as giving the concluſion to be as likely as not, unleſs
a = ( 1 - a ) or a= 382, a little more than 1. Were it not for
our peculiar introduction of teſtimony, then, the conclufion being
as likely as not to begin with, an argument which has any pro-
bability of proving it, would have made it more likely than not,
as before ſeen .

But that the introduced teſtimony ſhould be exactly as above,

is a mere ſuppoſition. If it were a mathematical propofition,
for instance, and Euler were to declare himſelf unable to give
more than a probability of proof, I, for one, ſhould confider him
 On Probability. 209
as giving a much higher rate of teftimony against the truth of
the aſſertion than is ſuppoſed in the preceding. But all this has
reference to the queſtion how to meaſure teftimonies and va-
lidities in particular caſes, which is quite a diſtinct thing from
the inveſtigation of the way to uſe them when meaſured.
In caſes in which the number of arguments is multiplied, it
generally happens that they ſtand or fall together, in parcels :
namely, that the ſame failure which makes one invalid, neceſſarily
makes others invalid. In this caſe, independent arguments muſt
be ſelected, and the probabilities for them alone employed.
We ſee in this problem an illuſtration of the commonly ob-
ſerved reſult, that the ſame argument produces very different
final conclufions in two different minds ; and this when, ſo far as
can be judged, both are diſpoſed to give the ſame probabilities to
the ſeveral premiſes of the argument. The initial odds, come
how they may, or u to 1 - μ, ſhould be altered by the arguments
in the proportion of 1 - b to 1- a. Accordingly, b and a being
the ſame to both parties, their belief in the concluſion may have
any kind of difference, if u be not the ſame thing to both.
Problem 6. Given an aſſertion, A, which has the probability
a ; what does that probability become, when it is made known
that there is the probability m that B is a neceſſary conſequence
of A, B having the probability b ? And what does the probabi-
lity of B then become ?
Firſt, let A and B not be inconfiftent. The cafes are now as
follows, with reſpect to A. Either A is true, and it is not true
that both the connexion exiſts and B is falſe : or A is falſe. This
is much too conciſe a ſtatement for the beginner, except when it
is ſuppoſed left to him to verify it by collecting all the caſes. The
odds for the truth of A, either as above or by the collection,
are a { 1 - m( 1 - b )} to 1 - a. As to B, either B is true, or B is
falſe and it is not true that A and the connexion are both true.
Accordingly, the odds for B are as b to ( 1-6)( 1 - ma) .
The reader muſt remember that when B neceſſarily follows
from A, B muſt be true when A is true, but may be true when
A is falſe ; while A must be falſe when B is falſe. And now we
ſee that a propoſition is not neceſſarily unlikely, becauſe it is very
likely to lead to an incredibility, or even to an abſolute impoſſi-
bility. Let b = 0, or let B be impoffible : then the odds for A
P
 210
On Probability .
are as a( 1 - m) to I- a. Say that it is 9 to I that the connec-
*tion exiſts ; then theſe odds are as a to 10(1 - a) . If a be
greater than i, ſtill A remains more likely than not, even when
it is 9 to I that it leads to the abſurdity B.
Secondly, let A and B be inconfiftent, ſo that both cannot be
true. Either then A is true, B falſe, and the connexion does
not exist ; or A is falſe. The odds for A are then as a( 1-6)
(1 -m) to I - а. With reſpect to B, either B is true and A is
falſe, or B is falſe, and A and the connexion are not both true.
The odds for B are then as b( 1 - a) to ( 1 - b)( 1 -ma) .
1

Among the early ſophiſms with which the Greeks tried the
power of logic, as a formal mode of detecting fallacies, was the
conſtruction of what we may call fuicidal propofitions, aſſertions
the truth of which would be their own falſehood. If a man
ſhould say ' I lie,' he ſpeaks neither truth nor falſehood ; for if
he ſay true, he lies, and if he lie, he ſpeaks truth. Such a ſpeech
cannot be interpreted. Again, the Cretan, Epimenides, faid that
all the Cretans were incredible liars ; is he to be believed or not ?
If we believe him, we must, he being a Cretan, diſbelieve him .
Some ſtated it thus ;- If we believe him, then the Cretans are
liars, and we ſhould not believe him ; then there is no evidence
against the Cretans, or we may believe him, ſo that the evidence
against the Cretans revives, &c. &c. &c. Refer ſuch a propo-
ſition to the theory of probabilities, and the difficulty immediately
diſappears . Whatever the credit of Epimenides as a witneſs
may be, that is, whatever, upon his word, the odds may be for his
propoſition, the fame odds are there against him from the propo-
ſition itſelf. Theſe equal conflicting teftimonies balance one
another (problem 1) and leave the effect of other teftimonies to
the ſame point unaltered. The ſophiſm of Epimenides, as ſtated,
is but an extreme caſe of the ſecond of the problems before us .
The propofition B is inconfiftent with A, and the connexion is
certain (m = 1 ) : the odds for B muſt then be as b( 1 -a) to
(1 - b)( 1-a), or as b to I - b, exactly what they are independ-
ently of the previous aſſertion.
 On Induction. 211

CHAPTER XI .

On Induction .

THE theory of what is now called induction muſt occupy a

large ſpace in every work which profeſſes to treat of the
matter of arguments ; but there is not much to ſay upon the gen-
uine meaning of the word, in any ſyſtem offormal logic. And that
little would be leſs, if it were not for the miſtaken oppofition
which it has long been cuſtomary to confider as exifting between
the inductive proceſs and the rest of our ſubject .
By induction (ἐπαγωγή) is meant the inference of a univerſal
propofition by the ſeparate inference of all the particulars of
which it is compoſed : whether theſe particulars deſcend ſo low
as ſingle inſtances or not. Thus if X be a name which includes
P,Q,R, ſo that every thing which is X must be one of the
three : then if it be ſhown ſeparately that every P is Y, and that
every Q is Y, and that every Ris Y ; it follows that every X
is Y. And this laſt is ſaid to be proved by induction. Thus
(Chapter VI) .
X)P, Q, R + P)Y + Q) Y + R)Y = X) Y
is an inductive proceſs. In form, it may be reduced as in page
123, to one ordinary fyllogifm.
Complete induction is demonſtration, and ſtrictly fyllogiftic in
its character. In the preceding proceſs we have y)p, y)q, y)r,
which give y)pqr : and X)P,Q,R is pqr)x ; whence y)x, or X)Y.
It is a queſtion of names, that is, it depends upon the existence
or nonexistence of names, whether a complete induction ſhall
preſerve that form, or loſe it in the appearance of a Barbara fyllo-
giſm, formed by help of the conjunctive poftulate of Chapter VI.
But when the number of ſpecies or inſtances contained under
a name X is above enumeration, and it is therefore practically
impoſſible to collect and examine all the caſes, the final induc-
tion, that is, the ſtatement of a univerſal from its particulars,
becomes impoffible, except as a probable ſtatement : unleſs it
ſhould happen that we can detect ſome law connecting the ſpe-
cies or instances, by which the reſult, when obtained as to a
certain number, may be inferred as to the reft.
212 On Induction .
This last named kind of induction by connexion, is common
enough in mathematics, but can hardly occur in any other kind
of knowledge. In an innumerable ſeries of propoſitions, repre-
ſented by P1,P2,P3,P4, &c, it may and does happen that means
will exiſt of ſhowing that when any conſecutive number, ſuppoſe
three, of them are true, the next must be true. When this
happens, a formal induction may be made, as soon as the three
firſt are eſtabliſhed. For by the law of connexion, P1,P2, and P3,
eſtabliſh P4 ; but P2,P3, and P4, eſtabliſh Ps ; and then P3,P4,
and P5, eſtabliſh P.; and ſo on ad infinitum. It is to be obſerved
that this is really induction : there is no way, in this proceſs, of
compelling an opponent to admit the truth of P100 without
forcing him, if he decline to admit it otherwise, through all the
previous caſes.
As an eaſy inſtance, obſerve the proof that the fquare of any
number is equal to the ſum ofas many conſecutive odd numbers,
beginning with unity, as there are units in that number : as ſeen
in
6 × 6 = 1 + 3 + 5 + 7 + 9 + 11

Take any number, n ; and write n ns (repreſenting a unit by

adot) in rank and file. To enlarge this figure into (n + 1 ) (n + 1)s,
we must place n more dots at each of two adjacent fides, and
one more at the corner. So that the ſquare of n is turned into
the ſquare of n + 1 by adding 2n + 1 , which is the (n + 1 )th odd
number. Thus 100 × 100 is turned into 101X 101 by adding
the 101ſt odd number, or 201. If then the theorem alleged be
true of n x n, it is therefore true of (n + 1 ) x (n + 1). But it is
true of the firſt number, 1 x I being I ; therefore it is true ofthe
ſecond, or 2x2 = 1 + 3 ; therefore it is true of the third, or
3 × 3 = 1 + 3 + 5 ; and ſo on .
But when we can neither examine every caſe, nor frame a
method of connecting one cafe with another, no abſolutely de-
monſtrative induction can exiſt. That which is usually called by .
the name is the declaration of a univerſal truth from the enumer-
ation of ſome particulars, being the aſſumption that the unex-
amined particulars will agree with those which have been ex-
amined, in every point in which thoſe which have been examined
agree with one another. The reſult thus obtained is one of
 On Induction. 213
probability ; and though a moral certainty, or an unimpeachably
high degree of probability, can easily be obtained, and actually is
obtained, and though most of our concluſions with reſpect to the
external world are really thus obtained, yet it is an error to put
the reſult of ſuch an induction in the ſame claſs with that ofa de-
monstration. There is no objection whatever to any one ſaying
that the former reſults are to his mind more certain than thoſe of
the latter : the fact may be that they are ſo. The difference
between neceſſary and contingent propoſitions lies in the quali-
ties from which they receive thoſe adjectives, more than in
difference of credibility. I know that a stone will fall to the
ground, when let go : and I know that a ſquare number must be
equal to the ſum of the odd numbers, as above : and though,
when I ſtop to think, I do become ſenſible of more aſſurance for
the ſecond than for the firſt, yet it is only on reflection that I
can diftinguiſh the certainty from that which is ſo near to it.
The rule of probability of a pure induction is easily given.
Suppoſing the fſimple queſtion to be whether X is or is not Y,
there being no previous circumſtances whatsoever to make us
think that any one X is more likely than not to be Y, or leſs
likely than not. Theſe are the circumstances of what I call a
pure induction. To begin with, it is I to I that the firſt X ex-
amined ſhall be a Y : if this be done, and X₁ be a Y, then it is
2 to 1 that X₂ ſhall be a Y ; ſhould it ſo happen, then it is 3 to
I that X3 ſhall be a Y. Generally, when the firſt m Xs have
all been examined, and all turn out to be Ys, it is m + I to I
that the (m + 1 )th X ſhall be a Y.
The fimplicity of this rule must not lead the ſtudent to ſuppoſe
he can find a ſimple reaſon for it. Let 10 Xs have been exam-
ined and found to be Ys : what do we affert when we ſay it is
II to I that the 11th X ſhall be a Y ? We aſſert that if an in-
finite number of urns were collected, each having white balls and
black balls in infinite number but in a definite ratio, and ſo that
every poſſible ratio of white balls to black ones occurs once ; and
if every poſſible way of drawing eleven balls, the firſt ten of
which are white, were ſelected and put aſide : then, of thoſe put
afide, there are eleven in which the eleventh ball is white, for
one in which the eleventh ball is black. The reader will find
ſome difficulty in forming a diſtinct conception of this, and of
214 On Induction.
courſe will find it impoſſible to have any axiomatic perception of
the truth or falsehood of the reſult.
It may be worth while to ſhow that a ſuppoſition making fome
degree of approach to the preceding circumstances will give fome
approach to the reſult. Firſt, in lieu of an infinite number of
balls in each box, which is ſuppoſed only that withdrawal of a
definite number may not alter the ratio, let each ball drawn be
put back again, which will answer the fame purpoſe. Let there
be only ten urns with ten balls in each, of which let the firſt
have one white, the ſecond two white, &c. and the last all white.
The number of ways of drawing eleven white balls ſucceſſively
out of any one urn is the eleventh power of the number of white
balls in the urn : that of drawing ten white balls followed by one
black one is the tenth power of the number of white balls mul-
tiplied by the number of black ones. If we were to put together
all the firſt, and then all the ſecond, we ſhould find about 21
times as many ways of arriving at the firſt reſult (ten white, fol-
lowed by a white) as the ſecond (ten white followed by a black).
But if we now increaſed the number of urns, and took a hundred,
having one, two, &c. white balls, we ſhould find instead of 21 ,
a number much nearer to 11 ; and ſo on.
Accordingly, when without any previously formed bias, we
find that m Xs, ſucceſſively examined, are each of them a Y, we
ought then to believe it to be m + 1 to I that the next, or
(m + 1 )th X, will be a Y. And further, a being a fraction leſs
than unity, we have a right to ſay there is the probability 1 - am + 1
that the Xs make up the fraction a or more, of the Ys. Or
thus ;-if the fraction a be, ſay , and if m be 10 : then if the
10 firſt Xs be all Ys, the probability that or more of the Xs
are Ys is just that of drawing one or more black balls in II
drawings, from an urn in which of the balls are always white.
If, for example, the firſt 100 Xs were all Ys, it would be
found to be 1000 to I that 93 per cent, at least, of all the Xs
are Ys .
If as before, the firſt m Xs obſerved have all been Ys, and we
aſk what probability thence, and thence only, ariſes that the next
n Xs examined ſhall all be Ys, the answer is that the odds in fa-
vour of it are m + I to n, and against it n to m + I. No induc-
tion then, however extenſive, can by itſelf, afford much probability
 On Induction. 215
to a univerſal conclufion, if the number of inſtances to be exam-
ined be very great compared with thoſe which have been exam-
ined. If 100 inſtances have been examined, and 1000 remain, it
is 1000 to 101 againſt all the thouſand being as the hundred.
This reſult is at variance with all our notions ; and yet it is
demonftrably as rational as any other reſult of the theory. The
truth is, that our notions are not wholly formed on what I have
called the pure induction. In this it is ſuppoſed that we know no
reaſon to judge, except the mere mode of occurrence of the in-
duced inſtances. Accordingly, the probabilities ſhown by the
above rules are merely minima, which may be augmented by
other ſources of knowledge. For instance, the ſtrong belief,
founded upon the moſt extenfive previous induction, that pheno-
mena are regulated by uniform laws, makes the firſt inſtance ofa
new cafe, by itſelf, furniſh as ſtrong a preſumption as many in-
ſtances would do, independently of ſuch belief and reaſon for it.
With this however I have nothing farther to do, except to
obſerve that, in the language of many, induction is uſed in a ſenſe
very different from its original and logical one. It is made to
mean, not the collection of a univerſal from particulars, but the
mode of arrival at a common cauſe for varied, but fimilar, phe-
nomena. A great part of what is thus called induction confifts
in diſcovery of differences, not resemblances. Under this confuſed
uſe of language, the uſual theory is introduced, namely, that
Aristotle was oppoſed to all induction, that Bacon was oppoſed
to every thing elſe, that the whole world up to the time of Bacon
followed Ariftotle, that the former was the first who ſhowed the
way to oppoſe the latter, that each had a logic of his own, &c.
&c. The whole of this account abounds with miſtatements .
The admitted and ſufficiently ſtriking difference between the
philofophy of modern and ancient times, in all natural and mate-
rial branches of inquiry, is not ſo easily explained as by choofing
two men, one to bear all the blame, the other all the credit : nor
are Copernicus, Gilbert, Tycho Brahé, Galileo, and the other
predeceffors of the Novum Organum, deſtined to be always de-
prived of their proper rank.
What is now called induction, meaning the diſcovery of laws
from inſtances, and higher laws from lower ones, is beyond the
province of formal logic. Its inſtruments are induction properly
216 On Induction .
ſo called, ſeparation of apparently related, but really diftinct par-
ticulars (the neglect of which was far more hurtful to the old
philoſophy than a neglect of induction proper would have been,
even had it exiſted) mathematical deduction, ordinary logic, &c.
&c. &c. It is the use of the whole box of tools : and it would
be as abſurd to attempt it here, as to append a chapter on car-
pentry to a deſcription ofthe mode of cutting the teeth of a ſaw.
The proceſſes ofAristotle and of Bacon are equally thoſe which
we are in the habit of performing every day of our lives. But
ſome perform them well, and ſome ill. It is extraordinary that
there ſhould be ſuch diviſion of opinion on the queſtion whether
a careful analyſis ofthem, and ſtudy of the parts into which they
decompoſe, is of any uſe towards performing them well. On
this point, and on the character of Bacon's office in philoſophy,
a living writer, to whom I ſhould think it likely that many yet
unborn would owe their firſt notions of Bacon's writings, ex-
preſſes himself in a manner which I quote, and comment on at
length, as the beſt expoſition I can find, of a claſs of opinions
which is very prevalent, and, I fully believe, to the prejudice of
fober thought and accurate knowledge.

The vulgar notion about Bacon we take to be this, that he invented a

new method of arriving at truth, which method is called Induction, and that
he detected ſome fallacy in the ſyllogiſtic reaſoning which had been in vogue
before his time. This notion is about as well founded as that of the people
who, in the middle ages, imagined that Virgil was a great conjuror. Many
who are far too well informed to talk ſuch extravagant nonſenſe, entertain
what we think incorrect notions as to what Bacon really effected in this
matter .

The inductive method has been practiſed ever since the beginning of the
world, by every human being. It is conſtantly practiſed by the moſt igno-
rant clown, by the moſt thoughtleſs ſchoolboy, by the very child at the
breast . That method leads the clown to the concluſion that if he ſows
barley, he ſhall not reap wheat. By that method a ſchoolboy learns that a
cloudy day is the beſt for catching trout. The very infant, we imagine, is
led by induction to expect milk from his mother or nurſe, and none from
hisfather.
Not only is it not true that Bacon invented the inductive method ; but
it is not true that he was the firſt perſon who correctly analyſed that method
and explained its uſes. Aristotle had long before pointed out the abfurdity
of ſuppoſing that ſyllogiſtic reaſoning could ever conduct men to the diſco-
very of any new principle, had ſhown that fuch diſcoveries muſt be made by
 On Induction. 217
induction, and by induction alone, and had given the hiſtory of the inductive
proceſs, conciſely indeed, but with great perfpicuity and preciſion.
Again, we are not inclined to aſcribe much practical value to that analy-
fis of the inductive method which Bacon has given in the ſecond book of
the Novum Organum. It is indeed an elaborate and correct analyſis. But
it is an analyſis of that which we are all doing from morning to night, and
which we continue to do even in our dreams. A plain man finds his ſto-
mach out of order. He never heard Lord Bacon's name. But he proceeds
in the ſtricteſt conformity with the rules laid down in the ſecond book of
the Novum Organum, and ſatisfies himself that minced pies have done the
miſchief. " I eat minced pies on Monday and Wednesday, and I was kept
awake by indigeſtion all night." This is the comparentia ad intellectum in-
ftantiarum convenientium. " I did not eat any on Tuesday and Friday, and
I was quite well. " This is the comparentia inftantiarum in proximo quæ
natura data privantur. " I ate very ſparingly of them on Sunday, and was
very flightly indiſpoſed in the evening. But on Chriſtmas-day I almoſt
dined on them, and was ſo ill that I was in great danger. " This is the
comparentia inftantiarum fecundum magis et minus. " It cannot have been
the brandy which I took with them ; for I have drunk brandy daily for
years without being the worſe for it. " This is the rejectio naturarum. Our
invalid then proceeds to what is termed by Bacon the Vindemiatio, and pro-
nounces that minced pies do not agree with him.
We repeat that we diſpute neither the ingenuity nor the accuracy of the
theory contained in the ſecond book of the Novum Organum ; but we think
that Bacon greatly overrated its utility. We conceive that the inductive
proceſs, like many other proceſſes, is not likely to be better performed
merely becauſe men know how they perform it. William Tell would not
have been one whit more likely to cleave the apple if he had known that
his arrow would deſcribe a parabola under the influence of the attraction of
the earth. Captain Barclay would not have been more likely to walk a
thouſand miles in a thouſand hours, if he had known the place and name of
every muſcle in his legs. Monfieur Jourdain probably did not pronounce
Dand F more correctly after he had been appriſed that D is pronounced
by touching the teeth with the end of the tongue, and F by putting the
upper teeth on the lower lip. We cannot perceive that the ſtudy of gram-
mar makes the ſmalleſt difference in the ſpeech of people who have always
lived in good ſociety. Not one Londoner in ten thouſand can lay down
the proper rules for the use of will andshall. Yet not one Londoner in a
million ever miſplaces his will andshall. Dr. Robertſon could, undoubtedly,
have written a luminous diſſertation on the uſe of theſe words. Yet, even in
his latest work, he ſometimes miſplaced them ludicrouſly. No man uſes
figures of ſpeech with more propriety becauſe he knows that one figure of
ſpeech is called a metonymy, and another a ſynecdoche. A drayman in a
paſſion calls out ' You are a pretty fellow,' without ſuſpecting that he is
uttering irony, and that irony is one of the four primary tropes. The old
ſyſtems of rhetoric were never regarded by the moſt experienced and dif-
cerning judges as of any uſe for the purpose offorming an orator. "Ego
218 On Induction.
hanc vim intelligo" faid Cicero " eſſe in præceptis omnibus, non ut ea ſecuti
oratores eloquentiæ laudem fint adepti, ſed quæ ſua ſponte homines eloquen-
tes facerent, ea quofdam obſervaſſe, atque id egiſſe ; fic eſſe non eloquentiam
ex artificio, fed artificium ex eloquentia natum. " We muſt own that we
entertain the fame opinion concerning the ſtudy of Logic, which Cicero
entertained concerning the ſtudy of Rhetoric. A man of ſenſe ſyllogizes in
celarent and cefare all day long without ſuſpecting it : and though he may
not know what an ignoratio elenchi is, has no difficulty in expofing it when-
ever he falls in with it.-(' Lord Bacon,' in Critical and Historical Eſſays
contributed to the Edinburgh Review. By Thomas Babington Macaulay.)

This brilliant paſſage has, I have no doubt, appeared to many

completely deciſive of the queſtion which it affirms : and, as fo
often happens in like caſes, there is a certain exaggeration againft
which it is of truth. It is good against those who confound
analyſis and recombination of exifting materials with introduction
of them : and who might profeſs to ſee in agriculture ſomething
which would have benefited mankind, though plants and animals
had not been natural products of the foil. But I now proceed
to examine it, against those who affirm that Ariftotle and Bacon
are of no uſe, and who very frequently fall into the common
logical fallacy of ſuppoſing that their caſe is proved, as foon as it
is made out that they are not of all the uſe : which Mr. Macau-
lay himſelf has done, except as against the exaggerators aforeſaid.
We reaſon inductively from morning till night, and even in
our dreams. True : and how badly we often do it, particularly
in ſleep. A plain man is then produced, to reaſon on Bacon's
principles : and Mr. Macaulay has imitated a plain man better
than he intended, by making him do it wrongly. Look over
the induction, and it will appear that the cafe is not made out ;
an exclufion is wanting : it may have been the mixture of minced
pies and brandy which did the miſchief. The plain man ſhould
have tried minced pies without brandy ; but he had drunk the
latter daily for years, and it never ſtruck him. This is preciſely
one of the points in which we are moſt apt to deceive ourſelves,
and for which we moſt need to have recourſe to the complete-
neſs of a ſyſtem of rules ; ſomething is left taken for granted.
The things of courſe, our daily habits, are neglected in the
confideration ofanything of a leſs uſual character : the plain man
left off the minced pies upon trial ; but not the brandy : Chrift
 On Induction. 219
mas miſchief must be referred, he thinks, entirely to Chriſtmas
fare, if at all.
Buteven ifthis omiffion had been ſupplied, and the reſult found
to confirm the conclufion, yet the plain man has ſtopped where
the plain man frequently does ſtop, at what Bacon calls the
Vindemiatio prima, the rudiments of interpretation. Complete-
neſs is ſeldom anything but ſtudy and ſyſtem. Philofophy ought
to bring him to the reſult that daily brandy has made that ſpirit
ceaſe to give the ſtimulus which, were its uſe only occafional,
would enable his ſtomach to bear an unusually rich diet for a
ſhort time. Our plain friend is preciſely in the poſition of a
bankrupt who curſes the times, on reaſoning ſtrictly Baconian as
far as it goes, and forgets that a caſual tightneſs in the money
market would never have upſet him, if it had not been for the
previous years of extravagant living and raſh ſpeculation.
But there are many proceſſes which are not better performed
becauſe men know how they perform them." Mr. Macaulay
here means " becauſe men know the laws of that part of the
proceſs which nature does for them." That men ſhould not
know better how to perform for knowing how they perform
is almoſt a contradiction in terms . William Tell knew how to
ſhoot all the better for knowing which end of the arrow he was
accuſtomed to fit to the ſtring : had he wanted this knowledge,
his chance of cleaving the apple would have been much dimi-
niſhed. But he would not have been improved by knowing
that his arrow deſcribed a parabola. True, becauſe it did not do
ſo. The centre of gravity of the arrow would deſcribe a para-
bola, if it were not for the reſiſtance of the air ; or fomething fo
near it as to be undiftinguishable. But, taking the deſcription
as roughly correct, William Tell did know, inductively, that the
arrow deſcribes a curve, concave to the earth: and had made
thousands of experiments in connexion of the two ends of that
curve, which were all that he was concerned with. It is no ar-
gument against the ſtudy, as a ſtudy, of induction, that the amount
of useful reſult which it had recorded in the mind of William
Tell in the ſhape of habit, would not have been augmented by
deductive knowledge of an intermediateſtatus with which he had
nothing to do. But let knowledge advance, under both modes
of progreſs, and Tell becomes an artillery officer, the rude arrow
220 On Induction .

a truly ſhaped and balanced ball, means of meaſurement are ap-

plied, the true curve is more correctly repreſented than by the
parabola, and thirty pounds of iron are thrown to four times the
diſtance which an arrow ever reached, and with a certainty al-
moſt equal to that of the legend.
But if Captain Barclay had known the places and names of
the muscles, he would not have been more likely to walk a thou-
fand miles in a thousand hours. The inſtance is far fetched :
becauſe the feat conſiſted in the exhibition of power of endurance
acquired by practice. If my denial ſeem as far fetched, it is the
fault of the propoſer. Captain Barclay must, by habit, by in-
duction, have acquired facility in varying his pace and geſture
ſo as to eaſe the muscles. Had he been well acquainted with the
diſpoſition and uſes of theſe organs to begin with (towards which
knowledge of their places and names would have contributed) he
would have learnt this art more easily. Though not altogether
ad elenchum, yet I may ſay that in this caſe the effect of ſuch
knowledge would have been that he would have been less likely
to have performed the feat. Had he directed his attention to
fome ſcience of obſervation, he would not have needed to have
fought fame, or exhaustion of remarkable energy, in ſuch a tri-
fling purſuit. And further, in a very common caſe, mechanics
has taught what few ever learn by induction, though they have
conſtant opportunities of doing it : namely, that in walking, the
ordinary practice of ſwinging the arms is injurious and tiring ;
that a very trifling amount of it tells ſeriouſly in a long journey.
Here is one uſeful reſult, which natural induction does not com-
monly teach, and there may be many more of the fame kind :
the queſtion between it and regular ſtudy requires the confidera-
tion, not only of what is done, and whether it might be done
better, but of what is not done.
Next, M. Jourdain did not pronounce D and F more cor-
rectly after his attention had been called to the details of the act
of pronunciation. None but Molière ever knew whether he did
or not : but all who have watched the progreſs of inſtruction
know that the bad habits or natural imperfections of children are
removed or alleviated by making them practice mechanical pro-
nunciation, with perceptive adoption of rules. In every one of
a few detached inſtances in which I have ſeen children at their
 On Induction. 221

reading leſſons in France, I have noticed that a return upon the

habits of pronunciation is always a part of the exerciſe: and that
the letters are pronounced with that diſtinct effort which makes
the pupil ſenſible of the action required. I have always attri-
buted to this practice the more uniform ſtandard of pronunciation
which prevails among the educated French, as compared with
ourſelves .
But the ſtudy of grammar makes no difference in the ſpeech
ofpeople who have always lived in good ſociety. If Mr. Macau-
lay mean merely as to the use of shall and will, and the like, it
may certainly be ſaid that the perpetual uſe of ſpeech (which is
not reaſoning) does enable every one to form the habits of thoſe
about him. But that grammar, as a whole, produces no effect
upon the ſpeech ofgood ſociety, is one ſide of a balanced matter
of opinion. Many contend that it has produced, in our gene-
ration and the one above it, a very unfortunate effect : they aver
that the purity and character of our Engliſh has been deteriorated
by Lindley Murray and his ſchool, and that we much want better
grammar teaching. On the ſubject of shall and will, it is re-
markable that Mr. Macaulay, whom a vigorous faculty of illuf-
tration, combined with immenſe reading, enables to ſtrew his
path with inſtances, has to invent his caſe, and to refer to a
treatiſe which Robertſon could have written. But it is not
enough : if we grant that ſuch a treatiſe would have been lumi-
nous, we may be ſafe ; but would it have been correct ? And
further, knowledge muſt abdicate at once, if we pronounce uſe-
leſs all that has been clearly explained by thoſe who have not
rightly practiſed. Bacon himself might have taken exfors ipfa
fecandi for his motto.
Next, it is ſaid that no man uſes figures of ſpeech more cor-
rectly becauſe he knows that one is metonymy and anotherfynec-
doche. True ; and in like manner no man conſults his books
more easily becauſe he has a bookcaſe. But, having the book-
caſe, he arranges his books in it, and then he knows where to
find them. Mr. Macaulay dwells throughout upon nomencla-
ture. I might inſiſt upon its ſuperſtructure : but even mere
naming is useful, when the meaning of the name is clearly under-
ſtood. A mind well ſtocked with understood names cannot
keep itſelf from being conftantly in the act of claſſification,
222 On Induction .
which contains induction. The mere involuntary reference of
inſtance number two to inſtance number one, which is made
when we remember that the ſecond must have the same name as
the firſt, is compariſon and induction, leads to reflection, culti-
vates taſte, and gives power. The drayman, who calls out in a
paffion, " You are a pretty fellow ! " without knowing that he
is uttering irony, is an incomplete picture : there is omitted a
wiſh relative to the eyes of his opponent, and an adjective which
is (in ſuch quarrels) ſometimes prophetically, but ſeldom deſcrip-
tively, true. The value of the difference between this ſavage
irony and the more elegant form of it which is ſo pleaſing in the
deſcription of the plain man's induction quoted above, is not
within the comprehenfion of the drayman : the foundation of a
better mode of expreſſion than undiſciplined rhetoric furnishes, ſo
far as its adoption is matter of taſte, was laid by those who placed
irony among the primary tropes. Good taste is a reſult of com-
pariſons, which could not have been made without nomenclature.
Did Cicero declare that ſyſtems of rhetoric are not of any uſe ?
The very quotation appears to mean that theſe ſyſtems, præcepta,
have their power ; that men get them by obſervation, and put
them into practice. The eafecuti oratores refers to what was
done in the firſt inſtance, by the firſt eloquent men,fuâfponte.
Moſt truly does he ſay that the art of rhetoric is derived from
eloquence, and not vice verſa : moſt falſely, as far as can be
judged, does he ſeem to infinuate that it was all done at one
ſtep ; firſt, ſome one or more confummate orators, ſecondly, a
finiſhed ſyſtem, drawn from obſervation oftheir methods. Per-
haps he intended a particular reference to a certain orator then
nameleſs : the ſentence, thus conſtrued, contains nothing but
matter which Tully is likely enough to have whiſpered to
Cicero.
A ſyſtem is a tool, and it must be employed upon materials
which different men furnish from their different means . But
the coat must be cut according to the cloth, both in fize and
quality : no reproach to the ſciſſors, nor prejudice to their ſupe-
riority over the ſharpened wood of the ſavage, even though prac-
tice will enable him to uſe the latter better than any civilized
man who is not a tailor can uſe the former. The formation of
tools, mental or material, is a cyclical proceſs. The firſt iron
 On Induction. 223
was obtained by help of wood ; one of the firſt uſes of it was to
make better tools, to get more iron, with which better tools ſtill
were made, and ſo on. And in this way we may trace back any
art to natural tools, and to materials which are to be had for the
gathering. The aſſertion made by Mr. Macaulay, and many
others, that in logic only, of all the abſtract ſciences, our natural
means are as good as thoſe which reſult from diligent analyſis, is
one which terminates in an iſſue of fact. The inſtances given
are contained in the aſſertion that a man offenſeſyllogizes in cefare
and celarent all day long without ſuſpecting it, and though he
does not know what an ignoratio elenchi is, can always detect
it when he meets with it.
Mr. Macaulay begins with an indefinite term, a man offenfe :
and the clauſe is deficient in logical perfpicuity. Firſt, what is a
man offenfe ? I grant that I should doubt the ſenſe of a man who
could not make the inferences deſcribed by cefare and celarent.
But do men become men of ſenſe by nature, without education ?
if yes, I deny the aſſertion that men of ſenſe reaſon (correctly) in
cefare, &c. The man of ſenſe who is not educated is as likely
to affert that cefaro is all that can be obtained, or to invent the
formfefape, as the plain man to forget to try the mince pies
without brandy before he concludes. If no, then the affertion
is itſelf ignoratio elenchi : for the very queſtion is how to make
men of ſenſe ; can they not be, ceteris paribus, formed better and
fafter with ſtudy of logic than without : it being agreed on all
hands that this man of ſenſe is always a practical logician.
Next, a man of ſenſe reaſons, &c. without ſuſpecting it.
Suſpecting what ? that he is reaſoning, or that he is reaſoning in
cefare ? I ſuppoſe the latter : that is to ſay, I take it to be meant
that a man of ſenſe may (not must, for ſome Ariftotelians are men
of ſenſe) not know that the logicians call the form of reaſoning
he uſes cefare. This is easily granted : but what is it but the
celebrated ignoratio elenchi of Locke, who fancied that he raiſed
an objection against the pretenſions of the logicians, when he
declared he never could believe that God had made men only
two-legged, and left it to Aristotle to make them rational. No
one ever denied that men reaſoned before Aristotle, and would
have reaſoned ſtill if he had never lived .
Mr. Macaulay, probably without ſo much as a new application
224 On Induction.
to the inkſtand, after falling into the ignoratio elenchi, ſingles out
this very fallacy as the one which a man of ſenſe is ſure to detect.
But if there be a fallacy which is the ſtaple of paralogiſm, it is
this one. Delectat domi, for ordinary diſcuſſion (eſpecially after
dinner) is little elſe ; impedit foris, for three fourths of public
debate, from the Houſes of Parliament downwards, is made up
of it. A man who expoſes it in converſation is confidered a
tireſome, and if he do it often, an uncourteous perſon : he " has
no converſation," he " harps upon one ſubject," he " won't let
you ſpeak."
I have made the above comments upon a very marked paſſage
of an eminent writer, in preference to introducing their ſub-
ſtance as a diſſertation of my own, that I might have the advan-
tage of the reader ſeeing that I meet real arguments, instead of
my own verſion or ſelection. It would probably be difficult to
find a better concentration of the ſubſtance of the antagoniſt
views, with reſpect to the formal ſtudy of reaſoning, than is
contained in my quotation from Mr. Macaulay : and I may fafely
take his adoption of them as proof that theſe views yet require
the notice of a writer on logic.

There is one reſult of the theory of probabilities, cloſely con-

nected with induction proper, which it will be adviſable to notice
here.
When the ſyllogiſm is declared illegitimate, on account of
both premiſes being particular, a probable concluſion of great
ſtrength may be admitted in many cases. This must be the
more infiſted on, becauſe it is too common to attend to nothing
but the demonftrative ſyllogiſm, leaving all of which the con-
cluſions are only probable, however probable, entirely out of
view.
I take as the inſtance the ſyllogiſm, or imperfect fyllogifm,
Some Xs are Ys, ſome Zs are Ys, therefore there is ſome pro-
bability that ſome Xs are Zs.' If the number of Xs and Zs
together exceed the number of Ys (as in Chapter VIII) there is
a certainty that ſome Xs are Zs. Let us then ſuppoſe this is
not the cafe.
Let the whole number of Ys in exiſtence be n, and let m and
 On Induction. 225
n be the numbers of Xs and Zs which are among them. I
ſhall confider two distinct caſes :-Firſt, when the diſtribution
of the Xs and Zs among the Ys is utterly unknown ; ſecondly,
when their diftribution is that of contiguity, that is, when the Ys
being for fome reaſon arranged in a particular order, the Xs
which are Ys are ſucceſſive Ys, and the ſame of the Zs which
are Ys .

For the firſt caſe a very rough notion will do, confined to the
ſuppoſition that few Xs and Zs are mentioned, compared with
the whole number of Ys. When the Xs and Zs together make
a large proportion of the Ys in number, then, if we have no
reaſon for making them contiguous, or otherwiſe limiting the
equally probable arrangements, it may be faid to be a moral cer-
tainty that ſome Xs are Zs .
In the firſt caſe, if we divide 43 times the product of m and n
by 100 times n, it gives us a ſufficient notion (not large enough)
of the common logarithm of k, the odds in favour of fome
Xs being Zs being k to 1. Say there are 1000 Ys, and that
100 Xs are Ys and 100 Zs are Ys. Then 43 × 100 × 100
divided by 100 X 1000 is 4.3, which is the logarithm of 20,000.
It is then more than 20,000 to I that, in this cafe, one or
more Xs are Zs. A more exact rule is as follows. To 43mn
divided by Ioon add its hundredth part, and to the reſult add
ſuch a fraction of itſelf as m + n is of 2n. Thus 43mn÷ 100 %
being 4.3, which, with its hundredth part is 4.343, and m + n
(200) being the tenth part of 2n (or 2000), we add to 4.343 its
tenth part, giving 4.777, which is about the logarithm of 60,000,
ſtill under the mark. It is more than 60,000 to 1 that ſome Xs
are Zs. When the fractions are very ſmall, this rule is accurate
enough, if n be confiderable. Its reſult is, that if n be very con-
ſiderable, and if a perceptible fraction of the Ys be Xs, and a
perceptible fraction Zs, and if we really have no reason to make
the limitation of contiguity or the like, then we are justified in
treating it as a moral certainty that ſome Xs are Zs. But I ſuf-
pect the relation of contiguity, to which I now proceed, better
repreſents the actual ſtate of the caſe in ordinary argument.
When the Xs which are Ys are contiguous, and alſo the Zs
which are Ys, the probability that no Xs are Zs is the fraction
having the product of n - m - n + I and n - m- n + 2 for nu
 226 On Induction.
merator, and the product of n - m + 1 and n - n + 1 for denomi-
nator. Thus in the example above propoſed, 1000 Ys containing
among them 100 Xs and 100 Zs (each ſet contiguous) we have
801 × 802 for numerator and 901 × 901 for denominator. This
fraction is about 8-tenths ; ſo that it is now 8 to 2, or 4 to 1,
against any Xs being Zs.
In order to find the probability against the number of Xs
which are Zs exceeding k, add k to both the multipliers in the
numerator, which then become n - m- n + k + 1 and n - m -n
+k + 2. For example, there are 100 Ys, containing 30 Xs and
60 Zs (each ſet contiguouſly) : what is the chance against the
number of Xs which are Zs exceeding 10 ? The numerator is
21X22 : the denominator is 71 ×41 . This fraction is 462 by
2911 ; whence it is 462 to 2449 against, or 2449 to 462 (more
than 5 to 1 )for, the number of Xs which are Zs exceeding 10.
The chances, it is to be remembered, are all minima : ex-
cept when we mean that m Xs, and not more, are Ys, &c. Theſe
queſtions may ſerve to give ſome notion of the manner in which
arguments not logically concluſive, may be morally fo.
What is called circumstantial evidence is a ſpecies of induction
by probability. The thing required to be found has the marks
P,Q,R,S, &c.: this Y has the marks P,Q,R,S, &c.: there is then
a certain amount of circumſtantial evidence that this Y is the
thing we want to find. If it can be ſhown that there is but one
thing which has all theſe marks, then the circumſtantial evidence
is demonftrative. But if there were, ſay 100 Ys, of which 5
have the mark P, 5 the mark Q, &c., then having aſcertained
one Y which has all the marks, the question is, what chance is
there against another Y having them all: the ſame chance, at
leaſt, is there that the Y found is the one fought. Instead how-
ever, of attempting the problem in this way, which is never
reſorted to for want of data (I mean that the reſemblance which
the rough proceſſes of our minds bear to thoſe of the theory of
probabilities does not here exiſt) I take it as follows. If the
poſſeſſion of the mark P give a certain probability to the Y
found being that fought, it is as a witneſs whoſe teftimony has a
certain credibility. Similarly for Q,R,S, &c. Compound theſe
teftimonies, when known, by the rule in page 195, and the reſult
is the value of the circumſtantial evidence .
 227

CHAPTER XII .

On old Logical Terms.

N this chapter I propoſe to ſay ſomething on a few terms of
INthe old Logic, which though they keep their places inworks
on the ſubject, and have ſome of them paſſed into common lan-
guage, are very little uſed. They relate generally to the ſimple
notion, and the name by which it is expreſſed : and have little of
ſpecial reference, either to the propoſition or fyllogifm. They
are mostly derived from Ariftotle, whoſe incidental expreſſions
became or give riſe to technical terms, and whoſe ſingle ſentences
were amplified into chapters. And here, as in other places, I
have nothing to do with the degree of correctneſs with which
Aristotle's meaning was apprehended, nor even with how much
was drawn from Aristotle and how much added to him, but only
with the actual phrases and their uſual meaning.
The words logic and dialectics * are now uſually taken as
meaning the ſame thing : the old distinction is that dialectics is
the part of logic in which common and probable, but not necef-
ſary, principles, are uſed. But the distinction is neither clearly
laid down, nor faithfully adhered to, even by Aristotle himſelf.
The term (in this work always called name) was divided into
fimple and complex : the ſimple term was the mere name, the
complex term was what all moderns call the affirmative propofi-
tion. Thus man and run were ſimple terms : man runs, a com-
plex term. Later writers rejected this confufion : and divided
the acts of the mind confidered in logic into apprehenfion, judg-
ment, and discourse, taking cognizance of notions, propoſitions,
and arguments. The common meaning of the word discourse,
* Our language is capricious with regard to the uſe of fingular and plural
ofwords in ic : thus we have logic and dialectics, arithmetic and mathema-
tics, phyſic and phyſics for medicine and natural philoſophy. Some modern
writers are beginning to adhere uniformly to the ſingular, in which I cannot
follow them, for I am afraid an Engliſh ear would not bear with mathe-
matic as a ſubſtantive. Would it not better conſiſt with the genius of our
language if the plurals were to be always uſed, and the ſingulars made ad-
jectives without the termination al ?
228 On old Logical Terms.
(which now generally applies to ſomething ſpoken) is derived
from its place in this diviſion. The word argument, which is
now equivalent to reasoning against oppoſition expreſſed or implied,
was originally nothing but the middle term of a fyllogifm .
The ſimple term was univerſal or fingular : univerſal, when
of more inftances than one, as man, horſe, ſtar ; fingular, when
of one inſtance only, as the fun, the firſt man, the pole-ſtar, this
book. Singular names were called individuals, from the etymo-
logy of the word, as belonging to objects not diviſible into
inſtances to each of which the name could be applied. I have
not dwelt upon the distinction between fingular and univerſal,
becauſe it is ineffective in inference. And moreover, a fingular
propoſition is only objectively fingular, but ideally plural. ' Julius
Cæfar was a Roman' : in point of fact, there was but one Cæfar.
But take any imaginary repetition of the circumſtances of Cæfar's
life ; fuch, for inſtance as occurs to thoſe who have thought of
the poſſibility of the ſame courſe of events returning into exiſt-
ence after a certain cycle : and then the term Cæfar becomes
plural. Or, even without ſo forced a ſuppoſition, we may ſay
that, if we deſcribe Cæfar, we must deſcribe a Roman : that our
definition of Cæfar is ſo cloſe as to fit only one man that ever
lived, makes no eſſential difference in the character of the pro-
poſition.
But a further distinction which was made divided fingular
terms into ſubjects of univerſal, and ſubjects of particular, propo-
ſitions. A determinate (or definite) individual, as Cæfar, this
man, was the former : a vague (or indefinite) individual, as a
certain man, the firſt comer, was the latter. The distinction is
that of fome man ' and ' this one man.'
Certain notions of eſſence or relation, accompanying the ap-
prehenfion of a name, were called categories, or predicaments,
meaning ' modes of aſſertion with reſpect to the object named.
Aristotle gave ten categories, and might have given ten hundred.
In their uſual Latin form they were ſubſtantia, quantitas, quali-
tas, relatio, actio, paſſio, ubi, quando, fitus, habitus.
The word tranflated by ſubſtance, ἐσία, means mode of being :
and its literal Latin is eſſentia, eſſence. It is calledſubſtance (that
which ſtands under) as ſupporting accidents, preſently explained.
It is far too metaphyfical a term to come into common life with
 On old Logical Terms. 229

out ſome degradation : and accordingly it there means that of

which a thing is compoſed, whether material or not. Accordingly
we have the material ſubſtance of a coat, the intellectual ſub-
ſtance of an argument. But, as we uſe the word, its meaning
belongs to the other predicaments. In fact, the ſubſtance of the
old logicians ſtands, as to exiſtence, in the ſame ſituation as mat-
ter (page 30) with reſpect to our ſenſible perceptions, or object
with reſpect to our ideas. The ſubſtance, it was ſaid, is per fe
fubfiftens, while the accident could not be ſaid eſſe, but ineffe.
The distinction between the ſubſtance (mode of being) and the
material ſubſtance (in the modern ſenſe) may be helped by the
diſtinction between ſubſtantia prima and ſubſtantia fecunda, the
firſt referring to the individual, the ſecond to the general term.
Thus the ſubſtance ofJohn, as John, wasſubſtantia prima ; as
man, fubftantia fecunda. All theſe very metaphyſical notions
were the ſtudent's firſt introduction to logic, and were confidered
as of the utmost importance.
The predicament of quantity, derived from the notion ofwhole
and part, was conceived as either continuous or difcrete. In con-
tinuous quantity, the unit was diviſible, in difcrete, indiviſible.
Thus ten feet is continuous, ten men diſcrete. The distinction
is precisely that of magnitudinal and numerical.
Quality was ſubdivided into 1. Habit and diſpoſition, the latter
term being uſed for the imperfect ſtate of the former 2. Power
and want of it 3. Patibilis qualitas and paſſio, applied to the
ideas ofthat which is undergone, the firſt permanently, the ſecond
for a time. 4. Form andfigure.
Relation then, as now, referred to the ſuggeſtions derived from
comparison of two things or ideas. It was divided into verbal
and real (fecundum dici and ſecundum eſſe) . Thus the relation of
profit to profitable was verbal : that of father to fon, or of above
to below, real. The two things related, or correlatives, were called
Subject and term : ſo that of two correlatives, giving two oppofite
relations, the ſubject of either was the term of the other. The
fundamentum of the relation was that in which it took its riſe,
when it had a beginning.
Action and paſſion, the production and reception of an effect,
requiring the producing agent, and the receiving patient, were
divided into immanent, or enduring in the agent, and tranfient,
230 On old Logical Terms.
or paſſing out to another. Actions were univocal, or æquivocal,
according as their effects were of the ſame or different ſpecies.
A few years before the publication of Newton's Principia, it was
taught in a work imported into Cambridge that when mice bred
mice, the action was univocal, but when the sun bred mice (the
writer muſt have been thinking of Aristotle and ſome of the
ſchoolmen) æquivocal. There was alſo the terminus à quo and the
terminus ad quem to represent the ſtate before and the ſtate after
the action. Thus, when all this nonſenſe was ſent to Coventry,
the terminus à quo was an immenſe quantity of univocally bred
learning of the preceding kind ; the terminus ad quem was the
rooting up of the wheat of logic with the tares.
The where (as to abſolute poſition), the when, and the fite
(relative pofition) gave no peculiar terms of ſubdivifion. The
habitus (ἔχειν) referring to poſſeſſion generally in the firſt inſtance,
was materialized by ſome of the old logicians till it related to
drefs only, or habit in the thence acquired meaning.
The word predicament (and category as well) has been intro-
cuced into common language to ſignify a ſet of circumstances
under which any thing takes place. It is then no longer con-
fined to the above predicaments, nor is there any occafion that it
ſhould be.
The predicables (κατηγορέμενα) are diftinguiſhed from predica-
ments (κατηγορίαι) in that the former belong to any ſimple notion
or name, and may be predicated of it : the latter belong to the
connexion (when affirmative) between two names. They are
ſaid to be five in number, genus, ſpecies, differentia, proprium,
and accidens.
The words genus andſpecies have preſerved their old meaning.
If there be a number of names of which each is ſubidentical of
the one which follows, ſay V, W, X, Y, Z : then ofany two, ſay
W and X, X is a genus containing the ſpecies W. Here Z is
the fummum genus, and V the infima ſpecies : X is the genus
proximum of W, Y the genus remotum. In what I have called a
univerſe, which is a fummum genus, having for its infima ſpecies
the individual inſtance of any name in it, the ſuperidentical is the
genus, the fubidentical the ſpecies. Subcontraries (and contraries)
are oppoſiteſpecies ; fupercontraries and complex particulars have
no ancient name .
 On old Logical Terms. 231
The differentia is that by which one claſs (be it ſpecies or
genus, the difference being accordingly termedſpecific or generic)
is diftinguiſhed from another. Thus the difference (or one differ-
ence) ſeparating the ſpecies man from the other ſpecies of the
genus animal, is the epithet rational. :

The proprium (or property) is that which belongs to the ſpecies

only, whether it be to all or only to ſome : thus to ſtudy, and to
ſpeak, are equally propria of man. But the old commentators
give definitions of the property as follows. There are four
kinds. 1. That which belongs to the ſpecies alone, but not to
all. 2. To all the ſpecies, but not to that alone. 3. To the
ſpecies only, and to all of it, but not at all times. 4. To the
ſpecies alone, to all, and always.
The accidens (or accident) is that which may ſometimes be-
long to the individual of a ſpecies, but not neceſſarily, nor to that
ſpecies alone. In modern language, the term is limited to what
is unusual and unexpected.
The word cauſe was used by the ancients in a wider ſenſe
than by us : more nearly in the ſenſe of the Latin caufa, or the
Italian cofa. Cauſes were diftinguiſhed into material, formal,
efficient, andfinal. The material cauſe was the very matter of
a thing, confidered as a kind of giver of existence ; the formal
cauſe was its form, in the ſame light ; the efficient cauſe (our
common Engliſh word) the agent or precedent ; and the final
cauſe, the ultimate end or object, confidered as a reaſon for the
exiſtence of the thing. Sometimes writers ſtill talk of final
cauſes, and are as unintelligible to moſt readers as if they had
talked of final beginnings.
The word form was uſed in a wider ſenſe than that of figure
or ſhape, to mean, as it were, law of exiſtence, mode, diſpoſition,
arrangement. Mere figure or ſhape was only one of the acci-
dentalforms, as diftinguiſhed fromſubſtantialforms, belonging to
the ſubſtance. And motion was as widely used as form : it meant
any alteration. Thus, corruption was one of the motions of mat-
ter. Change from place to place, to which the modern word is
confined, was local motion.
The original uſe of the termsfubject and object is to denote a
thing confidered as that which may have ſomething inherent in
it, or attached to it, or ſpoken of it, &c.; and as that which may
 232 On old Logical Terms.
be objected to the mind or reaſon, or made to come in its way.
Thus it was ſaid that matter is the subject of thoſe properties
which are the objects of the mind in natural philoſophy. The
tranſition to the modern ſenſe of object, namely, end propoſed, is
natural enough. In modern times, ſubject and object are uſed *
with reſpect to knowledge : the ſubject being the mind in which
it is, the object being the external ſource from which it comes.
Forfubjective and objective I have in this work uſed ideal and ob-
jective (page 29). Adjunct was the technical term for that
which is in the ſubject.
A modal propofition was one in which the affirmation or nega-
wa expreſſed as more or leſs probable : including all that is
tionwas
technically under probability (Chapter IX) from neceffity to
impoffibility. The theory of probabilities I take to be the un-
known God which the ſchoolmen ignorantly worſhipped when
they ſo dealt with this ſpecies of enunciation, that it was ſaid to be
beyond human determination whether they moſt tortured the
modals, or the modals them. Their gradations were neceſſary,
contingent, poſſible, impoſſible ; contingent meaning more likely
than not, poſſible leſs likely than not. Theſe they connected
with the four modes of enunciation, A, I, O, E, and when by
Some is meant more than half, the connexion is good. The con-
troversy about modal forms continues up to this day among
logicians who are not mathematicians : I ſhould ſuppoſe that the
latter would never give it a thought, except as a branch of the
theory of probabilities, and except as to the confideration how
the terms by which the non-mathematical logician indicates his
degrees of belief are to be placed upon the numerical ſcale. In
like manner he reads the thermometer by graduation, and though
he admits the freezing and boiling point, which have an origin
in nature, he leaves temperate, ſummer heat, blood heat, &c. to
the fancy of thoſe who choose to employ them.
At the fame time it is clear that theſe modal forms were con-
fidered not merely as uſeful in expreſſion ofthe nature and amount
of belief, but as ſuggeſtive of real branches of inquiry, ſubſervient
to that great à priori inquiry into the nature of things to which

* See a full account of theſe words in Sir William Hamilton's notes to

Reid, p. 806, &c.
 On old Logical Terms. 233

mediæval logic was applied. We are not fit to judge of the in-
ſtrumental part of this philoſophy, unleſs we conſider alſo the
materials on which it was founded. In an age in which much
more faith was demanded of the ſtudent than now ; when he was
much more frequently required to decide in one way or the
other upon a ſingle teſtimony ; when, in addition to the non-
mythic wonders recorded in ancient writers, which there was no
mode of contradicting, all that was known of immenſe regions
and countries reſted upon very few accounts, and thoſe filled
with ſtories quite as ſtrange :-the absence of other means of
diftinguiſhing truth from falſehood obliged those who thought to
lay much ſtreſs upon à priori confiderations. It matters little to
us whether we infer the neceſſity of man being a walking animal
from the non-arrival of exceptions, and thence the univerfality
of the rule, or the univerſality from the ſuppoſed perfect induction
of inſtances, and thence the neceſſity. But it was of much more
conſequence to the old logician : of more real conſequence. He
did not know but that any day of the week might bring from
Cathay or Tartary an account of men who ran on four wheels
of flesh and blood, or grew planted in the ground like Polydorus
in the Æneid, as well evidenced as a great many nearly as mar-
vellous ſtories . As he could not pretend to inductive and demon-
ſtrative univerſality, even upon the queſtion of the form of his
own race, he was obliged to combine with his argument the an-
tecedent teſtimony of his own and other minds, in the manner
which the real doctrine of modals (page 205) ſhows to be necef
ſary in all non-demonftrated concluſions. It is true that he fre-
quently confounded the prediſpoſition of minds with the confti-
tution of objects ; the teftimony with the thing teſtified about.
We shall never have true knowledge of the ſchools of the
middle ages, until thoſe who have ſtudied both their philoſophy,
their phyſics, and their ſtate of tradition, will look at their
weapons of controversy as both offenſive and defenfive, and give
a fair account of the amount of protection afforded by the firſt,
in the exiſting ſtate of the ſecond and third. It would alſo be
adviſable to confider whether, looking at the power of communi-
cation by land and ſea, and all the circumſtances of literary inter-
courſe, it would have been practicable to place the knowledge of
the earth and its details upon any better footing of evidence.
 234 On old Logical Terms.
One leading feature of the ſchoolmen, acute as they were, and
as to repreſentation of notions, inventive, and which is ſhared by
many more modern writers who have not diſciplined themselves
mathematically, is ſeen in their employment of quantity : there
are inſtances of the ſtrange uſe, the wrong uſe, and the no-uſe.
Moſt of them ariſe from indiſtinct apprehenfion of continuity,
which obliges them to accept ſuch ſtages of quantity as are ex-
preſſed by exiſting terms, without any effort to fill up gaps.
There is alſo a ſlovenlineſs of definition in what relates to quan-
tity. Thus dozens of inſtances might be given in which the
Some of the particular propoſition is ſo defined that we might
ſuppoſe it is ſome, not all, instead of ' ſome, it may be all, and
the former is the expreſs definition of ſome writers : and it is only
when we find in rules that XY does not allow us to infer X :Y,
nor to contradict X)Y, that we aſcertain the real intended
meaning. " Logicians," ſays Sir William Hamilton, “ have
referred the quantifying predeſignations plurimi, and the like, to
the moſt oppofite heads ; ſome making them univerſal, ſome
particular, and ſome between both." They must have had curi-
ous ideas of quantity who made the propoſition ' moſt Xs are
Ys' either univerſal, or between the univerſal and particular : I
ſhould ſuppoſe that those who did the latter must have imagined
fome to refer to a minority.
There is a ſtrange notion ofquantity revived in modern times ,
which conſiſts in making plurality ofattributes a part ofthe quan-
tityofa notion. It is called its intenſive quantity, or its intenfion, or
comprehenfion. It is oppoſed to extenſive quantity, or extenfion,
which is the more common notion of quantity, referring to the
number of ſpecies or ofindividuals (it may be either, the individual
is the real infimaſpecies) contained under the name. Thus man is
not ſo extenſive as animal, but more intenſive ; the attribute ratio-
nalgives greater comprehenfion. But ' man refiding in Europe ' is
leſs extenſive and more comprehenſive than either. It is said that
the greater the intenſive quantity the leſs the extenſive, but this is
not true, unleſs no two of the figns ofintenfion be properties of
the ſame ſpecies. Thus, according to ſuch ſtatements as I have
ſeen, ' man, reſiding in Europe, drawing breath north of the
equator, feeing the ſun riſe after thoſe in America,' would be a
more intenfively quantified notion than ' man reſiding in Europe ' ;
 On old Logical Terms. 235
but certainly not more extenſive, for the third and fourth elements
ofthe notion muſt belong to thoſe men to whom the firſt and ſe-
cond belong. Thus, in the Port-Royal Logic, one ofthe earlieſt
modern works (according to Sir W. Hamilton), in which the dif-
tinction is drawn, it is ſaid that the comprehenfion of the idea of a
triangle includes ſpace, figure, three fides, three angles, and the
equality of the angles to two right angles. But the idea ofrecti-
linear three-fidedfigure has juſt as much extenfion.
The relation between comprehenfion and extenfion exiſts, and
is useful : but not, I think, as that of different kinds of quantity.
In page 148, where I hold that the propofition is contained in its
neceſſary conſequence, the view is one of extenfion : the ordinary
view is one of comprehenfion. Every cafe in which P is true,
is a caſe in which Q is true,' tells us that all the P-caſes are con-
tained, as to extent (number and location of inſtances), among
the Q-caſes. But, as to comprehenfion, every P-caſe contains
all that diftinguiſhes a Q-caſe from other things. When, in
page 47, it is ſaid that the idea of man is contained in that of
animal, I ſpeak of extenſion : all the inſtances to which the firſt
idea applies are among thoſe to which the ſecond applies. But,
as to comprehenfion, the idea of animal is contained in that of
man : all that defines animal goes to the definition of man, and
other things beſides. In page 50, the " is of poffeffion of all
eſſential characteriſtics," refers to comprehenfion ; the " is of
identity" to extenfion : both poſſeſſing equally the characters
under which the verb may occur in logic. There is no diftinction
which affects inference : for X)Y has exactly the fame proper-
ties whether we interpret it as expreſſing that Y has all the ex-
tenfion of X, and may be more ; or that X has all that Y has in
comprehenfion, and may be more.
In pages 115, &c. we have the mode of repreſenting names
of more or leſs comprehenfion. Thus, P, Q, R, &c. being cha-
racteriſtics, the obvious propofition PQ)P, illuſtrates the theorem
that where the comprehenfion of one name has all that of a ſe-
cond (as PQ has that of P) the extent of the ſecond is at leaſt
as great as that of the firſt. And the ſelf-evident poftulate in page
115, by which we may diminiſh the extent of a term univerſally
uſed, or increaſe that of one particularly uſed, may be expreſſed in
language of comprehenfion. That is, we may augment the com
236 On old Logical Terms.
prehenfion of a univerſal, or diminiſh that of a particular. Thus,
X)Y gives XP)Y, and X.Y gives XP.Y : but X)YP gives
X)Y.
It will be easily ſeen that comprehenfion has the firſt attribute
of quantity (page 174) : there is more and leſs about it. But it is
not of the measurable kind (page 175). As to extent, 200 in-
ſtances bear a definite ratio to 100, which we can uſe, becauſe
our inſtances are homogeneous. But different qualities or defcrip-
tions can never be numerically ſummed as attributes, to any pur-
poſe arifing out of their number. Does the idea of rational ani-
mal, two deſcriptive terms, ſuggeſt any useful idea of duplication,
when compared with that of animal alone. When we say that
a chair and a table are more furniture than a chair, which is true,
we never can cumulate them to any purpoſe, except by abſtract-
ing ſome homogeneous idea, as of bulk, price, weight, &c. To
give equal quantitative weight to attributes, as attributes, ſeems
to me abſurd : to uſe them numerically otherwise, is at preſent
impoffible.
The reader will have ſeen the origin of ſeveral very common
terms, which are uſed in a ſenſe coinciding with, or at leaſt much
reſembling, that put upon them by the ſchoolmen. But there is
one which has diametrically changed its meaning ; it is the word
instance. The word inftantia (and alſo ἔνστασις) implied a cafe
against, notfor ; the latter was exemplum : ſo that inſtance to the
contrary would have been tautology.
I have referred the word enthymeme to this chapter, though it
is always regularly explained in connexion with the ſyllogifm .
According to Ariftotle, Ἐνθύμημά ἐστι συλλογισμὸς ἀτελὴς ἐξ εἰκότων
καὶ σημείων, an enthymeme is an imperfect fyllogiſm from probables
and ſigns : the modern critics reject the word ἀτελὴς, imperfect, as
interpolated. The wordſign ſeems to mean indication, ſymptom,
or effect, which makes the cauſe almoſt neceſſary or highly
probable. But the ſchools took the word enthymeme to mean a
ſyllogiſm with a ſuppreſſed and implied premiſe, ſuch as ' He must
be mortal, being a man.' I cannot help ſuſpecting that Ariftotle *

* He ſays all that is communicated (λέγεται) of the predicate, will be

aſſerted in words (ῥηθήσεται) of the ſubject. Theſe two different tenfes of
two different verbs are often both translated by dicitur. Why did they
 On Fallacies. 237

made no difference between a ſuppreſſed premiſe, clearly intended

and diftinctly received, and one formally given. It seems to me
that we might as well diſtinguiſh a written from a ſpoken fyllo-
giſm, as to the logical character of the two.

CHAPTER XIII .

On Fallacies.

THERE is no ſuch thing as a claſſification of the ways in

which men may arrive at an error : it is much to be
doubted whether there ever can be. As to mere inference, the
main object of this work, it is reducible to rules : theſe rules
being all obeyed, an inference, as an inference, is good ; conſe-
quently a bad inference is a breach of one or more of theſe rules.
Except, then, by the production of examples to exerciſe a be-
ginner in the detection of breaches of rule, there is nothing to do
in a chapter on fallacies, ſo far as thoſe of inference are con-
cerned. Nevertheless, there are many points connected with the
matter of premiſes, to which it is very defirable to draw a reader's
attention : and above all to queſtions in which it is not at firſt
obvious whether the miſtake be in the matter or in the form ; or
in which it may be the one or the other, according to the ſenſe
put upon the words.
If there be anything ridentem dicere verum quod vetat, writers
on logic have in all ages moſt grievouſly neglected the prohibition
in treating this ſubject, and have given the ſtudent a preſcriptive
right to ſome amusement. One reaſon of this was, that the

occur ? For various reasons, I allow myself to ſuſpect, though not ſcholar
enough to maintain, that λόγος generally meant communication, paſſage
from one mind to another by any means, as much at least with reference to
the receiving, as to the imparting, mind : and that it is here oppoſed to ῥῆσις,
ſpeech, in that ſenſe. Throw the verbs back to their primary meanings,
and it will be ' That which is picked up of the predicate, ſhallflow out about
the ſubject. ' If my conjecture be correct, the modern enthymeme is here
put on the fame footing as the fully expreſſed ſyllogifm .
 238 On Fallacies.

Greeks endeavoured to try the new art by inventing inferences

the falſehood of which could not be detected by its rules. Theſe,
as may be ſuppoſed, were whimsical efforts of reaſoning : never-
theless, they have been handed down from book to book, unfur-
paſſed in their way. Another reaſon is, that jeſts, puns, &c. are
for the moſt part only fallacies ſo obvious that they excite laugh-
1

ter ; and the greater number of them can be ſhown to break one
or another of the rules of logic. Accordingly, they furnish
ſtriking examples of theſe rules ; the application of which, in ſe-
rious terms, has itſelf a taſte of the ludicrous. Boccacio has, by
his inimitable mode of narration, made a good ſtory the jest of
which could be deſcribed as confifting in nothing more than the
aſſumption that what can be predicated of ſtorks * in general
can be predicated of roaſted ſtorks : which is what logicians
would call the fallacia accidentis, or arguing a dicto fimpliciter,
ad dictumfecundum quid.
The terms fallacy, fophifm, paradox, and paralogiſm, are ap-
plied to offences against logic ; but not with equal propriety,
Fallacy and fophifm may technically have been firſt applied to
arguments in which there is a failure of logic : but it is now very
common to apply them alſo to arguments in which there is a
falſehood of fact, or error of principle, though logically treated ;
and if this laſt uſe be not correct, writers on logic have fanc-
tioned it in their examples. Many perſons go further, and call
the erroneous ſtatement itſelf a fallacy : that men are in the
habit of walking on their heads, they would ſay is a very obvious
fallacy. A paradox is properly ſomething which is contrary to
general opinion : but it is frequently uſed to ſignify ſomething
ſelf-contradictory: thus the newſpaper which recently avowed

* A ſervant who was roasting a ſtork for his maſter was prevailed upon
by his ſweetheart to cut off a leg for her to eat. When the bird came upon
table, the maſter defired to know what was become of the other leg. The
man anſwered that ſtorks had never more than one leg. The maſter, very
angry, but determined to ſtrike his fervant dumb before he puniſhed him,
took him next day into the fields where they ſaw ſtorks, ſtanding each on
one leg, as ſtorks do. The ſervant turned triumphantly to his maſter : on
which the latter ſhouted, and the birds put down their other legs and flew
away. " Ah, Sir," said the ſervant, "you did not ſhout to the ſtork at din-
ner yeſterday : ifyou had done ſo, he would have ſhown his other leg too."
 On Fallacies. 239
its opinion that the repeal of the corn laws would make food
both cheap and dear is ſaid to have maintained a paradox. The
modern uſe of the word implies diſreſpect, but it was not fo for-
merly. Thus in the fixteenth century the opinion of the earth's
motion was ſtyled the paradox of Copernicus by writers who
meant neither praiſe nor blame, but only reference to the opinion
of Copernicus as an unusual one. The more preciſe writers of
our day uſe the word paradox for an opinion ſo very fingular and
improbable, that the holder of it is chargeable with an undue
bias in favor of fingularity or improbability for its own fake.
Paralogism, by its etymology, is beſt fitted to fignify an offence
against the formal rules of inference. It has been frequently
abuſed by mathematical writers, who have fignified by it errors
of ſtatement, and undue aſſumptions : but it is not completely
ſpoiled for the purpoſe, and I ſhall therefore uſe it to denote a
formal error in inference, as a particular claſs of fallacy or fo-
phiſm, words which it would now be difficult to diftinguiſh in
meaning. Some have defined paralogiſm to be that by which a
man deceives himself, and ſophism that by which he tries to de-
ceive others : on what grounds I do not know.
The queſtion of a premiſe being right or wrong in fact or
principle, unleſs indeed it contradict itſelf, does not belong to
logic : nor could it ſo belong unleſs logic were made, in the wideft
ſenſe, that attempt at the attainment of the cognitio veri which
ſome have defined it to be. All that relates to the collection of
true premiſes with reſpect to the vegetable world belongs to
botany; with reſpect to the heavenly bodies, to aſtronomy ; with
reſpect to the relation of man to his Creator, to theology. Even
were it within the province of logic, it would be impoffible, in
leſs ſpace than an encyclopædia, to enter upon queſtions con-
nected with the matter of fyllogifms. With regard to paralogiſms,
or logical fallacies, (ſo called, as an error about the meaſure of
Space is called a geometrical error) the claſſification under breach
of rules would be good in form, but would afford no bafis for
the treatment of the ſubject. Those who bring them forward
ſeldom proceed in direct defiance of rule, but in various modes
of evaſion. Theſe it would be almoſt impoſſible to arrange in
ſatisfactory order.
Aristotle made a claſſification of fallacies, which was of courſe
240 On Fallacies.
adhered to by the writers of the middle ages. In this, as in
every other place, when I speak ofAristotle and his ſyſtem, I
ſpeak of it as understood by thoſe writers. How far they dif-
tinctly comprehended their maſter is a queſtion into which I
could not enter here, even if I were competent to write on the
ſubject. It is, however, ſufficiently apparent that the logic of
Ariftotle is not of the purely formal character which marked the
dialectics of the middle ages : there is a much more decided
introduction of the attempt to write on the matter of fyllogifm
than many perſons think there is. The claſſification of fallacies
ſeems to be one proof of this : and the interpretation of that claf-
fification by the middle writers ſeems to add their teſtimony to
the aſſertion : in this part of the ſubject they abandon techni-
calities almoſt entirely.
It ought to be eſpecially remembered that we are very diffe-
ently ſituated from thoſe writers, not as to what is fallacy, but
as to what the ſpecimens of it produced are likely to be. Out
of a world of general principles declared by authority, or declared
to be ſelf-evident by authority, they had to produce logical de-
ductions ; and, of course, the pure ſyllogiſm and its rules were
to them as familiar as the alphabet. The idea of an abſolute and
glaring offence against the ſtructure of the fyllogifm being fup-
ported one moment after it was challenged, would no more
ſuggeſt itſelf to the mind of a writer on logic than it would now
occur to a writer on aſtronomy that the accidental error (which
might happen to any one) of affixing four ciphers inſtead of five
in multiplying by a hundred thouſand would be maintained after
expoſure. Accordingly, their formal chapters on fallacies would
naturally relate, if not entirely to fallacies of matter, at least to
thoſe in which the fallacy of matter very cloſely hinges upon that
of form. And ſo it is in all the old ſyſtems which I have exam-
ined. The Ariftotelian diviſion (or rather ſelection, for it is far
from including everything) lends itſelf easily to this adaptation.
We, on the contrary, live in an age in which formal logic has
long been nearly baniſhed from education : entirely, we may ſay,
from the education of the habits. The ſtudents of all our uni-
verſities (Cambridge excepted) may have heard lectures and
learnt the forms of ſyllogiſm to this day: but the practice has
been ſmall : and out of the univerſities (and too often in them)
the very name of logic is a bye-word.
 On Fallacies. 241
The philoſophers who made the diſcovery (or what has been
allowed to paſs for one) that Bacon invented a new ſpecies
of logic which was to ſuperſede that of Aristotle, and their fol-
lowers, have ſucceeded by falſe hiftory and falſer theory, in driv-
ing out from our ſyſtem all ſtudy of the connexion between
thought and language. The growth of inaccurate expreſſion
which this has produced, gives us ſwarms of legiſlators, preachers,
and teachers of all kinds, who can only deal with their own
meaning as bad ſpellers deal with a hard word, put together
letters which give a certain reſemblance, more or leſs as the caſe
may be. Hence, what have been aptly called " the flipſhod judg-
ments and crippled arguments which every-day talkers are content
to uſe." Offences against the laws of fyllogifm (which are all laws
of common ſenſe) are as common as any ſpecies of fallacy : not
that they are always offences in the ſpeaker's or writer's mind,
but that they frequently originate in his attempt to ſpeak his
mind. And the excuſe is, that he meant differently from what
he said : which is received becauſe no one can throw the firſt ſtone
at it, but which in the middle ages would have been regarded
as a plea of guilty. The current notions about what logic is, are
beautiful and wonderful. I have heard a diſputant, an educated
man, a graduate, eſcape from allowing himself to be convinced
that he was arguing with a middle term particular in both pre-
miſes by declaring thatfacts were better thanfyllogifms : the form
of his argument would have proved that men are plants, becauſe
both require air. " I" he said, " produce you facts, like Bacon :
you quibble about their combination, like Ariftotle."
The Ariftotelian ſyſtem of fallacies contains two fubdiviſions.
In the firſt, which are in dictione, or in voce, the mistake is faid
to confift in the use of words : in the ſecond, which are extra
dictionem, or in re, it is ſaid to be in the matter.
Ofthe firſt ſet ſix kinds were diftinguiſhed, as follows :-
1. Æquivocatio or Homonymia, in which a word is uſed in two
different ſenſes ; giving really no middle term (if the middle term
be in queſtion) or a term in the conclufion which is not the fame
name as that uſed in the premiſes. For example, ' All criminal
actions ought to be punished by law : profecutions for theft are
criminal actions ; therefore, proſecutions for theft ought to be
puniſhed by law.' Here the middle term is doubly ambiguous,
R
242 On Fallacies .
both criminal and action having different ſenſes in the two pre-
miſes. But here, as in many other cafes, the choice lies with
the ſophift to bring the fallacy under the head to which we refer
it or not. It may pleaſe him to affert that he means the ſame
thing by criminal action in both premiſes ; in which case, the in-
ference is logical, but one or the other premiſe must be denied
as to the matter. Again, Finis rei eſt illius perfectio ; mors eſt
finis vitæ ; ergo mors eſt vitæ perfectio.' Here the ambiguity
may be thrown either on finis or on perfectio. The following
example can be traced through books for three centuries. 'Every
dog runs on four legs ; Sirius (the dog-ſtar) is a dog ; therefore
Sirius runs on four legs.' It has been the defect of many old
works on logic that all their examples have been of that obvious
abſurdity, which is well enough in one or two inſtances. Such
as ' Nothing is better than wisdom and virtue ; dry bread is bet-
ter than nothing ; therefore, dry bread is better than wisdom and
virtue.' Some of the old examples are ' A mouſe eats cheeſe ;
a mouſe is one fyllable ; therefore one fyllable eats cheeſe.' And
again, ' Iſte pannus eſt de Anglia ; Anglia eſt terra ; ergo, ifte
pannus eft de terra.'
Where the fyllogifm is formally put, equivocation of the mid-
dle term is generally ſeen with great eaſe. The moſt difficult
exception is, I think, the old fallacy, in which giving the name
of the genus is confounded with giving the name of the ſpecies,
and thereby, of courſe, giving the name of the genus. As in ' To
call you an animal is to ſpeak truth ; to call you an aſs is to call
you an animal ; therefore, to call you an aſs is to ſpeak truth.'
This equivocation will puzzle a beginner as to its form, and the
more fo from the evident falſehood of the matter. The middle
term is " He who ſays that you are one among all animals." He
ſpeaks truth ; and the one who calls you an aſs or a gooſe, cer-
tainly ſays that you are one among all animals. The equivocation
is in the two different uſes of the word one ; in the firſt premiſe,
it is an entirely indefinite one ; in the ſecond it is a leſs indefinite
one. This one is not attached to the quantity of the middle term,
which is univerſal in the firſt premiſe, and particular in the ſe-
cond : but is part of the middle term itſelf.
The manner in which the ſerious fallacy of equivocation moſt
frequently appears, is in the connection of the old afſociations of
 On Fallacies. 243
a word which has ſhifted its meaning with the altered meaning of
the fame. The word loyal, for instance, originally meaning no
more (and no leſs) than lawful, which, as applied to a man, meant
one who reſpected the laws, and had not forfeited any right by
miſbehaviour, now means attached to the Crown and to the title
of the holder of it. In contefts for fucceffion, the winner would,
of courſe, aſſume that lawful men were on his ſide. In more
recent times, the term was always felf-applied, at elections, by
thoſe who ſupported the party which had the confidence of the
Crown for the time being : but on ſuch occafions, abftinence
from the fallacy which the French call the voie du fait is the
utmoſt which can be expected of human nature.
The word publication has gradually changed its meaning, ex-
cept in the courts oflaw. It ſtood for communication to others,
without reference to the mode of communication, or the number
of recipients. Gradually, as printing became the eaſieſt and moſt
ufſual mode of publication, and conſequently the one moſt fre-
quently reſorted to, the word acquired its modern meaning : if
we ſay a man publiſhes his travels, we mean that he writes and
prints a book deſcriptive of them. I ſuſpect that many perſons
have come within the danger of the law, by not knowing that to
write a letter which contains defamation, and to fend it to another
perſon to read, is publishing a libel ; that is, by imagining that
they were ſafe from the conſequences of publiſhing, as long as
they did not print. In the ſame manner, the well-eſtabliſhed
rule that the firſt publiſher of a diſcovery is to be held the diſco-
verer, unless the contrary can be proved, is misunderstood by
many, who put the word printer in the place of publiſher. I
could almoſt fancy that ſome perſons think rules ought to travel
in meaning, with the words in which they are expreſſed.
A fimilar change has taken place in the meaning of the word
to utter, the ſenſe of which is to give out, but which now means
uſually to give out of the mouth in words. As yet, I am not
aware that any perſon charged with the utterance of counterfeit
coin has pleaded that no one ever uttered coin except the prin-
ceſs in the fairy tale : but there is no ſaying to what we may
come, with good example, and under high authority.
It may almoſt be a queſtion whether, in the time of Ariftotle,
ſucceſsful equivocation, that is, undetected at the moment, won'
244 On Fallacies.

not have been held binding on the diſputant who had failed to
detect it. The genius of uncultivated nations leads them to
place undue force in the verbal meaning of engagements and
admiſſions, independently of the understanding with which they
are made. Jacob kept the bleffing which he obtained by a trick,
though it was intended for Eſau : Lycurgus ſeems to have fairly
bound the Spartans to follow his laws till he returned, though he
only intimated a ſhort abſence, and made it eternal : and the
Hindoo god who begged for three ſteps of land in the ſhape of
a dwarf, and took earth, ſea and ſky in that of a giant, ſeems to
have been held as claiming no more than was granted. The
great ſtreſs laid by Aristotle on ſo many different forms of verbal
deception, may have ariſen from a remaining tendency among
diſputants to be very ferious about what we should now call play
upon words.
Governments permit what would otherwise be equivocation to
take a ſtrong air of truth, by legiſlating in detail against the prin-
ciples of their own meaſures. The window-tax is a ſpecial in-
ſtance. A newſpaper calls it a tax upon the light which God's
beneficence has given to all. The answer would be plain enough,
namely, that it is an income tax levied upon a uſe of that light
which (how truly matters not here) is aſſerted to be a fair criterion
of income. But this anſwer is destroyed by the permiffion to
block up windows, and thereby evade the tax : which is thus
made to fall upon the light uſed, and not upon the means of
uſing it which the size of the houſe affords. According to the
principle of this impoſt, the blocked window is as fair a crite-
rion of the income of the occupant as the open one, and ſhould
have been ſo confidered.
Among the forms which the fallacy of equivocation frequently
aſſumes, is that of the ſophift altering or qualifying the known
meaning of a word in his own mind, without giving the other
party any notice : ſo that there may be, if not two meanings in
ce mind, yet different meanings in the two minds concerned.
4
perſon aſſerts that ' Nobody denies, &c. &c.' Should this go
down, the point is gained ; what nobody denies must be undeni-
able. But ſhould it be conteſted (and it will generally be found
that the things which nobody denies are matters of ſome diffe-
rence of opinion, while thoſe which nobody can deny are quite
 On Fallacies. 245
ſure to be points of conſtant controverſy) the evaſion is ready.
It is no ſenſible perſon, or nobody that understands the ſubject,
nobody that is anybody, in ſhort : while perhaps it cannot be
ſettled who does, or who does not, underſtand the ſubject, until,
among other things, the very point in diſpute is determined.
There is a wide range of equivocations arifing out of mean-
ings which are ſometimes implied and ſometimes not. A large
claſs of them is made by the uſual, but not univerſal, practice,
ofgiving to the thing the name of that which it is intended
to be, whether the attempt be ſucceſsful or not. This is now
abbreviation or courteſy ; but it was the rule. According to old
definitions, bad reaſoning is reaſoning, fyllogifmus ſophiſticus is a
fyllogifm, and in an old book now before me, the fruits and effects
of demonſtration are ſcience, opinion, and ignorance, the latter
containing belief of falſehood derived from bad demonftration,
which we should now call no demonſtration .
One fallacy of our time, and a very favourite one, is the fet-
tlement of the merit of a perſon, or an opinion, not by arguing
the place of that perſon or opinion in its ſpecies, but by arbitrary
alteration of the boundary of the ſpecies, with the intent of ex-
cluding the individual in queſtion altogether.
It is ſomewhat analogous to the proceeding of the landlord
who unroofs the houſe to get rid of a tenant. Thus we have
had the controversy whether Pope was a poet, not whether he
was a good poet or a bad one, but whether he was a poet at all.
The difputants, or fome of them, claimed a right to define a poet,
and decided that none but verſe-makers of a certain goodneſs (to
be ſettled by themſelves) were poets. They might juſt as well
have decided, on their own authority, that none but men of a
certain amount of reasoning power were men. Had they done
this laſt, as long as they fixed the amount at a figure which in-
cluded themſelves under the name, nobody would have thought
they materially altered the extent of the term : it is not eaſy to
ſee why they have rights ſo arbitrary, over words the objective
definitions of which are nearly as well fixed as that of man.
Another form of the fallacy of equivocation is the aſſuming,
without expreſs ſtatement, that the meaning of a phrafe can be
determined by joining the meanings of its ſeveral words : which
is not always true in any language. When two words come to
246 On Fallacies .
gether, it often happens that their dictionary meanings would
never enable us to arrive at their known and uſual (and therefore
proper) compound meaning : though they might help us in ex-
plaining how that laſt meaning aroſe. A perſon undertakes to
croſs a bridge in an incredibly ſhort time : and redeems his pledge
by croffing the bridge as one would croſs a ſtreet, that is, by
traverſing the breadth. Now, though it be true that, in general,
to croſs is to go over the breadth, or ſhorter dimenſion, yet in
the cafe before us, the phrafe is elliptical, and ſignifies croſſing
the river upon the bridge. Nor can it be faid that this common
meaning is incorrect : that which is common and well known is,
in language, always correct. No reaſonable perſon would ſay
that a French newſpaper is wrong in reporting an army to be à
chevalfur la rivière, becauſe a river is not a horſe. This literal
(or rather unlettered) mode of interpretation is adopted among
gamblers in fettling bets : and is of itſelf enough to raiſe a ſtrong
preſumption that their occupation is not that of well-educated
men.

It is common enough in controversy, for one ſide or the other

to have fixed meanings of words in his own mind, on which he
proceeds without any inquiry as to whether thoſe meanings will
be conveyed by the words to the other fide, or to the reader. It
is very difficult to avoid this form of the fallacy, without giving
the meanings of the moſt eſſential terms, on the firſt occafions
of their occurrence . It is not uncommon to meet with a writer
who appears to believe, at least who certainly acts upon, the
notion that the right over words refides in him, and that others
are wrong fo far as they differ from him. I do not only mean
that there are many who have an undue belief in their own
judgments, both as to words and things : but I ſpeak of thoſe
who, though ſhowing a proper modeſty in reſpect to their own
conclufions, feem to be unable to do the fame with reſpect to
their definitions of words. If all mankind had ſpoken one lan-
guage, we cannot doubt that there would have been a powerful,
perhaps a univerfal, ſchool of philoſophers who would have be-
lieved in the inherent connexion between names and things ;
who would have taken the found man to be the mode of agita-
ting the air which is effentially communicative of the ideas of
reafon, cookery, bipedality, &c. The writers of whom I ſpeak,
 On Fallacies. 247

are more or leſs of this ſchool ; they treat words as abſolute

images of things by right of the letters which ſpell them. "The
French," ſaid the failor, " call a cabbage aſhoe ; the fools ! why
can't they call it a cabbage, when they must know it is one ? "
Equivocation may be uſed in the form of a propofition ; as
for instance, in throwing what ought to be an affirmative into
the form of a qualified negative, with the view of making the
negative form produce an impreſſion. Thus a controverfial
writer will affert that his opponent has not attempted to touch a
certain point, except by the abſurd affertion, &c. &c. &c. To
which the other party might juſtly reply, " Your own words
ſhow that I have made the attempt, though your phraſe has a
tendency, perhaps intended, to make your reader think that there
is none, or at least to blind him to the difference between none
and none that you approve of."
2. The fallacia amphiboliæ, or amphibologiæ, differs in nothing
from the laſt, except in the equivocation being in the conſtruc-
tion of a phrafe, and not in a ſingle term : as in confounding that
which is Plato's (property) with that which is Plato's (writing).
Or, as in ' Qui funt domini fui funt ſui juris ; ſervi funt do-
mini fui ; ergo ſervi funt ſui juris.' The ambiguities of con-
ſtruction in our language, arising from want of inflexions and
genders are tolerably (and intolerably) numerous. The dif-
ficulty of determining the emphatic word often gives a doubt as
to the meaning. But very often indeed there is a want of the
distinction which the algebraiſt makes when he writes three-and-
four tens as diftinguiſhed from three and four-tens : (3 + 4) .10
and 3 + 4.10. It cannot, for instance, be faid whether ' I intend
to do it and to go there to-morrow ' means that it will be done
to-morrow or not. It may be either-(I intend to do it and to go
there) to-morrow, or-I intend to do it and (to go there to-mor-
row). The preſumption may be for the firſt conſtruction : but
it is only a preſumption, not a rule ofthe language. In an inſtance
cited by Dr. Whateley-" If this day happen to be Sunday,
this form of prayer ſhall be uſed and the faſt kept the next day
following," the conſtruction is ambiguous, and the intended mean-
ing probably against the preſumption. There is a book of the
laft century, written by a " teacher of mathematics, and writing
mafter to Eton College." Were mathematics taught at Eton,
248 On Fallacies.
or not ? Punctuation may be an aſſiſtance ; but it ſo often hap-
pens that the author leaves that point to the printer, that it is
hardly ſafe to rely upon it. Printers punctuate correctly when
the meaning is clear : but when it is ambiguous, they may be
as apt to take the wrong meaning as any other readers.
3, 4. Thefallacia compofitionis, and fallacia diviſionis, conſiſt
in joining or ſeparating thoſe things which ought not to be joined
or ſeparated. If we may say that A is X and B is Y, ſo that A
and B is X and Y, we have no right to infer that we may form
the compound and collective names ' A and B,' and ' X and Y,'
and ſay that ' A and B' is 'X and Y.' Thus two and three are
even and odd : but five is not even and odd. Again, two and
five are four and three ; but neither is two four, nor five three.
It must be remembered that the word all, in a propofition, is not
neceſſarily fignificative of a univerſal propoſition : it may be a
part of the deſcription of the ſubject. Thus in ' all the peers are
a houſe of Parliament, we do not uſe the words all the peers in
the ſame ſenſe as when we say ' all the peers derive their titles
from the Crown.' In the ſecond caſe the ſubject of the propo-
ſition is peer ; and the term all is diſtributive, ſynonymous with
each and every. In the firſt caſe the ſubject is all the peers, and
the term all is collective, no more diftinguiſhing one peer from
another than one ofJohn's fingers is diftinguiſhed from another
in the phrafe, John is a man.' The ſame remarks may be made
on the wordfome ; as in ' ſome peers are dukes,' and ' fome peers
are the committee of privileges.' The all andsome of the quan-
tity of the propoſition are diftributive terms ; the all and fome of
the ſubject are collective. Again, all men are a ſpecies (of ani-
mals) which no number of men are, wanting the reſt. All men
here make the one individual object of thought of a fingular pro-
pofition. This amounts to an ambiguity of conſtruction, an
amphibologia, as do moſt ſources of fallacy falling under this head,
which can therefore hardly be confidered as anything more than
a caſe of the laſt. We want another idiom or the algebraical
diftinction, as in ' All (peers) hold of the Crown ; (all peers) are
a houſe of Parliament. '
5. The fallacia profodiæ or accentus was an ambiguity arifing
from pronunciation, and its introduction ſeems to lead to very
minute fubdiviſion of the ſubject, and to enſure the entrance of
 On Fallacies. 249
none but ludicrous examples. Burgerſdicius does not think it
unworthy of himself to defcend to the following, ' Omnis equus
eſt beftia ; omnis juſtus eft æquus, ergo omnis juſtus eft beftia.'
An older writer has ' Tu es qui es ; quies eſt requies ; ergo, tu
es requies. ' Theſe are mere puns ; and the makers of them
were fairly beaten by the contriver of Two men eat oyſters for
a wager, one eat ninety-nine, the other eat two more, for he eat
a hundred and won. ' But more ferious fallacies may be referred
to this head. A very forced emphafis upon one word may, ac-
cording to uſual notions, ſuggeſt falſe meanings. Thus, ' thou
ſhalt not bear falſe witneſs against thy neighbour,' is frequently
read from the pulpit either ſo as to convey the oppofite of a pro-
hibition, or to ſuggeſt that fubornation is not forbidden, or that
anything falſe except evidence is permitted, or that it may be
given for him, or that it is only against neighbours that falſe wit-
neſs may not be borne.
A ſtatement of what was ſaid, with the fuppreffion of ſuch
tone as was meant to accompany it, is thefallacia accentus. Gef-
ture and manner often make the difference between irony or
ſarcaſm, and ordinary aſſertion. A person who quotes another,
omitting anything which ſerves to ſhow the animus ofthe meaning ;
or one who without notice puts any word of the author he cites
in italics, ſo as to alter its emphafis ; or one who attempts to
heighten his own aſſertions, ſo as to make them imply more than
he would openly ſay, by italics, or notes of exclamation, or
otherwise, is guilty of the fallacia accentus.
To this fallacy I ſhould refer one of very common occurrence,
the alteration of an opponent's propoſition ſo as to preſent it in
a manner which is logically equivalent, but which alters the em-
phafis, either as noticed in page 134, or in any other manner. It
is generally not reaſoning, but retort, which is the object of the
alteration : for inference cannot be altered by changing a propo-
ſition into a logical equivalent, but a ſmart repartee may be very
effective against ' Some Xs are Ys,' but flat enough against ' fome
Ys are Xs. ' And even when the proponent mistakes his own
meaning, and miſcalculates his own emphasis, ſtill, if the mistake
be obvious, there is fallacy in taking advantage of it ; for he who
communicates in ſuch incorrect terms as ſhow what the correct
ones are, does, in fact, communicate in correct terms, to all who
250 On Fallacies.

ſee the ſhowing. Of course, reſpect for logic never ſtood in the
way of a fucceſsful retort from the time of Aristotle till now, nor
will on this fide of the millenium. A fpeculator once wrote to
a ſcientific ſociety, to challenge them to an (on his part) anti-
Newtonion controversy, relying on it that he could contend in
mechanics, though avowedly ignorant of geometry. He was
anſwered by a recommendation to ſtudy mathematics and dyna-
mics. His rejoinder was an angry pamphlet, in which, indignant
at the unfairneſs, as he took it to be, of the recommendation, he
exclaimed, ' I did not confeſs my ignorance of dynamics.' Had
he been worth the answering, it would have been impoffible to
refift the reply ' No, but you ſhowed it.' Had he written, as he
meant ' It was not dynamics of which I confefſed ignorance,'
and had an opponent written, as many would have done, ' You
ſay, fir, that you did not confeſs your ignorance of dynamics :
indeed you did not, you contented yourſelf with an ample diſplay
of it,' he would have uſed thefallacia accentus. Nor would he, in
my opinion, have been clear of it though he had only taken advan-
tage of a wrong, but evidently wrong, placement ofemphasis on
the part of the aſſailant. The use of ſuch a weapon, as to its
legitimacy, depends entirely upon the manner in which the quef-
tion ſhall be ſettled how far irony is allowable. Where the anſwer
is in the affirmative, a very obvious fallacy, as a farcafm, may be
permitted. But I may here obſerve, that irony itſelf is generally
accompanied by the fallacia accentus ; perhaps cannot be aſſumed
without it. A writer diſclaims attempting a certain taſk as above
his powers, or doubts about deciding a propoſition as beyond his
knowledge. A felf-fufficient opponent is very effective in aſſur-
ing him that his diffidence is highly commendable, and fully jufti-
fied by the circumſtances.
6. The fallacia figuræ dictionis, as explained, means literally
a miſtake in grammar and nothing elſe ; as that becauſefluvius
is aqua it is humidA, or that becauſe aqua is feminine, ſo is poeta.
All theſe fallacies in dictione come under the head of ambiguous
language, and amount to nothing but giving the fyllogifm four
terms, two of them under the ſame name. The fallacies extra
dictionem are ſet down as follows .
1. Thefallacia accidentis ; and 2. That à dictofecundum quid
ad dictum fimpliciter. The firſt of theſe ought to be called that
 On Fallacies. 251
of à dicto fimpliciter ad dictum fecundum quid, for the two are
correlative in the manner deſcribed in the two phrases. The firſt
confifts in inferring of the ſubject with an accident that which
was premiſed of the ſubject only: the ſecond in inferring of the
ſubject only that which was premiſed of the ſubject with an acci-
dent. The firſt example of the ſecond must needs be ' What you
bought yesterday, you eat to-day ; you bought raw meat yeſter-
day ; therefore, you eat raw meat to-day.' This piece of meat
has remained uncooked, as freſh as ever, a prodigious time. It
was raw when Reiſch mentioned it in the Margarita Philofo-
phica in 1496 : and Dr. Whateley found it in juſt the ſame ſtate
in 1826. Of the firſt, we may give the inſtance ' Wine is per-
nicious ; therefore, it ought to be forbidden.' The expreſſed
premiſe refers to wine uſed immoderately : the concluſion is
meant to refer to wine however uſed. This ſpecies of fallacy
occurs whenever more or leſs ſtreſs is laid upon an accident,
or upon any view of the ſubject, in the conclufion, than was
done in the premiſes. As in the following :- All that leads to
ſuch philoſophy as that of the ſchoolmen, with their logic, muſt
be unworthy to be ſtudied, except hiſtorically.' The intent of
ſuch a fentence is not formally to propoſe the falſe fyllogifm,
،
The ſchoolmen had that which led them to a falſe philoſophy ;
the ſchoolmen had logic ; therefore, logic led them to a falſe phi-
loſophy,' but only to take the chance of the ſtreſs thus laid upon
logic producing a diſpoſition to ſuppoſe that the logic was in fault.
The premiſes are really :-
The philofophy of the ſchool-
men (who paid particular atten- is a falſe philoſophy.
tion to logic)
that the guides to which
Every falfe philofophy is ſhould be neglected, except
}* * as hiſtory.

whence it is rightly inferred that the guides to ſuch a philoſophy

as that of the ſchoolmen (who ſtudied logic) are only of hiſtorical
uſe. And the ſame thing might equally be inferred of the ſchool-
men who ate mutton, a practice to which most of them were as
much addicted, no doubt, as to making fyllogifms. The art of
252 On Fallacies.
the ſophift conſiſts in making the accident which is either un-
fairly introduced, or withdrawn, or ſubſtituted, have an apparently
relevant relation to the ſubject itſelf. Undoubtedly, the ſchool-
men's logic has a connexion with their philoſophy which the
mutton they ate has not : but as long as it is not the connexion
which permits the inference, it is abſolutely irrelevant.
All the fallacies which attempt the ſubſtitution of a thing in
one form for the fame thing (as it is called) in another, belong to
this head : fuch as that of the man who claimed to have had one
knife twenty years, giving it ſometimes a new handle, and fome-
times a new blade. The answer given by the calculating boy
(page 54, note) was, relatively to the queſtion, a worthy anſwer,
and took advantage of the common notion that a bean, after
being ſkinned, is ſtill a bean, as before. More ſerious difficulties
have ariſen from the attempt to ſeparate the effential from the
accidental, particularly with regard to material objects. The
Cartefians denied weight, hardneſs, &c. to be eſſential to mat-
ter, until at laſt they made it nothing but ſpace, and contended
that a cubic foot of iron contained no more matter than a cubic
foot of air.
The law, in criminal caſes, demands a degree of accuracy in
the ſtatement of the fecundum quid which many people think is
abfurd : and it appears to me that the lawyers often help the
popular miſapprehenfion, and give it excuſe, by confounding
errors of things with errors of words, after the example of the
world at large. Any error of any kind, provided it be ſmall in
amount, paſſes for a mistake in words only, by virtue of its ſmall-
neſs. By a mistake in words, I mean the addition or omiffion
of words which, whatever they might do under another ſtate of
things, do not, as matters ſtand, affect the meaning.
Take two inſtances, as follows ;-Some years ago, a man was
tried for ſtealing a ham, and was acquitted upon the ground that
what was proved against him was that he had ſtolen a portion of
a ham. Very recently, a man was convicted of perjury, ' in the
year 1846,' and an objection (which the judge thought of impor-
tance enough to reſerve) was taken, on the ground that it ought
to have been in the year of our Lord 1846.' There may, of
courſe, be acknowledged rules, which, as long as they are rules,
must be obeyed, and which may make the ſecond miſtake as ne
 On Fallacies. 253
ceſſarily vitiate an indictment as the firſt. But, in diſcuſſing the
policy of the rules, it would seem to me that the two cafes are
entirely different. In both, no doubt, the rest of the indictment
might, by implication, make good the meaning required : but
there ſeems a great difference between allowing the remainder to
correct an error, and allowing it to make good an inſufficiency
(fuppofing the date, in the ſecond cafe, to be really inſufficient) .
In the ſecond cafe, the accuſed may ſee the omiſſion as well as
another, and may confider of his defence againſt every alterna-
tive : in the firſt, he may be actually led to appear in court
with a defence not relevant to what will be brought against him.
The ſecond may be a hardſhip, the firſt is an injustice. And this,
even on the ſuppoſition that the rest of the indictment is to be
allowed in explanation : for we have no more right to ſuppoſe
that the true parts will correct the erroneous ones, than that the
erroneous parts will affect the conſtruction of the true ones. But
there is good reaſon to think that the ſufficient deſcription of one
ſentence may ſupply what is wanted in the inſufficient deſcription
of another, when inſufficiency is all.
But, perhaps, it will be held to be the better rule, that the re-
mainder of the indictment ſhould not be allowed in explanation .
It will then be admitted by all that a material error, or a material
infufficiency, ſhould be allowed to nullify the charge. The dif-
ference between the law and common opinion entirely relates
to what conſtitutes a material amount of one or the other. And
here it is impoſſible to bring the two together : for the law muſt
judge ſpecies, while the common opinion will never riſe above
the cafe before it. In the two inſtances, which by many will be
held equally abſurd, a great difference will be ſeen by any who
will imagine the two deſcriptions, in each cafe, to be put before
two different perſons. One is told that a man has ſtolen a ham ;
another that he has ſtolen a part of a ham. The firſt will think
he has robbed a proviſion warehouſe, and is a deliberate thief:
the ſecond may ſuppoſe that he has pilfered from a cook-ſhop,
poſſibly from hunger. As things ſtand, the two deſcriptions
may ſuggeſt different amounts of criminality, and different mo-
tives. But put the ſecond pair of deſcriptions in the fame way.
One perſon is told that a man perjured himself in the year 1846 ;
and another, that he perjured himself in the year of our Lord
254 On Fallacies.
1846. As things ſtand, there is no imaginable difference : for
there is only one era from which we reckon. The two deſcrip-
tions mean the ſame thing : nor can it even be ſaid that one is
complete and the other incomplete ; but only that one is leſs
incomplete than the other. The next queſtion might have been,
what lord was meant, our Lord Jesus Chrift, or our Lord the
King ? both being phrases of law. The anſwer will be, that the
number 1846 leaves no doubt which was meant. A very good
anfwer, certainly ; but equally conclufive as to the ſimple phrafe
' in the year 1846.' The firſt caſe is one in which the two de-
ſcriptions have a real difference of meaning : it is not ſo in the
fecond.
3. The petitio principii is one of the logical terms which has
almoſt found its way into ordinary life. It is tranflated by the
phraſe begging the question, that is, aſſuming the thing which is
to be proved. This is alſo called reasoning in a circle, coming
round, in the way of conclufion, to what has been already for-
mally aſſumed, in a manner expreſſed or implied. I ſhall reſerve
what I have to ſay on the justice of this tranflation, and take it
for the preſent as good .
Every collective ſet of premiſes contains all its valid conclu-
ſions ; and we may fairly ſay that, ſpeaking objectively of the
premiſes, the aſſumption of them is the aſſumption of the con-
cluſion ; though, ideally ſpeaking, the prefence of the premiſes
in the mind is not neceſſarily the presence of the conclufion. But
by this fallacy is meant the abſolute aſſumption of the ſingle con-
cluſion, or a mere equivalent to it, as a ſingle premiſe. If the
concluſion be ' Every X is Z ' and if it be formally known that
A and X are identical names, and alſo B and Z, then to aſſume
' Every A is B ' as a premiſe in proving ' Every X is Z ' would
be a manifeft petitio principii, or begging of the queſtion. But
even this must be faid hypothetically ; it is ſuppoſed fully agreed
between the difputants that the two identities are granted. Let
it be, otherwise, and there is no petitio principii : it is then fair to
propound A)B, which, if disputed, is to be proved, and afterwards
to reaſon as in A)B + B)Z = A)Z, X)A + A)Z = X) Z. Strictly
ſpeaking, there is no formal petitio principii except when the very
propoſition to be proved, and not a mere ſynonyme of it, is
aſſumed. This of course, rarely occurs : ſo that the fallacy to
 On Fallacies. 255
be guarded againſt is the aſſumption of that which is too nearly
the fame as the conclufion required. And then the fallacy is
nothing distinct in itſelf: but merely amounts to putting forward
and claiming to have granted that which ſhould not be granted.
When this is done, it matters little as to the character of the
fallacy, whether the undue claim be made for a propoſition which
is nearer to, or further from, the conclufion to be proved. When
proof is offered, the advancement of the conclufion in other words
is of courſe not petitio principii : when proof is not offered, the
aſſumption ofthat which (with other things proved) would prove
the conclufion, is a fallacy of the ſame character in all cafes,
There is an opponent fallacy to the petitio principii which, I ſuf-
pect, is of the more frequent occurrence : it is the habit of many
to treat an advanced propofition as a begging of the question the
moment they ſee that, if eſtabliſhed, it would eſtabliſh the quef-
tion. Before the advancer has more than ſtated his theſis, and
before he has time to add that he propoſes to prove it, he is
treated as a fophift on his opponent's perception of the relevancy
(if proved) of his firſt ſtep. Are there not perſons who think
that to prove any previous propoſition, which neceffarily leads to
the conclufion adverſe to them, is taking an unfair advantage ?
There is another caſe in which begging the queſtion may be
unjuſtly imputed. It ſhould be remembered that demonstrative
inference is not the only kind of inference : there is elucidatory
inference, recapitulatory inference, &c. A propofition may have
its aſſerted explanation preſented as a ſyllogiſm, the inference of
which, as demonftration, might well be called a reſult of petitio
principii. Say ' it never could have been doubted that men would
apply ſcience to the production of food. If there ſhould be any
heſitation about this, the explanation of man under the phraſe
which is excluſively characteriſtic of him, rational animal, would
remove it : the animal must have food, the rational being will
have ſcience. But it would be begging the queſtion to affert that
the fyllogifm of elucidation ' A rational animal is, &c.; man is,
&c ; therefore man is, &c.' is a demonſtration. And out of this
ariſes the fallacy of preſuming that an author meant demonstration,
when he can only be fairly conftrued to have attempted elucida-
tion of what he ſuppoſed would, upon that elucidation, be granted.
The forms of language are much the fame in the two caſes.
256 On Fallacies.

It has been obſerved that Ariftotle hardly ever uſes the phrafe
ἀρχὴν ἀιτεῖσθαι, principium petere : it is τὸ ἐξ ἀρχῆς and τὸ ἐν ἀρχῇ,
that which is (ought to come) out of, or is in,the principle. By
the word principium he distinctly means that which can be known
ofitself. He lays down five ways of aſſuming that which ought
to come out of a ſelf-known principle, of which begging the quef-
tion is the firſt. The others are aſſuming the univerſal to prove
the particular ; aſſuming a particular to help to prove the uni-
verſal ; aſſuming all the particulars of which the univerſal may
be compoſed ; and aſſuming ſomething which obviously demon-
ſtrates the conclufion .
Among the earlier modern writers, as far as I have ſeen them,
there is ſome diverſity in their deſcription of the petitio principii.
That the principium was meant to be the thing known of itself,
the ἀρχή of Ariftotle, as far as the introduction of the word is
concerned, ſeems clear enough. Was it not then by a mere cor-
ruption that it was frequently confounded with the conclufion,
the ' quod in principio quæfitum fuit ? ' Did not the fame in-
accuracy, * which confounds the τὸ ἐν αρχῇ of Ariftotle with the
ἀρχή itſelf, govern the change of the word ? Moſt writers take
the fallacy of the petitio principii as meaning that in which the
conclufion is deduced either from itſelf, or from ſomething which
requires proof more, or at least as much, ignotius aut æque igno-
tum. But ſome, in their definitions, and ſtill more in their ex-
amples, ſupport the following meaning, which I ſtrongly ſuſpect
to be the true derivation of the phrase, however the principium
and quod in principio might afterwards have been confounded with
one another. The philoſophy of the time conſiſted in a large
variety of general propofitions (principles) deduced from autho-
rity, and ſuppoſed to be ultimatelyderived from intrinfic evidence,
ſelf-known, or elſe by logical derivation from ſuch principles.
Theſe were at the command of the diſputant, his opponent could
not but admit each and all of them : the laws of diſputation de-
manded + the aſſent which the geometer requires for his poftu-
* Sir W. Hamilton of Edinburgh (notes on Reid, p. 761,) ſays that
principium is always uſed for that on which ſomething elſe depends .
† Does a traditional remnant of this convention ſtill linger in the notun-
frequent notion that a difputant is entitled to the conceffion of his principia ?
We ufed to hear 'You must grant me my firſt principles, elfe I cannot
 On Fallacies. 257

lates. Except when, now and then, literary ſociety was ſhaken
to its very foundations by a diſpute which affected any of them,
as a nominaliſt controversy or the like moral earthquake. The
moſt frequent ſyllogiſm was one which, having the form Barbara,
had a principium for its major, and an exemplum for its minor :
as in ' All men are mortal (principium) ; Socrates is a man (ex-
emplum) ; therefore Socrates is mortal. The petitio principii,
then, occurred, when any one, to prove his caſe, made it an ex-
ample of a principle which was not among thoſe received, with-
out offering to bring the former under the logical empire of the
latter. And fome writers define the fallacy as occurring fi
contingat in fyllogifmo principium petere ; where by principium
they mean the principle which generally occurs in the major pre-
miſe, and by their inſtances they clearly ſhow that they mean to
include nothing but the ſimple ſyllogiſm ofprinciple and example.
They would leave us to infer that if any one ſhould happen to
conſtruct a ſyllogiſm in which both premiſes are principles, one
or both not received, the inference, though denied by fimple
denial of one or both premiſes, would not be confidered as tech-
nically the petitio principii, which with them was, as it were,
petitio principii exemplum continentis.
It has often been aſſerted that all fyllogifm is a begging of the
queſtion, or a petitio principii in the modern ſenſe, an aſſumption
of the conclufion. That all premiſes do, when the argument is
objectively confidered, contain their conclufion, is beyond a doubt :
and a writer on logic does but little who does not make his reader
fully alive to this. But the phrafe, as applied to a good fyllo-
giſm, is a miſapprehenfion of meaning : for its definition refers
it to what is aſſumed in one premise. The moſt fallacious pair
of premiſes, though expreſſly conſtructed to form a certain con-
cluſion, without the leaſt reference to their truth, would not be
aſſuming the queſtion, or an equivalent. But a further charge
has been made against the ſyllogiſm, namely that very often the
conclufion, ſo far from being deduced from the principle, is
actually required to deduce it : that for inſtance, in ' All men are
argue. ' Cardinal Richelieu's anſwer to his applicant's ilfaut vivre, namely,
Je n'en vois pas la néceſſité, had ſomething of inhumanity in it : but, as
applied to the Mais, Monfieur, il faut ſe diſputer of the preceding affumption,
it would generally be quite the reverſe.
S
 ‫ر‬
On Fallacies .
mortal ; Plato is a man ; therefore Plato is mortal' we do not
know that Plato is mortal becauſe all men are mortal, but that
we need to know that Plato is mortal, in order to know that it
is really true that all men are mortal. There is much ingenuity
in this argument : but I think a little confideration, not of the
fyllogifm, but of how we stand with reſpect to the fyllogifm, will
anfwer it.
When we say that A is B, we do not merely mean that the
thing called A is the thing called B : if we ſpoke of objects as
objects, it would not matter under what name, and ' A is B ' would
be no other than ' B is B ' and the very propoſition itſelf would
be ofits own nature a mere identity, an affertion that what is, is .
It seems to me that between objects, thus viewed, there can nei-
ther be propofitions nor fyllogifms. A may remind us of a thing
as ſuggeſting one idea to our minds ; B of the ſame thing as
ſuggeſting another : and the propoſition ' A is B' then aſſerts that
the two ſtates of our mind are from the ſame external ſource.
Our logic, in wholly ſeparating names from objects, and dealing
only with the former, makes a fort of ſymbolic repreſentation of
the diftinction between ideas and objects.
Now the objection above ſtated to the fyllogifm appears to me
to be founded upon thinking of the object, as if it had no names.
Suppoſe all things marked, each with every name which can be
applied to it. Undoubtedly then, each one marked man will
have the mark mortal upon him, and ſome the mark Plato, it
may be and by the time all the marks are put on, and to a per-
fon who is ſuppoſed to be immediately cognizant of the fimul-
taneous exiſtence of two or more marks on the fame thing, it
would be an abſurdity to attempt any fyllogifm at all. What
coexistence of marks could there be which he must not be ſup-
poſed to have noted in making the induction neceſſary for a uni-
verſal propoſition. When he collected the elements of All men
Plato
are mortal ' he saw man among the reft and ſet it down. But
ſuppoſe that his knowledge is not acquired, as to different marks,
all at once: but that each coincidence of marks is to be a ſepa-
rate acquifition to his mind. Then he does not know, by the
time he has found out that ' All men are mortal ' whether Plato
be mortal or not. Plato may be a ſtatue, a dog, or a book written
 On Fallacies. 259
by a man of that name. Plato does not carry man with it : his
major tells him nothing about Plato, until he has the minor,
' Plato is a man ' and then, no doubt, he has abſolutely acquired
the conclufion Plato is mortal.' The whole objection tacitly
aſſumes the ſuperfluity of the minor ; that is, tacitly affumes we
know Plato to be a man, as ſoon as we know him to be Plato.
Grant the minor to be ſuperfluous, and no doubt we grant the
neceffity of connecting the major and the concluſion to be ſuper-
fluous alſo. Grant any degree of neceſſity, or of want of necef-
ſity, to the minor, and the ſame is granted to the connection of
the major and conclufion.
In the preceding cafe, the ſyllogiſm is looked upon as one of
communication, by the authors of the objection ; while at the
ſame time it is tacitly aſſumed that the minor does not commu-
nicate : Plato, by virtue of our acquaintance with the name, is
taken to be a man .
Moreover, it is to be noted that the propofition uſed in argu-
ment, whether to ourſelves or to others, is very frequently not
ſo much the mere attribution of one idea to another, as a decla-
ration that pro hac vice the idea contained in the more extenſive
term is all that is wanted, and that the differences which con-
ſtitute the ſpecies are not to the purpoſe. Or (page 234) it is
the diminution of the comprehenfion which is neceſſary, and the
increaſe of extenſion is onlycontingent. It is ſtripping the com-
plex idea of the unneceſſary parts, to prevent only what is requi-
fite. Thus any one who will aſſert that, in the Mofaic account,
no animal life whatever was destroyed by flaughter before the
deluge, muſt be convinced by being reminded that an antedilu
vian (Cain ) killed Abel who was a man and therefore an animal.
With the petitio principii may be claſſed (for it might alſo be
referred to other fallacies) caſes of the imperfect dilemma. Sup-
poſe we say ' Either M or N muſt be true : if M be true, Z is
impoſſible ; if N be true, Z is impoffible ; therefore Z is impof-
ſible. ' Now if the disjunctive premiſe ought to have been ' ei-
ther M or N or Z is true,' here would have been almoſt an ex-
preſs petitio principii. For example, ſay ' A body muſt either be
in the ſtate A or the ſtate B ; it cannot change in the ſtate A ;
it cannot change in the ſtate B ; therefore, it cannot change at
all.' Now, if the alternative A or B be neceſſary, the correct
260 On Fallacies.
ſtatement may be ' A body muſt either be in the ſtate A, or in
the ſtate B, or in the ſtate of tranfition from one to the other.'
Of this kind is the celebrated fophifm of Diodorus Cronus, that
motion is impoffible, for all that a body does, it does either in
the place in which it is, or in the place in which it is not, and it
cannot move in the place in which it is, and certainly not in the
place in which it is not. Now, motion is merely the name of
the tranſition from the place in which it is (but will not be) to
that in which it is not (but will be). It is reported that the in-
ventor of this ſophiſm ſent for a furgeon to ſet his diſlocated
ſhoulder, and was anſwered that his ſhoulder could not have been
put out either in the place in which it was, or in the place in
which it was not ; and therefore, that it was not hurt at all.
4. The ignoratio elenchi, or ignorance ofthe refutation, is what
we ſhould now call anſwering to the wrong point : or proving
ſomething which is not contradictory of the thing aſſerted. It
may be confidered either as an error of form or of matter ; and
it is, of all the fallacies, that which has the widest range. Such,
for instance, as the caſe of a writer I have read, who admits that
certain evidence, if given at all, would prove a certain point ; and
admits that ſuch evidence has been given : but refuſes to admit
the point as proved, becauſe the evidence was given in anſwer to
objections, and in aſecond pamphlet. The pleadings in our courts
oflaw, previous to trial, are intended to produce,out ofthe varieties
of ſtatement which are made by parties, the real points at iſſue ;
ſo that the defence may not be ignoratio elenchi, nor the caſe the
counter-fallacy, which has no correlative name, but might be
called ignoratio conclufionis. If a man were to fue another for
debt, for goods fold and delivered, and if defendant were to reply
that he had paid for the goods furnished, and plaintiff were to
rejoin that he could find no record of that payment in his books ;
the fallacy would be palpably committed. The rejoinder, fup-
poſed true, ſhows that either defendant has not paid, or plaintiff
keeps negligent accounts ; and is a dilemma, one horn of which
only contradicts the defence. It is plaintiff's buſineſs to prove
the fale, from what is in his books, not the abſence of payment
from what is not ; and it is then defendant's buſineſs to prove
the payment by his vouchers.
It is commonly faid that no one can be required to prove a
 On Fallacies. 261

negative, and often that no one can prove a negative. There is

much confufion about this : for any one who proves a poſitive,
proves an infinite number of negatives. Every thing that can be
proved to be in St. Paul's Cathedral at any one moment is fairly
proved not to be in more places than I can undertake to enume-
rate. What is meant is, that it is difficult, and may be impoffible,
to prove a negative without proving a poſitive. Accordingly,
when the two fides of the queſtion conſiſt of a poſitive and nega-
tive, the burden of proof is generally confidered to lie upon the
perſon whoſe intereſt it is to eſtabliſh the poſitive. This being
understood, it is ignoratio elenchi to attempt to transfer the charge
of proving the negative to the other party. But this rule is by
no means without exception : there are many departures from it
in the law, for example, though not under the moſt logical
phraſes. For instance, a homicide, as ſuch, is confidered by the
law a murderer, unleſs, failing juſtification, he can prove that he
had no malice. Here, in the language of the law, the homicide,
ſuppoſed unjustifiable, is in itſelf a preſumption of malice, which
the accuſed is to rebut. It is not true, in point of fact, that fuch
preſumption exiſts on the mere caſe of homicide, independent of
the manner of it : if the law will conſult its own records, it will
find that, for one homicide with malice of which it has had to
take cognizance, there are dozens at least, done in heat of blood,
and called manſlaughters. But the caſe ſtands thus ;-the alter-
natives are few, ſo that proving the negative of one, which the
accuſed is called on to do, can be done by proving the affirmative
one out of a ſmall number. There are but malice, heat of blood,
miſadventure, insanity, &c. to which the action can be referred.
Of theſe few things, it is easier for the accuſed to eſtabliſh ſome
one out of ſeveral, above all when motive is in queſtion (ofwhich
only himself can be in poſſeſſion of the moſt perfect knowledge)
than it is for the profecutor to eſtabliſh a particular one. And
the principle on which he is called on to eſtabliſh a negative ( or
rather another poſitive) is that the burden of proof fairly lies on
the one to whom it will be by much the eaſieſt. The proof of
a negative, then, being as eaſy as, in fact identical with, the
proof of one of the poſitive alternatives, ſuch proof may, from
the circumstances, lie upon a diſputant, particularly when the
number of the alternatives is few. But the negative proof, a
262 On Fallacies.

very different thing, is of its own nature hardly attainable, and

therefore hardly to be required. A book has been miſlaid ; is it
in one room or the other ? If found in the ſecond room, there
is proof of the negative as to the firſt : and almost any one who
can read can be truſted to ſay, on his own knowledge, that in a
certain room there is a certain book. But to give negative proof
as to the firſt room, it must be made certain, firſt, that every
book in the room has been found and examined, ſecondly, that
it has been correctly examined. No one, in fact, can prove
more than that he cannot find the book : whether the book be
there or not, is another queſtion, to be ſettled by our opinion of
the vigilance and competency of the ſearcher, Controverfialiſts
conftantly lay too much ſtreſs on their own negative proofs, on
their I cannot find, even as to caſes in which it is palpably not
their intereſt to find .
Somewhat akin to the preceding is the conſtant fallacy of con-
troverfialiſts, conveyed in their ſtrong afſertion of the reſults of
their own arguments. Few can bear to admit that there is a
queſtion for others to decide ; and after ſumming up both fides,
to ſeparate the points which the reader is to pronounce upon.
They must decide for him, and thus act both counſel and judge :
probably becauſe their arguments are not ſo convincing to their
own minds as they wiſh them to be to the reader's. They prove,
at the utmoſt, their own conviction that they have the right ſide :
but the thing to be proved is that ſuch conviction is well founded.
They know the maxim Si vis me flere, dolendum est primum ipfi
tibi, and think it will hold good of the reaſon, as well as of the
feelings : as it will, to ſome. The conſequence is, that the deli-
berate reader ſuſpects them, and feels inclined rather to differ
than agree : he will not dance to a writer who pipes too much.
Just as " I'll tell you a capital thing," ſets the hearer upon avoid-
ing laughter, and gives him notice to try ; ſo ' I intend to give
moſt unimpeachable proof,' puts the judicious reader upon look-
ing for inadmiſſible aſſumptions, and he is ſeldom allowed by
fuch writers to look in vain. But, if the difputant who begins by
declaring his intention to be irreſiſtible, be ſuſpicious, the one
who ends by announcing that he is ſo, is abſolutely ſelf-convicted.
If it be very clear, why ſhould he ſay it ? Does he tell his reader
that he must remember to diftinguiſh the black letters from the
 On Fallacies . 263
white paper, or does he print at the top of the book ' keep this
ſide uppermoft ?" Theſe things (eſſential as they are) he really
does leave to the reader : but he dares not trust the latter to find
out (though he ſays it is as clear as black and white) that his ar-
guments are ſo ſtrong and ſo good, that nothing but wilful dif-
honesty, or hopeleſs prejudice, can refift their force.
Another common form of the ignoratio elenchi, lies in attri-
buting to the concluſion aſſerted ſome ultimate end or tendency.
Thus, an argument in favour of checking the power of the
Crown is called Jacobiniſm ; of an increaſe of that power, abſo-
lutiſm : though the argument propoſed may be found, indepen-
dently of its propoſer's wiſhes. This is a caſe in which the reſult
of the method is justifiable, though the method is wrong. Many
readers will remember the advice given by an old judge to a
young one, ' Give your judgments without reaſons ; moſt likely
your decifions will be right ; and it is just as likely that your
reaſons will be wrong.' This advice ſhould be followed by many
of thoſe who judge or decide arguments. The propoſer is of a
known opinion, which gives him a ſtrong bias towards the con-
cluſion of the argument. He is a witneſs (page 205), and the
effect upon the mind of the receiver is to be that of the united
argument and testimony. The teftimony is, in the receiver's
mind, of a low order ; the propoſer is a radical, and the receiver
is of opinion that a radical would pick a pocket : or elſe, perhaps,
the propofer is a tory, and the receiver is of the belief that a tory
must have picked a pocket. Theſe opinions may be right or
wrong ; but they exiſt : and there is certainly no formal fallacy
in admitting them, as affecting the testimony, to fubtract from
the probability of the truth of the conclufion. But there is a
formal fallacy, a decided ignoratio elenchi, in throwing all the in-
difpofition to receive upon the invalidity of the argument.
There is a much more culpable form of the ſame ſpecies.
If ſuch a conclufion were admitted, it would lead to ſuch and
fuch another conclufion, which is not to be admitted. In quef-
tions of abſolute demonftration, this proceſs is found : if B be
certainly falſe, and if it be the neceſſary conſequence of A, then
A muſt alſo be falſe. But it is unfound when it takes the form,
' I believe B to be falſe ; I believe it to follow from A ; there-
fore I aſſume a right to diſbelieve A whatever evidence may be
264 On Fallacies.

offered for it ? This fallacy is ſufficiently expoſed in page 209.

There is a tradition of a Cambridge profeſſor who was once
aſked in a mathematical diſcuſſion ' I ſuppoſe you will admit that
the whole is greater than its part, and who answered, ' Not I,
until I ſee what uſe you are going to make of it.' This was no
doubt the extreme caſe ; the more ordinary one ariſes in a great
meaſure from the great fallacy of all, the determination to have
a particular conclufion, and to find arguments for it. Obſerve a
certain perſon who is led on by a wily opponent in converſation :
nothing is preſented to him except what his reaſon fully concurs
in, and no inference except what is indiſputable. At a ſudden
turn of the argument, he ſees a favourite conclufion, which he
cares more for than for all the reaſonings that ever were put
together, upſet and broken to pieces. He conſiders himself an
ill-uſed man, entrapped, ſwindled out of his lawful goods ; and
he therefore returns upon his ſteps, and finds out that ſome of
the things which he admitted when he did not ſee their con-
ſequences, are no longer admiſſible. Neither he nor the oppo-
nent has the leaft idea of the nature of probable arguments, and
of their oppofition : both proceed as if the train of reaſoning were
either demonſtration or nothing. The concluſion, formed perhaps
upon teſtimony, which is more likely to be a guide to truth for
the mind in queſtion than any appreciation of argument which
that mind could make, must, according to the maxims of the age,
be referred to argument, and argument only. The perpetual and
wilful fallacy of that mind is the determination that all argument
ſhall ſupport, and no argument ſhall ſhake, the concluſion. If
there were only a diſtinct perception of another ſource of con-
viction, ſo ſtrong that ordinary argument can neither materially
weaken, nor materially confirm it, there would be ſenſe in the
conclufion ; ſenſe, becauſe there is truth. Right or wrong, fuch
is the fource of moſt convictions in, perhaps, moſt minds : ſuch
ſource ought therefore to be acknowledged. It would be an ex-
cellent thing, if, in any disputed matter, thoſe who are better
ſatisfied by authority of the truth of one ſide of the conclufion
than of the validity of argument in general, would avow it, keep
their own fide, and let others do the fame. But here is the diffi-
culty : the perſons who ſhould avow ſuch a ſtate of mind are as
much diſpoſed to make converts as others : they do not like to
 On Fallacies. 265
debar themſelves from diſſemination of their opinions. Accor-
dingly they propound their beſt arguments, be they what they
may, as what ought to produce all the conviction which them-
ſelves feel. On this point ſee page 194.
The whole claſs of argumenta ad hominem, having ſome refe-
rence to the particular perſon to whom the argument is addreſſed,
will generally be found to partake of the fallacy in queſtion. Such
are recrimination and charge ofinconſiſtency, as, ' You cannot uſe
this aſſertion, becauſe in ſuch another caſe you oppoſe it.' But
if the original argument itſelf ſhould be a perſonal attack, then
ſuch a retort as the preceding may be a valid defence.
In many fuch argumenta ad hominem, it is not abſolutely the
ſame argument which is turned against the propoſer, but one
which is afſſerted to be like to it, or parallel to it. But parallel
cafes are dangerous things, liable to be parallel in immaterial
points, and divergent in material ones. A celebrated writer on
logic afferts, that no one who eats meat ought to object to the
occupation of a ſportſman on the ground of cruelty. The parallel
will not exiſt until, for the perfon who eats meat, we ſubſtitute
one who turns butcher for amusement. There is, or was, a vul-
gar notion that butchers cannot fit on a jury. Suppoſe that ſuch
a law were propoſed, on the ground of the habits arifing from
continual infliction of death. Would it really be a counter-argu-
ment that men who eat meat have the ſame animus and are liable
to acquire the ſame habits. It is contended (juſtly or not) that
a defire to take life for ſport is a cruel deſire ; to answer that
thoſe who eat fleſh from which life has been taken by others have
therefore alſo cruel defires, ought to be called arguing a dicto
fecundum quid ad dictum fecundum alterum quid. The matter is
clear enough. Cruelty of intention (the thing in queſtion) muſt
be ſettled by our judgment of the circumſtance in which the
ſport confifts. A person who ſeeks bodily exerciſe and the ex-
citement of the chaſe, and who can acknowledge to himself that
his object is gained on the birds which he miſſes, as well as upon
thoſe which he hits, even if thoughtless, cannot be faid to act
with cruelty of intention. But the ſportſman, as he calls himself,
who collects his game in one place, merely that he may kill,
without exerciſe, or feeling of ſkill, is either culpably thoughtleſs,
or elſe a ſavage, who delights in the infliction of death. Let any
 266 On Fallacies.
man aſk himſelf, whether in the event of his being called upon to
vote for a perfectly abſolute ſovereign, he would feel much con-
cerned to inquire whether the candidate was or was not a ſportf-
man of the firſt kind : and then let him aſk himself the ſame
queſtion with reſpect to the ſecond.
The moſt amusing, and perhaps the most common, example
of the ignoratio elenchi, is the taking exception to ſome part of
an illuſtration which has nothing to do with the parallel. The
word illuſtration (though it mean throwing light upon a thing)
is uſually confined to that fort of light which is derived from
ſhowing a proceſs of difficulty employed upon an eafier cafe.
The firſt fallacy may be committed by the illuſtrator. He has
before him the ſubject matter of the premiſes, their connexion in
the proceſs of inference, and the reſult produced. Either may
be illustrated ; thus, if it be doubtful whether ſuch premiſes
may be employed, the illuſtrator may throw away his mode of
connexion, and chooſe another : if the proceſs of inference be
doubtful, he may chooſe other premiſes : and ſo on. Buthe may
illuſtrate the wrong point : and this is a fallacy very common to
teachers and lecturers. The greatest difficulty in the way of
learners is not knowing exactly in what* their difficulty conſiſts ;
and they are apt to think that when ſomething is made clear, it
must be the ſomething. I am of opinion that the examples
given of fyllogifms in works of logic are examples of wrong
illuſtration. The point in queſtion is the form, the object is to
produce conviction of the form, of its neceſſary validity. If the
ſtudent receive help from an example ſtated both in matter and
form, the odds are that the help is derived from the plainneſs
of the matter, and from his conviction of the matter of the con-
clufion. If this be the cafe, he has not got over his difficulty.
Many learners are puzzled to ſee that ' Every Y is X' is not a
neceſſary conſequence of Every X is Y. If the want of con-
* Every learner, in every fubject, ſhould accuſtom himself to endeavour
`to ſtate the point of difficulty in writing, whether he want to ſhow the re-
fult to another or not. I wish I had kept a record of the number of times
which I have inſiſted on this being done, previously to undertaking the ex-
planation, and of the proportion of them in which the writer has acknow-
ledged that he saw his way as foon as he attempted to ask the road in precife
written language. That proportion is much more than one half. Truly
faid Bacon, that writing makes an exact man .
 On Fallacies. 267
nexion be eſtabliſhed by an inſtance, as by appealing to their
knowledge that every bird is not a gooſe, though every gooſe be
a bird, their knowledge of the propoſition is not logical. The
right perception may, no doubt, be acquired by reflection on in-
ſtances : but the minds which are beſt ſatisfied by material in-
ſtances, are alſo thoſe which give themſelves no further trouble.
The illuſtration being ſuppoſed correct, there is more than one
fallacious mode of oppofing it. Some perſons will diſpute the very
method of illustration of form, in which the ſame mode of infer-
enceis appliedtoeaſier matter ; but theſe are mere beginners,hardly
even entitled to a name which ſuppoſes the poſſibility of progreſs.
Others will deny the analogy of the matter, and theſe there is no
means of meeting: for illuſtration is ad hominem, and the per-
ception of it cannot be made purely and formally inferential : a
denier of the force of an illustration is inexpugnable as long as
he only denies. But when he attempts more, when he indicates
the point in which the illuſtration fails, he very often falls into the
error of attacking an immaterial point. If any one were to con-
tend (as ſome do) that it is unlawful to take the life of any ani-
mal, he might be aſked what he would say if Guy Faux had
trained a pigeon to carry the match to the vault, would it have
been lawful to ſhoot the bird on its way or not ? There are not
a few who would think it an anſwer to ſay that he could not
have trained the pigeon, or that pigeons were not then trained
to carry.
5. The fallacia confequentis (now very often called a nonfe-
quitur) is the ſimple affirmation of a conclufion which does not
follow from the premiſes. If the ſchoolmen had lived in our
day, they would have joined with this the affirmation of logical
form applied to that which wants it, a very common thing among
us. A little time ago, either the editor or a large-type correſpon-
dent ( I forget which) of a newſpaper imputed to the clergy the
maintenance of the ' logic ' of the following as ' conſecutive and
without flaw. This was hard on the clergy (particularly the
Oxonians) for there was no middle term, neither of the conclud-
ing terms was in the premiſes, and one negative premiſe gave a
poſitive conclufion. It ran thus,

Epiſcopacy is of Scripture origin .

268 On Fallacies.

The church of England is the only epiſcopal church in Eng-

land,
Ergo, the church eſtabliſhed is the church that ſhould be ſup-
ported.

Many caſes offend ſo ſlightly that the offence is not perceived.

For inſtance ' knowledge gives power, power is defirable, there-
fore knowledge is defirable ' is not a fyllogifm ; there is no mid-
dle term. It is a forites, as follows, knowledge is a giver of
power, the giver * of power is the giver of a defirable thing, the
giver of a defirable thing is defirable, therefore knowledge is de-
firable .'
It ſhould be noted, however, that the copula ' gives ' reſem-
bles ' is greater than ' (page 5) and is an admiſſible copula in in-
ferences with no converſion, provided that ' A gives B and B gives
C,' implies ' A gives C.' The ſame may be faid of the verbs to
bring, to make, to lift, &c. And many of theſe verbs are, by
the unſeen operation of their having the effect of is in inference,
often ſupplanted by the latter verb in phraseology. Thus we
fay ' murder is death to the perpetrator ' where the copula is
brings ; ' two and two are four ' the copula being ' have the
value of ' &c. But this practice may lead to fallacies, as above
ſhown : which must be avoided by attention to the claſs of verbs
which communicate their action or ſtate, ſuch as make, give,
bring, lift, draw, rule, hold, &c. &c. All theſe verbs are applied
to denote the cauſe of the ſeveral actions : ſo, to give that which
gives, or to bring that which brings, is to give or to bring. The
boy who was faid to rule the Greeks becauſe he ruled his mo-
ther, who ruled Alcibiades, who ruled the Athenians, who ruled
the Greeks, would have been correctly ſaid ſo to do, if the mat-
ters of rule had been the ſame throughout.
6. The non caufa pro cauſa. This is the miſtake of imagin-
ing neceſſary connexion where there is none, in the way of cause,
confidered in the wideſt ſenſe of the word. The idioms of lan-
guage abound in it, that is, make their mere expreſſions of phe-
nomena attribute them to apparent cauſes, without intent to
aſſert real connexion. Thus we say that a tree throws a ſhadow,

* Becauſe power is defirable. See page 115, as to this ſtep.

 On Fallacies. 269
to deſcribe that it hinders the light. When the level of a billiard
table is not good, the favoured pocket is ſaid to draw the balls.
A particular caſe of this fallacy, which is often illuſtrated by
the words poft hoc, ergo propter hoc, is the concluſion that what
follows in time follows as a conſequence. When things are ſeen
together, there is frequently an aſſumption of neceſſary connexion.
There is, of courſe, a preſumption ofconnexion : if A and B have
never been ſeen apart, there is probability (the amount of which
depends upon the number of inſtances obſerved) that the removal
of one would be the removal of the other. It is when there is
only one inſtance to proceed upon that the aſſumption falls under
this fallacy ; were there but two, inductive probability might be
faid to begin. The fallacy could then conſiſt only in eſtimating
the probability too high.
As may be ſuppoſed, the non caufa pro caufa ariſes more often
from mere ignorance than any other fallacy. To take the two
inſtances that I happened to meet with nearest to the time of
writing this page ;-Walpole, remarking onthe uniform practice
among the old writing-maſters of putting their portraits at the
beginning of their works, remarks that theſe men ſeem to think
their profeffion gives poſterity a particular intereſt in their fea-
tures. Probably they did not think about it : the uſage of the
day prevented any man from being chargeable with undue va-
nity who exhibited his phyſiognomy, and most of the writing
masters were themselves engravers, and either did their own por-
traits, or more probably made uſe of their acquaintance with the
more celebrated engravers for whom they did the under drudgery,
toget themſelves done on eaſy terms. Again, Noble (in his con-
tinuation of Granger) remarks that Saunderſon had ſuch a pro-
found knowledge of music, that he could diftinguiſh the fifth
part of a note. The author did not know, firſt, that any perſon
who cannot diftinguiſh leſs than the fifth part of a note to begin
with, ſhould be bound over to keep the peace if he exhibit the
leaſt intention of learning any muſical inſtrument in which in-
tonation depends upon the ear ; and ſecondly, that if Saunderſon
were not ſo gifted by nature, knowledge of muſic would no more
have ſupplied the defect, than knowledge of optics would give
him ſight.
The fallacia plurium interrogationum conſiſts in trying to get
270 On Fallacies.

one anſwer to ſeveral queſtions in one. It is ſometimes uſed by

barrifters in the examination of witneſſes, who endeavour to get
yes or no to a complex queſtion which ought to be partly anſwered
in each way, meaning to uſe the anſwer obtained, as for the whole,
when they have got it for a part. An advocate is ſometimes
guilty of the argument à dicto fecundum quid ad dictum fimpli-
citer : it is his buſineſs to do for his client all that his client might
honestly do for himself. Is not the word in Italics frequently
omitted ?. Might any man honestly try to do for himself all that
counſel frequently try to do for him ? We are often reminded of
the two men who ſtole the leg of mutton ; one could ſwear he
had not got it, the other that he had not taken it. The counſel
is doing his duty by his client ; the client has left the matter to
his counſel. Between the unexecuted intention of the client,
and the unintended execution of the counſel, there may be a
wrong done, and, if we are to believe the uſual maxims, no
wrong doer. The answer of the owner of the leg of mutton is
ſometimes to the point, Well, gentlemen, all I can fay is, there
is a rogue between you.' That a barriſter is able to put off his
forenfic principles with his wig, nay more, that he becomes an
upright and impartial judge in another wig, is curious, but cer-
tainly true.
The above were the forms of fallacy laid down as moſt eſſen-
tial to be ſtudied by those who were in the habit of appealing to
principles ſuppoſed to be univerſally admitted, and of throwing
all deduction into ſyllogiſtic form. Modern diſcuſſions, more
favourable, in ſeveral points, to the discovery of truth, are con-
ducted without any conventional authority which can compel
preciſion of ſtatement : and the neglect of formal logic occa-
ſions the frequent occurrence oftheſe offences againſt mere rules
which the old enumeration of fallacies ſeems to have conſidered
as ſufficiently guarded against by the rules themſelves, and ſuf-
ficiently deſcribed under one head, the fallacia confequentis. For
example, it would have been a childish mistake, under the old
ſyſtem, to have aſſerted the univerſal propoſition, meaning the
particular one, becauſe the thing is true in moſt caſes. The rule
was imperative: not all must be some, and even all, when not
known to be all, was fome. But in our day nothing is more
common than to hear and read aſſertions made in all the form,
 On Fallacies. 271
and intended to have all the power, of univerſals, ofwhich no-
thing can be faid except that most of the caſes are true. If a
contradiction be aſſerted and proved by an inſtance, the answer
is ' Oh ! that is an extreme cafe ' But the aſſertion had been
made of all caſes. It turns out that it was meant only for ordi-
nary cafes ; why it was not ſo ſtated must be referred to one of
three cauſes ;-a mind which wants the habit of preciſion which
formal logic has a tendency to foſter, a defire to give more
ſtrength to a conclufion than honeftly belongs to it, or a fallacy
intended to have its chance of reception .
The application of the extreme cafe is very often the only teſt
by which an ambiguous aſſumption can be dealt with : no won-
der that the aſſumer ſhould dread and protest against a proceſs
which is as powerful as the ſign of the croſs was once believed
to be against evil ſpirits. Where anything is aſſerted which is
true with exceptions, there is often great difficulty in forcing the
aſſertor to attempt to lay down a canon by which to diftinguiſh
the rule from the exception. Every thing depends upon it : for
the queſtion will always be whether the example belongs to the
rule or the exception. When one caſe is brought forward which
is certainly exception, the affertor will, in nine caſes out of ten,
refuſe to fee why it is brought forward. He will treat it as a
fallacious argument against the rule, instead of admitting that it
is a good reason why he ſhould define the method of diftinguiſh-
ing the exceptions : he will virtually, and perhaps abſolutely, de-
mand that all which is certainly exception ſhall be kept back,
ſimply that he may be able to aſſume that there is no occafion to
acknowledge the difficulty of the uncertain caſes.
The uſe of the extreme caſe, its deciſive effect in matters of
demonſtration, may furnish preſumption as to what it is likely to
be in matters of afferted near approach. As in the following in-
ſtance. It seems almoſt matter of courſe, when ſtated, to thoſe
who have not ſtudied the ſubject of life contingencies, that the
proper value of a life annuity is that of the annuity made certain
during the average exiſtence of ſuch lives as that ofthe annuitant.
That if, for example, perſons aged 22 live, one with another,
40 years, an office which receives from every ſuch perſon the
preſent value of forty payments certain, will, without gain or loſs,
in the long run, be able to pay the annuities. If this be (as was
272 On Fallacies.

ſtoutly contended by ſome writers of the last century) a univerſal

truth, it will hold in this extreme caſe. Let there be two per-
fons, one of whom is certain to die within a year from the grant
(and therefore never claims anything) and the other of whom is
certain to live for ever. It is clear that the value of an annuity
to both is o + the value of a perpetual annuity. But the average
life ofboth is eternal : one perpetual duration makes the average
of any ſet in which it is, perpetual. Hence by the falſe rule the
value is two perpetual annuities, or just double of the truth.
We might ſuppoſe that moſt perſons have no idea of a uni
verſal propoſition : but uſe the language, never intending all to
fignify more than most. And in the fame manner principles are
ſtated broadly and generally, which the aſſertor is afterwards at
liberty to deny under the phrase that he does not carry themfo
far as the inſtance named. It would not do to avow that the
principle is not always true : ſo it is ſtated to be always true, but
not capable of being carried more than a certain length. Are
not many perſons under ſome confufion about the meaning of
the word general? In ſcience it always has the meaning of uni-
verfal : and the ſame in old Engliſh. Thus the catechism of
the church of England aſſerts that there are two facraments
which are generally neceſſary to ſalvation : meaning neceſſary for
all of the genus in queſtion, be it man, Chriſtian, member of the
church, or any other. But in modern and vernacular Engliſh,
general means only usual, and generally means uſually.
A great deal of what is called evaſion belongs to this head,
or to that of the ignoratio elenchi, as the ſophiſt anſwers. The
advocates, for instance, of the abſolute unlawfulneſs of war never
tell, unleſs preſſed, what they think of the case of reſiſtance to
invaſion. Is the country to be given up to the firſt foreigner
who chooses to come for it ? Sometimes the extreme caſe comes
into play : ſometimes the aſſertion that no one will come ; which
is irrelevant as to the queſtion what would be right if he did
come .

Among amuſing modern evaſions are ' There is no occafion to

confider that' and ' I do'nt confider it in that point of view.
Any one who watches the manner in which men defend their
opinions will frequently ſee ' A is B and B is C, therefore A is
C' anſwered, not by denial of either premiſe, but by ' that is not
 On Fallacies. 273
the proper point of view' or ' I don't ſee it in that light. This
ſhould be called the confufion between logic and perſpective.
The denial of one univerſal is often made to amount to, or to
paſs into, the aſſertion of the oppoſite, or fubcontrary, univerſal.
This craving after general truths, the moſt manifeft fault of the
old logicians in their choice of premiſes, did not expire with them.
Bacon ſays ' the mind delights in ſpringing up to the moſt gene-
ral axioms, that it may find rest. Many perſons are defirous of
' ſettled opinions,' which is well ; unleſs by ſettled opinions they
mean univerſal, as is often the cafe. That ſome are and ſome
are not is no ſettlement : it makes every caſe require examina-
tion, to ſee under which it falls. And with the above we may
couple the tendency to believe that refutation of an argument is
proof of the falſehood of its conclufion, and that a falſe conſe-
quence must be a falſe propoſition. Hence it ariſes that ſo many
perſons dare not give up any argument in favour of a propofition
which they fully believe : they think they abandon the propo-
ſition.

It ſometimes happens that an aſſertion is made, which it is

difficult to ſuppoſe can be anything but a caſe of a univerſal pro-
poſition : and yet the aſſertor takes care not to make his pro-
poſition univerſal, but perſiſts in the particular cafe. A logician
in our day has aſſerted that when Calvin ſays that all officers of
the church ſhould be elected by the people, he must be under-
ſtood as ſpeaking in reference to deacons only, becauſe the
aſſertion is made in the chapter on deacons. If it had been
roundly ſtated that all univerſal propoſitions are to have their.
univerſes limited by the headings of the works or chapters in
which they occur-for instance, that the affertion that all men
are mortal, occurring in a history of England, is to be taken as
made of Engliſhmen only-there would have been at leaſt no
ambiguity. But as it is, we are left to ſurmiſe whether this be
meant, or whether the propoſition be to apply to Calvin only, or
to Reformers only, or to men whoſe names begin with C, &c .
The odds are that the application of a univerſal propofition will
be dictated by the heading of a chapter : but the extent to which
a premiſe is afferted as true is not to be judged of by that to which
it is wanted for use : and the leſs, the nearer we go to the day of
the old logicians. T
274 On Fallacies.
Wrong views ofthe quantity of a propofition are as frequent
as any fallacies. Some, meaning moſt, andfome, meaning few,
are frequently confounded. This is the neceſſary conſequence
of the nature of human knowledge, in which we can but rarely
form a definite idea of the proportion which the extent ſpoken of
bears to the whole. It is part ofthe value of the mathematical
theory of probabilities, that the mind is accuſtomed to the view
of reſults drawn from perfectly definite ſuppoſed cafes; as uſeleſs,
it may be, in themſelves, as many ofthe queſtions in a book of
arithmetic, but nevertheleſs good for exerciſe. It is not ſurpriſf-
ing that fallacies about quantity ſhould be capable of moſt ſtrik-
ing expoſure in queſtions concerning meaſurable quantity, that
is, in queſtions of mathematics : nor that there ſhould be claſſes
of fallacy of which it is difficult to illuſtrate the detection by any
other inſtances. What can be more clear, for example, to ordi-
nary apprehenfions than the broad ſtatement that ' of things of
the fame kind, that which is ſometimes right must be better than
that which is always wrong.' But a little confideration will
ſuggeſt that what is always wrong may be as good as that which
is ſometimes right, if we do not know how to diftinguiſh the
caſes in which the latter is right : and alſo that what is not
much wrong, generally, may be more uſeful than that which is
moftly very wrong, when it is not abfolutely right. A watch
which does not go is right twice a day : but it is not ſo uſeful as
one which does go, though very badly.
To give an account of all the fallacies which depend upon
wrong notions of quantity would require much ſpace, and more
aſſumption ofmathematical knowledge in my reader than is con-
ſiſtent with my plan. But I may mention the miſtaken uſe of
abſolute terms and notions in queſtions of degree. There can
be, a diſputant will ſay, but a right and a wrong ; and if this be
not right, it is wrong. Many perſons will announce that their
watches are quite right, abſolutely at the true time, to a ſecond :
and will end by giving the time which was ſhown when they
looked, as being accurately that of the inſtant at which they
announce it. The proverb Frustra fit per plura, quod fieri po-
teft per pauciora contains an inaccuracy of degree : a bargain
which coſts twenty ſhillings and is worth fifteen, is not twenty
ſhillings loft, but only five, though the vexation of the party
overreached will seldom ſuffer him to fee this.
 On Fallacies. 275
Proverbs in general are liable to this mistake. They are often
uſed in exactly the ſame manner as the firſt principles ofthe old
logicians. In fact, remembering that theſe firſt principles were
bandied from mouth to mouth till they were perfectly proverbial,
as we now call it, among the learned ; and obſerving the appli-
cation of our modern proverbs, as made by the maſs of thoſe
who have not profited by mental diſcipline, we may ſee that the
faults of the ſchoolmen are only thoſe of the ordinary human
mind. It is hard indeed if there be a purpoſe which a proverb
cannot be found to ſerve : it is a univerſal propofition of no very
definite meaning, ſanctioned by uſage, having the appearance of
authority, and capable of ſtretching or contracting like Prince
Ahmed's pavilion. One only is allowable- In generalibus latet
error : this deſtroys all the reſt, and then, when cloſely looked
at-commits ſuicide.

All mistakes of probability are eſſentially mistakes of quantity,

the ſubſtitution of one amount of knowledge and belief for an-
other. It is often difficult to convey a proper notion of the
degree of force which is meant to be given ; and ſtill more ſo
to retain it throughout the whole of a diſcuſſion. A perſon be-
gins by ſtating an explanation as poſſible, or probable enough to
require confideration, as the caſe may be. The forms of language
by which we endeavour to expreſs different degrees of probability
are easily interchanged ; ſo that, without intentional diſhoneſty
(but not always) the propoſition may be made to ſlide out of one
degree into another. I am satisfied that many writers would
ſhrink from ſetting down, in the margin, each time they make a
certain aſſertion, the numerical degree of probability with which
they think they are justified in preſenting it. Very often it hap-
pens that a conclufion produced from a balance of arguments,
andfirst presented with the appearance ofconfidence which might
be repreſented by a claim of ſuch odds as four to one in its favour,
is afterwards used as if it were a moral certainty. The writer
who thus proceeds, would not do ſo if he were required to write
in the margin every time he uſes that conclufion. This would
prevent his falling into the error in which his partiſan readers are
generally fure to be more than ready to go with him, namely,
turning all balances for, into demonſtration, and all balances
against, into evidences of impoffibility .
One ofthe great fallacies of evidence is the diſpoſition to dwell
276 On Fallacies.

on the actual poſſibility of its being falſe : a poſſibility which must

exiſt when it is not demonſtrative. Counſel can bewilder juries
in this way till they almoſt doubt their own ſenſes. A man is
ſhot, and another man, with a recently diſcharged piſtol in his
hand, is found hiding within fifty yards of the ſpot, and ten mi-
nutes of the time. It does not follow that the man ſo found
committed the murder : and cafes have happened, in which it has
turned out that a perſon convicted upon evidence as ſtrong as the
above, has been afterwards found to be innocent. An aſtute
defender makes theſe caſes his prominent ones : he omits to men-
tion that it is not one in a thouſand against whom ſuch evidence
exiſts, except when guilty.
All the makers of ſyſtems who arrange the univerſe, ſquare the
circle, and fo forth, not only comfort themselves by thinking of
the neglect which Copernicus and other real diſcoverers met
with for a time, but ſometimes fucceed in making followers.
Theſe laſt forget that for every true improvement which has
been for fome time unregarded, a thouſand abſurdities have met
that fate permanently. It is not wiſe to toſs up for a chance of
being in advance of the age, by taking up at hazard one of the
things which the age paſſes over. As little will it do to deſpiſe
the uſual track for attaining an object, becauſe (as always hap-
pens) there are ſome who are gifted with energies to make a road
for themselves. Dr. Johnſon tells a ſtory of a lady who ſerioufly
meditated leaving out the claſſics in her fon's education, becauſe
ſhe had heard Shakſpeare knew little of them. Telford is a
ſtanding proof (it is ſuppoſed by ſome) that ſpecial training is
not eſſential for an engineer.
The diſpoſition to judge the prudence of an action by its reſult,
contains a fallacy when it is applied to ſingle inſtances only, or
to few in number. That which, under the circumſtances, is the
prudent rule of conduct, may, nevertheleſs end in ſomething as bad
as could have reſulted from want of circumſpection. But upon
dozens ofinſtances, ſuch a balance would appear in favour of pru-
dence as would leave nodoubt in favourofthe rule of conduct, even
in the inſtances in which it failed. The fallacy conſiſts injudging
from the reſult about the conduct ofone who had only the previous
circumſtances to guide him. You acted unwiſely, as is proved
by the reſult,' is a paralogiſm, except when it implies ' You did,
 On Fallacies. 277
as it happens in this inſtance, take a courſe which did not lead to
the defired reſult.' Take a ſtrong cafe, and the abſurdity will be
ſeen. A chemiſt makes up a preſcription wrongly, and his cuf-
tomer leaves him for another : this other, ſo it may happen, makes
it up ſtill more wrongly, and poiſons the patient. Who would
venture to ſay that he acted unwiſely, as is proved by the reſult,
in leaving the tradeſman whom he knew to be careleſs, for another
of whom he knew no harm. The only way in which blame can
be imputed, is when it can be faid ' You acted unwiſely, in not
finding out, as you might have done, that the reſult which has
happened is the one which was likely to happen. One reſult
proves very little as to the ſuperior wisdom of the courſe which
produced it ; ſeveral may give a preſumption of it, and the greater
the number, the greater the preſumption.
So little is this thought of, that the common phrase, ' I acted
for the beſt,' meaning originally ' I acted in the manner which
under the circumſtances, appeared likely to lead to the beſt re-
ſults,' very often loſes its proper meaning, and is uſed as ſynony-
mous with ' I acted with good intentions.'
Theſe, and many other points, I can only flightly touch on :
I will proceed to notice a few other cauſes of error.
And firſt, of equivocations of ſtyle. I have before referred to
ſuch a phenomenon as the alteration of a good fyllogifm into a
bad one, to make the ſentence read better. But nothing ever
reads well (for a continuance) except the natural current of a
writer's thought. I ſhould like it to be the law of letters, that
every book ſhould have inſerted in it the printer's affidavit, ſetting
forth the number of verbal eraſures in the manufcript, fair copies
being illegal. It would be worth at least one review.
There is a wilful and deliberate equivocation, which it is
ſuppoſed the age demands. It is the uſe of ſynonymes, or fup-
poſed ſynonymes, to prevent the ſame word from occurring twice
in the fame paſſage. So far is the neceſſity ofthis practice recog-
nized, that there are few printing-offices in London, the readers
of which do not query the ſecond introduction of any word which
prominently appears twice. And then the author obeys the hint,
ſtrikes out one of the offenders, ſticks in a dictionary equivalent,
and would have been content if the printer's reader had done it
for him. And fo he writes a goodstyle. To be ſure, he does not
 278 On Fallacies.

ſay what he meant, exactly ; for ſynonymes are ſeldom or never

logical equivalents : but what is that to elegance of expreffion?
The demand for non-recurrence of words arifes from the pub-
lic (I beg its pardon) not knowing how to read. If, when a
word occurs twice, the proper emphaſes were looked for, and
obſerved, there would be nothing offenſive about the repetition.
It is the reader who makes one and one into two, by giving both
units equal value. Take this ſentence from Johnſon, (the firſt
I happened to light on, in the preface to Shakſpeare), and read
it firſt as follows :-" He therefore indulged his natural diſpoſition :
and his diſpoſition, as Rymer has remarked, led him to comedy :"
* and then as follows " He therefore indulged his natural difpofi-
tion ; and his diſpoſition, as Rymer has remarked, led him to
comedy." This reading is what the context requires, and the ill
effect of the repetition is next to nothing. Take the next ſen-
tence :-" In tragedy he often writes, with great appearance of
toil and ſtudy, what is written at last with little felicity : but in
his comic ſcenes he ſeems to produce, without labour, what no
labour can improve." Theſe were the firſt inſtances I found,
from a chance opening of the Elegant Extracts, purpoſely choſen
as a miſcellany. The laws of thought generally dictate this rule,
that the firſt occurrence of a word is the more emphatic of the
two : the leſſon of experience is, that a writer who prevents re-
currence by the use of the dictionary of ſynonymes, is a good
ſtyle-maker for none but a bad reader, and may very poffibly be
a good arguer for none but a bad logician. Of course, I ſhould
not deny that recurrence of both word and emphasis is a defect,
if it be frequent.
The confufion between the means and the end, and putting
one in the place of the other, is well enough known in morals :
but there is a correſponding tendency to forget the distinction
between the principle which is to be acted on, and the rule of
action by which adherence to that principle is ſecured. A refe-
rence to the derived rule is in all reſpects as good as one to the
firſt principle, between parties who underſtand both, and the
connexion between them. But thoſe who underſtand the rule
only, are apt to forget that a rule may or may not be the true
expreffion of a principle, according to the circumſtances in which
it is propoſed to apply it. If, indeed, it were of univerſal appli
 On Fallacies. 279
cation, thoſe who do and those who do not underſtand the prin-
ciple might be on the fame footing as to ſecurity: but there are
few fuch rules.
The preceding caution may be applied in all departments of
thought, in law and in logic, in morals and in arithmetic. It is
impoſſible, for instance, to ſtate the rule of three in ſuch a man-
her as easily to include the caſes in which it ſhall apply, and ex-
clude thoſe to which it does not. To ſay that it must be uſed
where the fourth quantity, the one fought, is to be a fourth pro-
portional to the three which are given, though correct, ſtill leaves
it open to inquiry what are the caſes in which this condition is to
be ſatisfied : and many caſes might be, and are propoſed, in which
the inquiry is not eaſy to a beginner. In law, there are not only
rules, but rules for their application. To an unlearned ſpectator,
particularly in the courts ofequity, in which the advocate addreſſes
a judge, and not a jury, the argument takes that technical form
which makes many perſons think that the whole law is, at beſt,
only arbitrary rule. It may be that ſome of thoſe who there
addreſs the court can make nothing better of it : and juſt as there
are arithmeticians, and good ones too, who are but the flaves, and
never the maſters, of their proceſſes, ſo there may be advocates,
and even judges, who have not one element of the legiſlator in
them. But there are enough of a higher ſpecies.
The great art of uſing rules is to apply them in aid, and not
in contravention, of the principles which they are intended to
embody. A rule may have exceptions, it is ſaid ; but this is
hardly a correct ſtatement. A rule with exceptions is no rule,
unleſs the exceptions be definite and determinable : in which cafe
the exceptions are exclufions by another rule. The parallel is
perfect between rules and propofitions (page 143). Thus, ' All
Europe, except Spain and Portugal ' is a univerſal propoſition ;
but ' All the ſtates of Europe except two ' is a particular one. A
rule which applies to all ſtates except Spain and Portugal is a
rule : but a rule which applies to all except two (unknown) is no
rule. When it is ſtated, in ordinary language, that every rule is
ſubject to exception, it is meant, for the moſt part, that the cir-
cumſtances under which adherence to the rule gains the object,
are thoſe which most frequently occur, and that the circumſtances
under which adherence to the rule would defeat the object are
280 On Fallacies.
rare. If this were remembered, much confufion would often be
ſaved. We want a word which ſhall ſo far expreſs rule, that it
ſhall imply that which will generally ſucceed, without thenotion
of obligation which accompanies that of rule, and which perpetu-
ally miſleads. We want, in fact, the rule niſi of the courts, which
is to be a rule unleſs cauſe be ſhown against it : and which will,
in moſt caſes, be ultimately made abſolute, but is not abſolute
from the beginning.
The common mistake is, that the rule niſi is an abſolute rule,
and that therefore it may be ſubſtituted for its leading object or
firſt principle, and that even the very words which expreſs that
object gained, may be taken as equally expreſſive of fatisfaction
of the rule, and vice verſa. For instance, it is commonly ſtated
that the rule by which a diſcoverer is determined, is publication ;
that he who firſt publiſhes the diſcovery, is to he held the diſco-
verer ; one lapſe more, and it is ſaid that he is the diſcoverer ;
yet one more, and it will be ſaid that the publication is the diſco-
very. The very remarkable circumſtances attending the recent
diſcovery of the planet Neptune, involving points of peculiar in-
tereſt and delicacy, have cauſed this rule to be much diſcuſſed, and
have brought out every variety of ſtatement of it. The thing to
be determined is the actual truth of the question, the real hiſtory
ofthe human mind with regard to it. No one has a right under
any rule, no matter what its authority, nor by whom impoſed, to
ſubſtitute the thing which is not, for the thing which is, or the
leſs probable for the more probable. If philoſophers were to at-
tempt, by a law of their own framing, to ſubſtitute the conven-
tional reſult for the real one, the common ſenſe of mankind would
difpute their authority, and reverſe their deciſion. The firſt rule
(niſsi) is undoubtedly that the firſt printer is the firſt publiſher, the
ſecond, that the firſt publiſher is the diſcoverer. Theſe will, un-
leſs cauſe be ſhown against them, be made abſolute in every caſe.
A notion which is very prevalent, namely, that the firſt publiſher
has therefore the rights of the diſcoverer, is as incorrect as that
the firſt printer is therefore the first publisher. To take the cur-
rent language, one would ſuppoſe that printing one hundred
copies would be held better than circulating one thousand in
manuſcript, and that even though the firſt publiſher could be
proved to have plagiariſed, he has ſtill the rights ofdiſcovery.
 On Fallacies. 281

Just as (page 244) early notions make laws ofliteral interpre-

tation ſuperſede thoſe of intended meaning, ſo, in the earlier
ſtages of law, rules are often made to over-ride the principles on
which they profeſs to be founded, and to defeat truth and common
ſenſe. There is more excuſe here than there would be in a
queſtion of ſcience, for peace and convenience are main objects
of law, and it may be that rigid adherence to a rule, as a rule, at
a certain avowed ſacrifice of truth and justice, may be the only
practicable means of preventing a larger ſacrifice of both. In old
times, the rule of affiliation, Pater est quem nuptiæ demonstrant,
was held ſo abſolutely, that the husband of the mother would be
the legal father, though the two had been confined in two diffe-
rent jails a hundred miles apart for twelve months preceding the
birth of the child. The modern law has made this rule to be no
more than it ought to be, namely, one which must hold unleſs
the contrary be proved.
It is not uncommon, in difputation, to fall into the fallacy of
making out conclufions for others by ſupplying premiſes. One
ſays that A is B ; another will take for granted that he must be-
lieve B is C, and will therefore conſider him as maintaining that
A is C. But it may be that the other party, maintaining that
A is B, may, by denying that A is C, really intend to deny that
B is C. In religious controversy, nothing is more common than
to repreſent ſects and individuals as avowing all that is eſteemed
by thoſe who make the repreſentation to be what, upon their pre-
miſes, they ought to avow. All parties ſeem more or leſs afraid
of allowing their opponents to ſpeak for themselves. Again, as
to ſubjects in which men go in parties, it is not very uncommon
to take one premiſe from ſome individuals of a party, another
from others, and to fix the logical concluſion of the two upon
the whole party : when perhaps the conclufion is denied by all,
fome of whom deny the firſt premiſe by affirming the ſecond,
while the reſt deny the ſecond by affirming the firſt. Any ſect
of Chriſtians might be made atheists by logical conſequence, if it
were permitted to join together the premiſes of different ſections
among them into one argument. This is a fallacy which, how-
ever common, could easily be avoided, and would be, if thoſe
who uſe it cared for anything but victory. But there is another
form of the fame, which every one is ſubject to, and which it is
282 On Fallacies.

not ſo eaſy to perceive. It is that of drawing upon our former

ſelves for the premiſes which are to guide us for the time being.
Conclufions remain in our minds long after the grounds on which
they were formed are abandoned : and it may happen that one
premiſe of an argument will ſtill have force, when the very rea-
fons on which the ſecond premiſe is now admitted are contra-
dictory of thoſe which once induced us to admit the firſt. Thus
many who have learnt to advocate the legal toleration of opinions
which they ſtill believe, by force of education, to be abſolute
crimes againſt ſociety, are logically the advocates oftoleration of
crime ; whereas, the arguments which they have learned to think
valid for the firſt premiſe, ought, if worth anything, to teach
them to deny the ſecond. I have myſelf heard from one mouth
in one converſation (ofcourſe not in one part of it) that all fins
against the Creator are ſins againſt ſociety, that all fins against
ſociety ought to be punished by ſociety, that certain opinions then
named are fins against the Creator, and that it is the height of
injustice to puniſh any one for his opinions.
In printed controversy, the ſtatement of the oppoſite opinion
or afſſertion may be made by deſcription without citation (by
chapter or page), by deſcription with citation, or by quotation
with or without deſcription. The firſt is not allowable. The
preſumption is ſtrong that a person who oppoſes an opinion, im-
putes an error, or makes a charge, upon the writings of another,
is bound at least to cite, in a manner which cannot be miſtaken,
the part of thoſe writings to which he refers. There are writers
who refer defcriptively and even commentatively, putting the
reference of citation, and thus (as Bayle ſays Moréri conſtantly
does) lead the reader to ſuppoſe that the words of their para-
phrafe and comment are thoſe of the paſſage itſelf. I do not fee
that quotation is obligatory, though highly defirable : but the
reader muſt remember, when there is only citation, that it is not
the author cited who ſpeaks, but the person who brings him for-
ward . It is a man's own account of his own witneſs : with the
advantage of an apparent offer of enabling the reader to go and
verify the ſtatement for himself. If the citer be honest, the paf-
ſage in queſtion exiſts : if judicious, it is to the effect ſtated.
Conſequently, whenever the citer's honesty or judgment is ex-
preſsly in queſtion, no mere citation is admiſſible.
 On Fallacies. 283
When citations are few they ought perhaps to be quotations :
when they are many, it may be impracticable to make them ſo.
But extenſive citation ought to be encouraged. Lazy readers
do not like it : they are not pleaſed to have a power of verifica-
tion offered of which they do not mean to avail themselves ; and
they would rather, in caſe of being miſled, have to throw the
blame upon the author than upon their own non-acceptance of
the offered means of verification. Accordingly, they expreſs
their diſguſt at " pages loaded with references." But the more
diligent readers conſider every citation as a boon. At the ſame
time it is to be remembered that there are writers who, relying
on the common diſinclination to verify, add a large number of
citations, and give the appearance of a ſtrong body of authorities,
which are often nothing to the purpoſe, and ſometimes not taken
from actual examination, but copied from other writers.
Perhaps the greatest and most dangerous vice of the day, in
the matter of reference, is the practice of citing citations, and
quoting quotations, as if they came from the original ſources,
instead of being only copies. It is in truth the reader's own
fault if he be taken in by this, or by the falſe appearance of au-
thority juſt alluded to ; for it is in his own power to certify him-
ſelf of the truth : though there may be difficulty when the cita-
tions are many, or when some ofthem are from very rare books.
Honesty and policy both demand the expreſs ſtatement of every
citation and quotation which is made through another ſource.
If a perſon quote what he finds of Cicero in Bacon, it ſhould be
' Cicero (cited by Bacon) ſays, &c. It has happened often
enough that a quoter has been convicted of altering his author,
and has had no answer to make except that he took the paſſage
from fome previous quoter.
Quotations are frequently made with intentional omiffion and
alteration. But no rule ought to be more inflexible than that
all which is within the marks of quotation ought to be a literal
tranſcript of the book quoted. Sometimes the omiſſion is made
becauſe part of the ſentence is unneceſſary, as the quoter thinks.
But this is just the point which he has no buſineſs to decide
without letting his reader know that he has decided it, which is
eafily done by the recognized mark of omiffion ( ..... ) If
a perſon would quote the Æneid for the antiquity of Carthage,
284 On Fallacies.
he has no buſineſs to write down, as from Virgil, Urbs antiqua
fuit Carthago : ' it ſhould be ' Urbs antiqua fuit ..... Carthago,'
if he decide upon omitting ' Tyrii tenuere coloni.' In this caſe,
not only may the omiffion make the propoſition appear more
categorical than it is in the original, turning it from ' There was
an old city, Carthage, rather towards ' Carthage was an old city ; '
but a reader may chooſe to think that the omitted words qualify
the epithet, or even offer proof deſtructive of it. What if he
ſhould deny the antiquity of Tyre ? The omiſſion may (or may
not) be right, but the omiffion without notice, or fuppreffion, is
certainly wrong.
Moreover, it is dangerous to truth to ſhorten without notice,
inaſmuch as thoſe who quote the quotation will be apt to do the
ſame thing ; that is, thinking they have the whole paſſage, to
ſhorten it further. What this may end in, no one can predict :
but miſtakes have been brought about in this way quite as abſurd
as any that ever were made. It may reaſonably be ſuppoſed
that many very ludicrous errors ariſe thus. A good many years
ago, I ſucceeded, by means of a ſhortened quotation, put away
until it was wanted, in arriving at, and publiſhing, the conclufion
that Archimedes was once ſuppoſed to have been an ancestor of
Henry IV. of France. The real purport of the ſentence was that
he was ſuppoſed to have been an ancestor of the Sicilian martyr
St. Lucia, on whoſe day Henry IV. was born. It has happened
that A has been ſaid to have aſſerted in a ſecond book, that B
related the death of C, when the truth is that A ſaid in the firſt
book that B died many years before C (See the Companion to the
Almanack for 1846, page 27). I do not ſpeak of omiffions made
becauſe the part omitted would prove more than the quoter
likes : this of courſe is fraud.
Unjustifiable as unnoted omiffions may be, ſtill more ſo are
additions and alterations. Writers have ſometimes inſerted gloſſes
of their own, into the text which they quote, either as addition
or alteration. Explanatory additions may easily be made within
brackets [ ], which are understood marks of ſuch a thing : but
alterations are intolerable. But why, the reader may aſk, are
ſuch things inſiſted on ? Is not the ſimple rule, Be honest,
enough to include theſe and hundreds of things like them, with-
out detail ? To this I reply that within a twelvemonth before
 On Fallacies. 285
the time I write this, a clergyman, a man of high education and
character both, publiſhed a fermon in which he gave a verſe
from the Bible within marks of quotation, in which he wilfully
ſtruck out one word, and inſerted another, without notice : and
his fermon went through ſeveral editions, either without detec-
tion, or without that detection leading to ſucceſsful remonſtrance.
I do not ſuppoſe there was diſhoneſty here ; but rather the fol-
lowing reasoning ;- I am sure it was meant; therefore I may
ſtate that it was ſaid. Such reaſoning is one ofthe curſes of our
literature.
There is one alteration within the marks ofquotation which
may at firſt ſeem reaſonable : it is alteration ofgrammar to bring
the quoted phrases into connected Engliſh with the quoter's
context. As when a man ſays " I know" and another perſon,
quoting him, ſays " He knows." But it is ſurely just as eaſy
to put down He says " I know." There is often an alteration
ofemphasis in this adaptation of grammar, and generally an in-
troduction of irony : and it is the premierpas to ſomething worſe.
As far as I have ſeen, thoſe who do it as a matter of courſe, are
apt ſometimes to put their own paraphrafes under marks of quo-
tation. A writer ſhould ſuit his own grammar to that of his
quotation, and not the converſe.
Omiſſion of context, preceding or following the quotation,
may alter its character entirely : and this is one of the moſt fre-
quent of the fallacies of reference, both intentional and uninten-
tional. The only way to inſure full confidence is to give the egg
in its ſhell : that is, to begin at a point which clearly precedes
the immediate ſubject of quotation, and to continue until the
matter is as clearly paſt : to give a ſentence preceding and a
ſentence following the matter quoted for its own fake, diftin-
guiſhing the latter. This is not always concluſive : becauſe the
ſubject may be reſumed in a ſentence or two, or in another part
of the book. But it will inform the reader, in most cases, whe-
ther he is or is not likely to differ from the quoter as to the
meaning of the part quoted. And this refers particularly to quo-
tations of opinion : thoſe of fact may often be more briefly
treated with ſafety.
In quoting ancient authors, in caſes where the text is not no-
torious, the various readings ſhould be given, eſpecially when it
286 On Fallacies.
is an author whoſe texthas an indifferent reputation for accuracy.
Or ifthis cannot be done, the edition ſhould be cited. Shameful
things have occurred in controversy, by omiſſion of a part of the
ordinary text, which the quoter choſe to confider as an interpola-
tion, without choofing to confider that the reader ought to have
liberty to judge for himself on that point.
Among the cafes of indirect citation, ſhould be included that
in which a book is mentioned as exiſting, not on the authority
of the writer's own eyes,but on that of a catalogue. The num-
ber of nonexisting books which are entered in catalogues and
copied, as to their titles, into other works, is greater than any
one who has not examined for himself would ſuppoſe poſſible.
In thoſe who know this, confidence is destroyed ; and this ſome-
times affects queſtions ofopinion. I am toldthat Dugald Stewart,
who had a strong notion of the practical impoſſibility of pre-
ſenting Euclid in a ſyllogiſtic form, never would believe that it
had been done by Herlinus and Daſypodius. Such a work is
entered in catalogues : but I muſt ſay that the ſtate of catalogues
is ſuch that Stewart or any one elſe had full right to doubt of
any work, upon no other than catalogue evidence. The work
does exiſt, and I have a copy of it. But, ſeeing how matters
ſtand, no one has a right to declare that an old book ever was
written, without informing his reader on what fort of evidence
he relies.

CHAPTER XIV.

On the Verbal Description of the Syllogifm.

firſt attempt to expreſs the rela-
INpage 75,propoſitions
tions of I have madein alanguage which will make fyllo-
giſms capable of verbal deſcription, and the inference of their
conclufions matter of ſelf-evidence. It is desirable that this ſhould
be more fully done, and I accordingly renew the attempt, with
the beſt words of deſcription which I can find or make. Any
one who can ſuggeſt words which better convey the meaning to
himſelf, will find it eaſy to ſubſtitute them for those which I
have uſed.
 Description of the Syllogifm. 287
The conditions to be ſatisfied are, that the words ſhould have
as much imported meaning as poſſible, that every word and its
contrary ſhould have the connexion of contrariety well marked,
and that the verbal deſcriptions ſhould be capable of being easily
formed from the ſymbolic notation. As may be ſuppoſed, theſe
conditions are to ſome extent contradictory of each other : the
facrifice of either to the others is then to be made to the moſt
advantageous effect.
There are two ways in which it may be neceſſary to deſcribe
the fyllogifm. First, the one hitherto uſed throughout this work,
in which one concluding term is referred to the other by the in-
tervention of the middle term : what X is of Y, and what Y is of
Z, determine what X is of Z. Secondly, that in which the two
terms are referred to one another by comparison ofboth with the
middle term : what X and Z ſeverally are of Y determine what
X is of Z.
In the firſt mode, the middle term is mentioned, and its de-
ſcription is middle in the ſentence ; while the reference term is
understood in the predicate of each deſcription. Thus when we
fay ' a fubcontrary of a ſupercontrary is a fubidentical, it is that
a fubcontrary of afupercontrary (of Z) is a ſubidentical (of Z) ;
and thefupercontrary of Z is the middle term.
In the ſecond mode, the middle term is understood in the ſub-
ject, and the concluding terms in the predicate, of the defcrip-
tion of the fyllogifm. Thus when we say ' genus and ſpecies are
genus and ſpecies,' it means that two terms which are ſeverally
genus and ſpecies of the middle term (one entirely containing,
the other entirely contained in, the middle term) are genus and
ſpecies to one another (the firſt genus, the ſecond ſpecies).
Now it will be very easily ſeen, that the way to change the
firſt deſcription into the ſecond is as follows. Say the deſcrip-
tion runs thus, ' P of Q is R.' If Q be its own correlative, as
happens when Y and Z are convertibly connected, then ' P of
Q' merely becomes ' P and Q : ' but if have another, Qº, for
its correlative, then ' P of Q' becomes ' P and Qº. Again, if
R be its own correlative, its plural takes its place : but if R have
Rº for its correlative, it becomes ' R. and Rº. Thus ' fubcon-
trary of ſupercontrary is ſubidentical ' of the firſt mode, becomes
' fubcontrary and ſupercontrary are ſubidentical and ſuperidenti
288 On the Verbal Defcription
cal ' meaning that C and C' ofthe middle term are Di and D'
of each other. But ' fubcontrary ofſuperidentical is ſubcontrary '
becomes ' fubcontrary and ſubidentical are ſubcontraries.'
I need hardly ſay that ' P of Q is R' with reſpect to X in
terms of Z, must be read Q of Po is Ro' with reſpect to Z in
terms of X. This rule we have already uſed.
It is thus ſhown that it is only neceſſary to dwell on the firſt
mode ; and now ariſes the queſtion what words are to be em-
ployed in deſcribing the eight ſtandard propofitions. After a
good deal of confideration, I prefer to denote the univerſal rela-
tions by poſitive terms, and their contrary particulars by the cor-
reſponding negative ones : not without full perception of the
ſacrifice which enſues of the firſt condition above mentioned to
the third.

Thewords genus andſpecies immediately ſuggeſt themſelves to

denote the relation of Y to X and X to Y in X)Y. Theſe are
to be understood as employed up to their limit ; or the genus and
ſpecies may be coextenſive. For two names which have no-
thing in common, as in X.Y, I propoſe to say that they are ex-
ternals of each other. And for two names which have nothing
out of one or the other, as in x.y, that they are complements of
each other. Remember that complemental does not mean only
just complemental (which is contrary), but may be contrary or
ſupercontrary.
In X :Y, I call X a non-ſpecies of Y, and Y a non-genus of X.
Theſe words have not as much as I could wiſh of imported
meaning, nor are there any poſitive terms which I can propoſe
to ſupply their places. They appear as ſynonymous with not
entirely contained in and not containing the whole. In XY , let
X and Y be non-externals ; and in xy, let X and Y be non-com-
plements. Accordingly, in deſcribing what X is with reſpect to
Y, we have as follows, ſhowing the ſubſtitutions which occur in
reading the fyllogiftic ſymbols into this language.
A1 , ſpecies Oi, non-ſpecies.
A', genus O', non-genus.
Eı, external I., non-external.
E', complement I', non-complement.
If we confider genus and complement as larger terms, andspecies
 ofthe Syllogifm. 289
and external asſmaller ones, and if we put down each univerſal
followed by its two weakened particulars, writing firſt that which
is of the fame accent, we have
Univerſal. Firſt weakened form . Second weakened form.

A' Genus I' non-complement I non-external .

A Species I non-external I' non-complement.
E' Complement O' non-genus O non-ſpecies.
E External O non-ſpecies O' non-genus.
Thus it appears that the primary weakened form of a larger
name contains a larger name, and of a ſmaller a ſmaller : and
the contrary for the ſecondary forms. The words primary and
ſecondary do not refer to importance, but only to order of deri-
vation : thus A was in our table X)Y, weakened into XY,
before it became y)x, weakened into yx or xy.
The rules for forming particular fyllogifms by weakening uni-
verſal premiſes may now be repeated. In a univerſal fyllogifm,
ſubſtitute for thefirst premiſe and for the concluſion their primary
weakened forms, or for thesecond premiſe and for the conclufion
their fecondary weakened forms. In a ſtrengthened ſyllogiſm,
ſubſtitute for thefirst premiſe itsſecondary form, or for thesecond
premiſe its primary form .
I now write down the whole body of fyllogifms, that the rea-
der may exerciſe himself in the independent comprehenfion of
their meaning, and in afſſent to their inferences ; deducing the
particular fyllogifms from the univerſals only.
Univerſal and particular Syllogisms.
Symbol . Deſcription of X with respect to Z.
AAA Species of ſpecies is ſpecies.
LAI Non-external of ſpecies is non-external.
AI'I' Species of non-complement is non-complement.
A'A'A' Genus of genus is genus.
I'A'I' Non-complement ofgenus is non-complement.
A'LL Genus of non-external is non-external.
AEE Species of external is external.
IEO Non-external of external is non-ſpecies.
AO'O' Species of non-genus is non-genus.
U
290 On the verbal Deſcription
'A'E'E' Genus of complement is complement.
I'E'O' Non-complement of complement is non-genus.
A'OO Genus of non-ſpecies is non-ſpecies.
ELA'E External of genus is external .
OA'O Non-ſpecies of genus is non-ſpecies.
ΕΙΟ ' External of non-external is non-genus.
E'AE' Complement ofſpecies is complement.
O'AO' Non-genus of ſpecies is non-genus.
Ε'Ι'Ο Complement of non-complement is non-ſpecies.
EE'A External of complement is ſpecies.
ΟΕΙ Non-ſpecies of complement is non-external.
EOI' External of non-ſpecies is non-complement.
Ε'Ε Α' Complement of external is genus.
Ο'ΕΙ' Non-genus of external is non-complement.
Ε'Ο'Ι Complement of non-genus is non-external.

Strengthened Syllogisms.
AA'I' Species of genus is non-complement.
A'AL Genus of ſpecies is non-external.
ΑΕ'Ο ' Species of complement is non-genus.
Α'ΕΟ Genus of external is non-ſpecies.
EAO' External of ſpecies is non-genus.
Ε'Α'Ο Complement of genus is non-ſpecies.
ΕΕΙ' External of external is non-complement.
Ε'ΕΙ Complement of complement is non-external.

No perſon could propoſe to himself a better exerciſe in the

acquifition of command over language, than practiſing the de-
monftrations of theſe relations, or more properly their reduction
into ſpecific ſhowing, as to the matter of the inference, in what its
extent confifts. For instance, ' the complement of a non-com-
plement is a non-ſpecies ' : How, and by how much ? The non-
complement leaves ſomething which is neither in the term un-
derstood, nor in that non-complement. This, the complement
of that non-complement must fill up : and by this then, at least,
thecomplement of the non-complement is not in the term under-
ſtood, of which it is therefore ſo far non-ſpecies.
 of the Syllogifm. 291
In the preceding view, I have particularly confidered the con-
nexion between contrary forms, and the adaptation of language
to that connexion. But in the firſt derivation of the ſimple fyl-
logiſms (page 88) the univerſals were related, not to their con-
traries, but to their particular concomitants. I now proceed to
the confideration of this view, and to the juſtification, on ſelf-
evident principles, of the aſſertion that there is a real and ſtriking
affinity between the univerſal fyllogifm and its concomitants, as
AAA and O'AO', E'EA' and E'I'OI, &c.
The complex propofitions D., D', and C₁ contain each a uni-
verſal which, in common language, is generally confounded with
it, and a particular, the existence of which is therefore for the
moſt part ſuppoſed in thought to accompany the univerſal. The
remaining univerſal, E', is differently circumstanced: if we ſay
that X and Y complete the univerſe, we ſhould generally mean
that they only juſt complete it, and ſhould not think of the ſuper-
contrary relation, or of their overcompleting it. To be contained
but not to fill ; to contain with room to ſpare, or to overfill ; to
exclude and be excluded without completion ; and to exclude and
be excluded with completion (or to complete and be completed
without inclufion) ; are our moſt uſual ideas of the relations of
the extent of names .
The reduction of the complex propoſition to the ſimple uni-
verſal, when done by removal of the concomitant particular, is
in all caſes a lowering of the quantity, by the removal of an ex-
ceſs, as follows : -

Da means that X is contained in Y, and more is contained.

D' means that X contains Y, and contains more.
Ca means that X excludes Y, and excludes more.
C' means that X completes Y, and * more than completes .

Drop the ſecond clauſes, and Di, &c. are reduced to A., &c.
Drop the firſt clauſes, and it would seem as if we had ſtill the

* The alteration of grammar here ſeen is in deference to the word com-

plete, the beſt I can get. In this propoſition, the verb refers to the universe,
and it is X(joins in completing the univerſe)Y and joins in completing more
(than the univerſe) .
292 On the verbal Defcription
complex propofitions ; for more will contain its tacit reference to
that which it is more than. Let this tacit reference be dropped,
and then we have, instead of the whole complex propoſition, only
its particular. And this abandonment is actually made in com-
mon language, by what would be called perhaps a lax, but is a
very logical, uſe of the word more. There are more than fiſh on
the dry land,' would be perfectly intelligible, and not as implying
that there were any fiſh: ' he was actuated by more than the
motive, &c.' very often means ' other than the motive' &c.
Now, in the complex fyllogifm, as we have ſeen (page 81 ), the
exceſſive part of the conclufion (whence comes its ſecond clauſe,
its additive more) is the ſum of the exceſſive parts of the premiſes .
If one of the complex premiſes be deprived of its aſſertion of
exceſs, or lowered into a ſimple univerſal, the concluſion ſtill
remains, though not àfortiori, neceſſarily. This being done, the
valid excess of the concluſion depends upon the excess ofthe remain-
ingpremiſe; and the concomitant particular ſyllogiſm, confidered
as part of the mixed complex fyllogifm, is the expreſſion of this,
without the reſt. Finally, the exceſs may be uſed in the lax, or
non-correlative, ſenſe, and then the concomitant fyllogifm ſtands
by itſelf.
For example, OA'O may be read thus :-Confider O as
concomitant of A' in D'. ' X contains more than [ſomething
that is not in] Y ; Z contains X ; therefore, Z contains more
than [ſomething that is not in] Y.' If ' more than Y ' mean ' Y
and more,' this would be D'A'D'. Again, O'E.I' is ' more than
X [ſomething not X) is contained in Y ; Y excludes Z ; there-
fore, X excludes more than Z [ſomething not in Z] .' If ' more
than X ' were ' X and more,' &c.: this would be DEC . And
fo on for other cafes .
I now proceed to what I may call the quantitative deſcription
of the fyllogifm : by which I mean the expreffion of its caſes in
terms of the quantities only of its names and propofitions, leaving
the alternative of affirmation and negation to be ſettled by the
law of theſe quantities. My reaſon for the preſentation ofthe
ſyſtem in ſo many different points ofview will be obvious enough :
that which claims to be complete, muſt ſhow itſelf to containjuſt
the fame, and no more, as to reſults, whatever may be the prin-
ciple which is choſen as the basis of construction.
 ofthe Syllogifm. 293
Every propoſition, in ſpeaking of two names, ſpeaks of their
contraries, and (page 63) ofthe four terms, two direct and two
contrary, two are univerſal and two are particular. Since univer-
fal and particular are themſelves properly contraries, (for ' Every
X' is ' Xs, known to be all' and ' Some Xs ' are ' Xs, not known to
be all') let us ſignify the univerſal and particular forms of the
propoſition by V and v. Again, ſpeaking of a name, let its mode
of entry, univerſal and particular, be denoted by T and t. Writ-
ing down V( or v) applied to T( or t), T( or t) we can make eight
varieties, which give us the eight ſtandard forms applied to one
order, ſay XY ; as follows : -

A = V(Tt) A' = V( T) | E = V(TT) E' =V(tt)

=v
O = v (t ) I' = v (TT) I =v (tt)

Thus I' or xy, may be deſcribed as the particular in which

both terms are univerſal : for X and Y are both univerſal in xy,
or x : Y, or y : X. And v( TT) deſcribes it thus.
If, understanding the order to be XY, YZ, XZ, we write
down any three propoſitions, we make an attempt at a fyllogifm,
valid or not, as the caſe may be : as in

V(Tt).v(tt).V(tT) or VvV(Tt,tt,tT)

which must be ALLA'. It will aſſiſt the memory to obſerve that

ſub-ſymbols have VT or vt at the beginning, ſuper-ſymbols vT or
Vt. Alſo, that affirmatives have an even number of capitals
(none* or two) and negatives an odd number (one or three). A
univerſal and its particular concomitant have the ſame entries of
T and t, and contranominals have inverted modes of entry of
theſe letters. The convertibles have T in both places, or t : the
inconvertibles have T and t.
Firſt, it is unneceſſary to write down the term-letters of the
conclufion, for they must be taken from the premiſes, in every
caſe in which the conclufion is the ſtrongest that can be drawn
from the premises ; and our ſyſtem has no others (nor, indeed,

* The reader muſt here follow the mathematician in confidering o as an

even number .
294 On the verbal Description
has the Ariftotelian any other except Bramantip). Thus, TT,tt
being the term letters of the premiſes, ſtrike out the ſecond T
and the firſt t, which refer to the middle term, and Tt muſt be-
long to the conclufion. To prove this, obſerve that we know
that t in the premiſe cannot give T in the conclufion : therefore
T cannot give t ; for if, the term being Z, T gave t, then, put-
ting z properly in its place, t would give T, which it cannot.
Again, we know that the valid forms, as to propofitions, are
VVV, VVv, vVv, Vvv ; ſo that v occurring once only, muſt
come third, and V must come in the firſt pair. Further, in the
four term letters of the premiſes, VVV, vVv, Vvv, require Tt,
or tT, to come in the middle, while VVv alone requires TT,
or tt. Obſerve theſe laws, and every formation which can take
place under them leads to a valid fyllogifm. Putting dots to re-
preſent a blank place, we form the eight univerſal ſyllogiſms by
filling up the blanks in VVV(.. t,T .. ) and VVV(.. T,t .. ) ;
the eight ſtrengthened ſyllogiſms from VVv(.. T,T .. ) and
VVv(.. t,t ..) ; the eight particulars which begin with a univer-
fal from Vvv( .. t,T ..) and Vvv(.. T,t ..) ; and the eight par-
ticulars which begin with a particular from vVv(.. t,T .. ) and
vVv( .. T,t .. ). And, under the rules juſt given, we have no
other cafes .
Taking the preceding as a baſis, we might make the rules of
accentuation follow from it. For, ſince the firſt blank in our
ſymbol, and the firſt concluding term, muſt agree, and fince ac-
cents depend only on the first two letters in the ſymbol of a pro-
poſition, we may proceed as follows. Let K and L, each of
them, mean T or t, as the cafe may be, but with the proviſo that
what it means in either place it ſhall mean
ean in the other. Then,
in VVV(KT,tL,KL) and in vVv(KT,tL,KL), in which ſym-
bols ofconclufion are introduced, we ſee that the firſt and third
accents muſt agree, which is part of the direct rule. As to the
firſt and ſecond accents, they agree in the firſt inſtance above, if
K be t, which puts an even number of capitals in the firſt ſym-
bol VKT, or an affirmative propofition at the commencement :
they differ if K be T, which puts a negative propoſition firſt.
In the ſecond inſtance, they agree if K be T, which puts an
affirmative firſt, &c. I leave it to the reader to deduce the other
caſes of this rule, the inverſe rule, and alſo that premiſes give an
 ofthe Syllogifm. 295
affirmative, or a negative, conclufion, according as they have like
or unlike ſigns. And thus it will appear, that the ſymbolic rules
given in chapter V, are really expreſſions of the general rules of
quantity.
It will be obſervedthat the concomitant fyllogifms of a univer-
fal have the fame term letters as that univerſal, and only change
VVV into Vvv, or vVv. Alſo, that the inverted fyllogifms of
page 96 only invert the order of all the term-letters, and the
letters of the premiſes, when different.
Thus, EA'E, being VVV(TT,tT), its concomitants I'A'I'
and EOI', are vVv(TT,tT) and Vvv(TT,tT). But the in-
verted form AEE is VVV(Tt,TT). Contranominals have
different quantities in all the term-letters. The weakened forms
of a univerſal change the firſt premiſe letter and the firſt term
letter, or the ſecond of both. Thus, E.E'A, being VVV(TT,tt),
its weakened forms, OEM and EOI", are vVv(tT,tt) and
Vvv(TT,tT) .
The forms of the numerical fyllogifm (page 161) may be re-
covered by few and eaſy rules, in which the premiſes as they ſtand
determine the conclufion, as follows :-Let & be deſignated as
the number of X, and ' as that of x ; and ſo on. Let a term
of the conclufion be called direct when it is in the premiſe, and
inverſe when its contrary is in the premiſe. Then,
1. In every caſe, the concluſion has the ſum of the quantities
mentioned in the premiſes, as part of the expreffion of its quan-
tity.
2. For every inverſe term in the conclufion, the number of
its direct term appears in the quantity of the conclufion, fub-
tracted. Thus, x in a premiſe, with X in the conclufion, muſt
have - in the concluding quantity. But the direct terms of
the conclufion never introduce anything into the concluding
number.
3. When the entrances of the middle term are fimilar (YY,
or yy), the terms of the two forms of concluſion are both direct
and both inverſe, with fubtraction of the number of the middle
term in the former, addition of the number of its contrary in the
latter. Thus, yy gives -n' in the direct, + n in the inverſe
form.
4. When the entrances of the middle term are diſſimilar (Yy,
296 On the verbal Description, &c.
or yY), each form of conclufion has one direct and one inverſe
term ; and no number from the middle term enters the conclud-
ing quantity.
Thus, the conclufions from mxY + nYZ are immediately
written down as

(m + n-n)xZ and (m + n + n' -'-;)Xz:

while thoſe from mxY + n'yz are at once
(m + n' - ')Xz and (m + n' - ')xZ

There are relations exiſting between the forms of the ſyllo-

giſm which I have not confidered. For instance, the deſcription
of X with refpect to Z being that it is a ſpecies, (A1), the de-
ſcription of its contrary, x, is that it is a ſupercontrary, (E'). If
then we give the name of contradeſcriptives to A. and E' we find
that A' and E , I, and O', I' and Oi, are alſo contradeſcriptives .
The arrangement offyllogifms by contradeſcriptives, and the laws
of connexion thence reſulting, will be an eaſy exerciſe for the
ſtudent.
 APPENDIX .

I.

Account of a Controversy between the Author ofthis Work and

Sir William Hamilton of Edinburgh ; and
final reply to the latter.
HIS appendix contains an account of a controverfy in which ſome
T ofthe matters treated in the preceding work involved me with
Sir William Hamilton, Profeſſor of Logic and Metaphyfics in the Uni-
verſity of Edinburgh. It has produced four publications (to which I
ſhall refer as I, II, III, IV) namely :
I. ' Statement in anſwer to an aſſertion made by Sir William Hamil-
ton, Bart..... by Auguſtus De Morgan, ....
(London, octavo, R. and
E. Taylor, pp. 16, publiſhed April 30, 1847.)
II. A letter to Auguſtus De Morgan, Eſq..... on his claim to an
independent rediſcovery of a new principle in the theory of fyllogifm .
From Sir William Hamilton, Bart. Subjoined, the whole previous cor-
refpondence, and a poſtſcript in anſwer to Profeſſor De Morgan's " State-
ment,"" (London and Edinburgh, octavo, Longman and Co., Maclachlan
and Co. pp. 44, excluſive of ' Profpectus ' hereinafter mentioned : re-
ceived by me May 22, 1847.)
III. Letter from me to Sir W. Hamilton, dated May 24, publiſhed
in the Atheneum Journal of May 29.
IV. Letter from Sir W. Hamilton to me, dated June 2, publiſhed in
the fame Journal ofJune 5.
There are two queſtions involved, one concerning my character, the
other purely literary. The former ſtands thus. March 13 , Sir W.
Hamilton informed me by letter that (the Italics are his own words) to
him it is manifeft that for a certain principle I was wholly indebted to his
information, and that if I should give itforth as aspeculation of my own
(which I had done to himself, and meant to do, as he knew, and have
fince done, in print) I ſhould, even though recognizing always his pri-
ority, be guilty both of an injurious breach of confidence towards him and
offalse dealing towards the public. This hypothetical charge, and dero-
gatorysupposition ofwhich he may formerly have surmised the poſſibility
(ſuch are his ſubſequent qualifications of it) is unreservedly retracted at
the beginning and end of II : but it is frequently infinuated in the mid-
dle, by propofing things as difficult to be explained otherwise, by hint
that others may believe it, by hopes that they will not, by charges of
falfehood, &c. &c. For theformal charge is ſubſtituted imputation of
 298 Appendix.
lapſe ofmemory, intellectual confufion, &c. The following is the pro-
gramme of the firſt intended argument, (II. p. 4.)
' I confefs, that, for a time, I regarded your pretenfion, as an attempt
'at plagiarifm, cool as it was contemptible.
' From this view, feeling, information, reflection turned me ; and I
' now, Sir, tender you my fincere apology, for admitting, though founded
' on your own ſtatements, an opinion ſo derogatory of one, otherwiſe ſo
• well entitled to reſpect.
• In itſelf, this view was, to me, painful and revolting. The cha-
' racter, too, which you bear among your friends, I found to be wholly
' incompatible with a ſuppoſition ſo odious. You are repreſented as an
' active and able man, profound in Mathematics, curious in Logic, wholly
• incapable of intentional deceit, but not incapable of chronological mif-
' takes . Your habitual confufion of times is, indeed, remarkable, even
•from our correſpondence. Your dates are there, not unfrequently of
• the wrong month, and not always, even ofthe right year. With much
' acuteness, your works ſhow you deficient in architectonic power, the
' concomitant of lucid thinking ; and, that you are not guiltleſs of intel-
lectual raſhneſs is ſufficiently manifeſt, from your pretention to advance
• Logic, without having even maſtered its principles.'
With regard to the ſubſequent infinuation of a retracted charge, my
explanation (believing as I do, that Sir W. Hamilton always ſpeaks fub-
jective truth) is that his mind inſenſibly fell back to its old bias as he
felt that the ſubſtitute for his charge wanted ſtrength : my conclufion is,
that it is unneceſſary henceforward to notice any thing he may fay or
write on my character : and my determination is to act accordingly.
Sir W. Hamilton's pamphlet contains about a ſcore and a half of quo-
tations, on which hang ſundry jokes and ſneers, ſome ofthem at mathe-
maticians in general, and myself as one of the body. On theſe I ſhall
only ſay that my notions of the common ſenſe of controversy, and my
determination to perfiſt, generally, in the tone of reſpect to my oppo-
nent's learning and character which I have hitherto preſerved, would,
were there nothing elſe, prevent my adopting the habit ofwhich they
are ſpecimens. But as no man willingly ſtands an unreturned fire of
facetiæ withoutdefiring to prove that his forbearance does not ariſe from
want of ammunition, I will permit myſelf(diſclaiming the animus under
which ſuch things are usually written) juſt to ſhow that quotation, ap-
plication, alluſion, ſneer, joke, and fling at an opponent's ſtudies, are all
among the weapons which I could have employed, if I had thought
them worthy of my antagoniſt, or of thoſe whom I want to convince.
I might, for instance, have written ſomething like the following ;-
Among the affets of the old logicians, diſcovered when the ſchools
were ſwept out, there was found, as is well known, the queſtion Utrum
chimæra bombinans in vacuo poſſet comedere fecundas intentiones : a very
good title, as Curll would have faid, wanting nothing but a treatiſe
written to it. Now whether it be comedere, or whether the ſchoolmen
invented comèdere, Sir W. Hamilton, on whom their mantle has fallen,
has written the treatiſe, and ſucceſsfully maintained the affirmative. His
 Appendix . 299
notion that his communication could give any hint, is clearly and aptly
defcribed by chimæra, his ſtyle by bombinans, his proofby vacuum : and
thesecond intentions, above noticed, chewed up and given forth with
his firſt ones, are a practical example of the poſſibility ofthe Q.E.I. He,
or rather the bombinating chimæra which has perſonified itſelf in his
form, as the ἶλος ὄνειρος did in that of Neftor, is thus both retractor
and detractor. But though the tranfition from flops to folids generally
indicates convalefcence, yet, as here made manifeſt, the paſſage from
liquid to dental may be only the growing weakness, the periſcence, of the
cafe.
I affert the following documents to be all that are relevant with
reſpect to the literary part of the controversy. They are given at the
end of this appendix.
A is an extract from a communication of mine to the Cambridge
Philofophical Society, made before I received any communication what-
foever from Sir W. Hamilton. I affert it to contain a diſtinct an-
nouncement and uſe ofthe principleof quantification ofthe middle term,
be that middle term fubject orpredicate. On this point the reader is to
judge.
B is a communication from Sir W. Hamilton to me. The reader
is to judge firſt, whether it contain anything which is intelligible with
reſpect to any ſyſtem of fyllogifm ; secondly, whether, if it ſhould ſo con-
tain anything, that ſomething would have been information to me who
had written A, on ſome matter afterwards found in C.
C is the relevant part of an addition made by me to A, when the
latter came before me in proof. The reader is to judge firſt, whether C
contain anything more than an application ofA ; ſecondly, if fo, whe-
ther that ſomething more is derived from anything intelligibly hinted at
in B.
The only bare fact on which Sir W. Hamilton and myself are at
iſſue is this . I aſſert and maintain that the matter of C was written in
my poſſeſſion before I received B : Sir W. Hamilton holds me mif-
taken, and thinks he can prove from the correfpondence that in this
point my memory has failed. This I continue to treat as irrelevant :
for we are both agreed that the corpus delicti, if delictum there be, lies
in C containing fomething not ſubſtantially contained in A, but fuffi-
ciently hinted at in B. Any reader who thinks that C does contain
ſomething ſuggeſted by B which is not in A, may declare against the
correctneſs of my memory ; any one who thinks the contrary, will hold
it of no conſequence whether my memory on the disputed fact be good
or bad. With the frit reader I have no caſe : with the ſecond I have
all I think worth caring about.
Sir. W. Hamilton maintains my letters to be eſſential parts of the
cafe. They may become ſo, as foon as it is pointed out what C contains
which is hinted at in B, and not contained in ſubſtance or principle, in
A. When Sir W. Hamilton points out, by citation from C, what he
alleges to have been taken, and by citation from B, what he thinks it
has been taken from, and when I thereupon fail to produce equivalent
300 Appendix.
knowledge from A or elſe to expoſe the irrelevance ofhis citation from
B-then thoſe letters may become ofimportance. This he has not done,
though ſpecially challenged to do ſo : and when I come to diſcuſs III
and IV, it ſhall appear that he admits he has not done it.
I now give the beſt account I can of the origin of the dispute, pre-
miſing, that up to this 3d of September, 1847, I do not abſolutely know
what the ſyſtem is which I am charged with appropriating. There is a
ſyſtem which I think is most probably the thing in queſtion : but a fyf-
tem containing a defect of ſo glaring a character, that I will not attribute
it to Sir W. Hamilton, who defcribes his own as " adequately teſted
and matured" until he expreſſly claims it, or until I have the moſt indu-
bitable proof.
In the common, or Ariftotelian propoſition, the quantities of the ſub-
ject and predicate are determined, the firſt by expreſſion or implication,
the ſecond by the nature of the copula (fee page 57 ofthis work). And
the only quantities confidered are all and some ; the latter meaning any-
thing that not none may mean, ſome, it may be all but not known to be
all, perhaps not more than one. The matter contained in A ſuggeſted
itſelf to me in the fummer of 1846, and was forwarded to Cambridge
with the rest of the memoir on the 4th of October.
I will now introduce Sir W. Hamilton's defcription of the various
kinds of quantity (II p. 31 , 32).
Your " Statement" is chiefly plausible from a wretched confufion
' ofdistinct things. This confufion, with which you delude yourſelf,
' and many of your readers, is of two independent ſchemes of logical
' quantification ; the one, aſſerting an increase in the expreſſly quantified
' terms, the other, a minuter diviſion of theforms of quantification itself.
• To diſintricate this entanglement, we have fimply to confider, in their
• contrafts, the three following ſchemes of quantification :-
• The firſt ſcheme is that which logically confines all expreſſed
' quantity to the Subject, prefuming the Predicate to be taken-in ne-
'gative propoſitions, always determinately in its greatest and least ex-
•tenfion (univerſally and fingularly), in affirmative propoſitions, always
' indeterminately in some part of its extenfion (particularly).
• Thesecond ſcheme is that which logically-extends the expreffion
' ofquantity to both the propofitional terms, and allows the Predicate to
• be of any quantity, in propoſitions of either quality. This not only
•ſupplies a capital defect, but affords a principle on which Logic ob-
'tains a new and general development.
• The third ſcheme is that which logically admits more expreſſed
' quantities than a determinately leaſt or greatest extenſion (quantity fin-
'gular and univerſal), and an indeterminately partial extenſion (quantity
،
particular.) This, though it corrects, perhaps, an omiffion, yields no
principle for a general logical development.
• The first doctrine is the common or Ariftotelic; the ſecond is mine ;
' and in the third-in ſo far as you have gone, and apart from the con-
• ſideration of right or wrong-I do not queſtion your originality.
Now, the ſecond and third ſchemes are both opposed to the firſt,
⚫ but in different reſpects ; conſequently the ſecond and third may, each
 Appendix . 301
• of them, combine with itſelf, either the whole other, or that part of
• the firſt to which it is not itſelf oppoſed. More is impoffible.
• Let the following be noted : *-Your old view (that in the body of
• the Cambridge Memoir) is a combination ofthe ThirdScheme ofquan-
* tification with the FIRST ; your new view (that in its Addition) is a
• combination ofthe THIRDScheme of quantification with the SECOND : and
• the confusion, ofwhich you are now guilty, is the recent and uniform,
' and perverse identification, in your PRESENT " Statement," of the SECOND
• Scheme with the third.
• Before, however, proceeding to comment on your confufion of the
• ſecond and third ſchemes, I may alſo relieve a confufion in the term
6

definite and its reverſe, indefinite, as applied to logical quantification.

In the first, common, or Ariftotelic meaning, definite, or more pre-
• ciſely predefinite ( διοριστὸς, προσδιοριστός,) is equivalent to expreſſed,
' overt, or, more proximately, to defignate and pre-designate; in this
' ſenſe, definite quantity denotes expreſſed, in oppoſition to merely under-
'stood, quantity.
• In thesecond meaning, that which I have always uſed, (and certain
ancients, I find, were before me,) definite is equivalent to determinately
• marked out ; a ſenſe in which definite quantity is extension undivided
' or indiviſible, univerfal or fingular (this including any collected plu-
• rality of individuals) as opposed to particular quantity.
• In the third meaning, which you have ufurped, definite is equivalent
' to numericallyspecified; and in this ſenſe, a definite is an arithmetically
' articulate quantity, as oppoſed to one arithmetically inarticulate.-
• This your meaning of the word I did not, before the appearance of
،

your " Statement," apprehend ; for of course I prefumed you to uſe it

• in its firſt or common meaning, from which you never hint that you
• confcioufly intend to deviate.'
Three ſchemes ofquantity are here mentioned.
Firſt, the ordinary one.
Secondly, that in which the ordinary quantities, all andsome, are ap-
plied in every way to both fubject and predicate.
Thirdly, that in which numerically definite quantity is applied to
ſubject or predicate or both the eſſential diſtinction of this case is nume-
rical definiteneſs : it really contains the ſecond ſyſtem, when numerical
quantity is algebraically expreſſed. Of theſe, it appears, Sir W. Ha-
milton claims the ſecond, or rather, the application of ſuch a ſcheme to
the fyllogifm. What then is it ? I ſuppoſe it to be the following. My
order of reference is XY.

* Let the following alſo be noted :-My old view (that in the body of the Cam-
bridge paper) is entirely on thefirſt ſcheme, except in one digreſſive ſection and one
ſubſequent paragraph (from both ofwhich A is quoted) in which thesecond and third
are combined : my next view (that in the addition) is alſo a combination of thefecond
and third ſchemes : and my " Statement " contained alſo a uniform, but not recent,
identification of the ſame ſecond and third ſchemes, which I never ſeparated in thought
until I saw this paragraph. Any one who can form an opinion of the way in which
the ſubject would preſent itſelfto the mind of a mathematician, will fee that the ſecond
ſcheme would preſent itſelfconcomitantly with, and as an effential part of, the alge-
braical form of the third. A. De M.
302 Appendix .
All X is all Y means that X and Y are identical : it is my D. All
X isfome Y is A. Some X is all Y is A'. Some X is some Y is I.
As to negative propoſitions, All X is not all Yis E. Some X is not
all Y is O. All X is notsomeris O'. Some X is notfome Y is true
ofall pairs of terms one ofwhich is plural. In its indefinite form, it is
what I have in Chapter VIII. calledspurious.
The propofitions of this ſyſtem are then the complex D, or A +A',
the fix Ariftotelian forms A₁, A', E , O , O', I , and the ſpurious form,
which may be called U. In looking over (Sept. 5) Sir W. Hamilton's
pamphlet, I happened to light on the affertion (incidentally made) that
his ſyſtem gives thirty-fix valid moods in each figure. On examining
the preceding ſyſtem, I find this to be the cafe. I should not have pub-
liſhed the refults, had not Sir W. Hamilton made it neceſſary for me
to comment on them. I ſhall denote the propoſition U, or ' Some Xs
are not ſome Ys ' by X :: Y ; and I ſhall, ſuppoſing each cafe to be formed
in the firſt figure, then tranſpoſe it into my own notation.
1. There are fifteen forms in which D enters. Whenever D is either
ofthe premiſes, the other premiſe and conclufion agree. Thus we have
ADA , DUU, &c. &c.
2. Fifteen Ariftotelian forms AAA, A'A'A' ; AEE , EA'E ;
AO'O', 'O ; A'OO , O'AO' ; A'II , IAI ; EIO , IEO ;
Α'ΑΙ ; Α'ΕΟ , EAO'.
3. Six more U fyllogifms A'O'U, OAU ; A'UU, UAĻU ; IO'U,
OIU.
The two things to be conſidered are ;-the introduction of the iden-
tical propofition ; and that ofthe ſpurious one, as I call it.
It is, I ſuppoſe, a fundamental rule of all formal logic, that every pro-
poſition must have its denial, its contradiction. Now D has no fimple
contradiction in this ſyſtem : that O' and O, both contradict it (and alſo
E) is true : but the mere contradiction is the disjunction ' O' or O ' .
A perſon who can ſhow that one or the other oftheſe is true, has de-
monſtratively contradicted D, even though it could be proved impoffible
to determine which of the two it is.
The propofition U is usually ſpurious. But if we introduce it, we
muſt introduce its contradictory alſo. Now if either X or Y be plural
names, it must be true : conſequently, the contradiction of U is ' X and
Y are fingular names, and X is Y.' When a ſyllogiſm having the pre-
miſe U is introduced, either that premiſe maybe contradicted, or it may
not. If it may, there is no form todo it in: ifit may not, then it is a
ſpurious propoſition, and cannot, by combination with others, prove
anything but a like ſpurious conclufion.
Let X :: Ydenote ' Some Xs are not ſome Ys,' andX,Y, denote ' there
is but one X and one Y, and X is Y.' Then either X:: Y or X,Y, muſt
be true, and one only. A logical ſyſtem which admits one and not the
other, which contains an affertion incapable of contradiction without
going out of the ſyſtem, can hardly be ſaid to be " adequately teſted and
matured," and is not ſelf-complete. The propoſition X,Y, includes in
itſelfthe conditions ofD, and is a kind offingularform of D.
 Appendix. 303
Iprefume, from the number of Sir W. Hamilton's moods, thirty-fix,
as above obtained, that the contradiction neither of D nor of U finds a
place. Admit them, and the contradiction ofU alone (call it V) de-
mands fixteen new moods in each figure. I will now proceed.
In my publication, ſpeaking now of(A) what was fent to Cambridge
before I communicated with Sir W. Hamilton, I had no quantification
intermediate between the ordinary one, and the numerical one applied
to either ſubject or predicate, as wanted in the canon ofthe middle term
there given. Look at the last ofthe ſeven ſyllogiſms in thesecond extract,
where both the predicates, being of the middle term are quantified, and the
condition ofvalidity is quantitatively ſtated. But for Y₁+Y, less than
I ' ſhould be read ' y₁+y, greater than 1.' The equivalence of this
to ' Y+Y, less than I ' is a mistake. In thefirst extract, the general
2

canon is given which is afterwards uſed in C.

Up to the time when Sir W. Hamilton publiſhed his letter in reply
to my ſtatement, (II), I never had ſeparated the idea ofhis ſecond ſcheme
ofquantification from that of the third.
Thus then we stood on October 3, when I fent my paper to Cam-
bridge. Sir W. Hamilton had been teaching the application of the
ordinary quantities to both ſubject and predicate : I had arrived at the
algebraical repreſentation ofthe numerical quantification of terms, whe-
ther fubject or predicate matters not, as long as they were middle terms.
1846, October 6. My communication (containing A) was in the
hands of Dr. Whewell (as he informs me) for tranſmiſſion to the Cam-
bridge Society : I never saw it again till the next February. October
7, Sir W. Hamilton wrote to me, in anſwer to an application of mine
on the history of the fyllogifm, further informing me that he taught an
extenfion and fimplification of its theory, which he offered to commu-
nicate. November 2, (the offer having been accepted) Sir W. Ha-
milton forwarded the communication B, which I give entire ; conſiſting
of a letter, and the Requisites which he had furnished to his ſtudents,
for a prize Eſſay. December 28, he wrote again, forwarding a printed
Prospectus of his intended work on logic. This is not material ; for,
on receiving it, I thought certain, what from the previous communica-
tion I had thought poſſible, that Sir W. Hamilton was in poſſeſſion of
the theory of numerically definite ſyllogiſms (but this was a mistake of
mine, as will preſently appear). I accordingly, to preſerve my own
rights, immediately forwarded (as will preſently be ſtated more in de-
tail) an identifying deſcription ofthe ſheets of paper on which my nu-
merical theory was written, and an account of both my ſyſtems (in
letters dated December 31 , 1846, and January 1, 1847). Of this, Sir
W. Hamilton (who has publiſhed both letters) is my witneſs. 1847,
February 27, I dated the addition to the proof ſheet of my Cambridge
paper, which was deſpatched to Cambridge the next day. This addi-
tion contains C, which itſelf contains (in ſubſtance) all that part of my
letter of January I which refers to the diſputed point. March 13 , Sir
W. Hamilton wrote the letter containing the charge ofplagiarism ; hav-
ing been for two months prevented by illneſs from reſuming the ſubject.
304 Appendix.
All ſubſequent correſpondence referred to proceedings, and not to the
ſubject matter of the charge.
Many days before the middle of October, I had applied the ſyſtem of
quantification in the manner ſhewn in C. Sir W. Hamilton thinks my
memory has failed here : I know better. My memory does not depend
upon a date, but upon the opening of the Univerſity College Seffion,
which takes place in the middle of October. But it matters nothing,
for the notion of the complete quantification of a predicate, when wanted
because it is the middle term, will prove the poſſeſſion ofthat proceſs as
well as quantification in all cafes whether wanted or not. On receiving
B, I looked with curiosity at 2º, on which, in fact, Sir W. Hamilton
grounds his declaration ofhaving made a communication. He demands
ofhis pupils,
• The reasons why common language makes an ellipsis ofthe expreſſed
quantity, frequently of the subject, and more frequently of the predicate,
though both have always their quantities in thought.'
On looking at this, and feeing mention of the quantities which the
terms have in thought, in common language, I took it for granted that
the common quantities were ſpoken of: namely, that of the fubject from
the tenor of the propoſition, that ofthe predicate from the nature of the
copula. I never ſhould have imagined that in the common language of
common people, there were any other quantities, even if, intheir minds,
the predicate have theſe. Had this been all, I ſhould have paſſed it over,
as referring to common quantities, and making common people a little
more of logicians, as to the predicate, than I have found them to be.
That this common language meant the language of any ſcientific ſyſtem, I
had not the least idea : ſtill leſs that it referred to the language of the
writer's own unprinted ſyſtem, current only between himself and his
hearers. And, though I gained a fufpicion that Sir W. Hamilton might
have (which he had not) adopted numerical quantification, it was not
from this paſſage, which by itselfwas nothing, but from what is now
coming, which made this paſſage ambiguous.
On looking further into B, (which fee) I found that Sir William's
ſyſtem, whatever it might be, noted defects in the conversion ofpropofi-
tions, and a general canon offyllogifm. Now I had two ſyſtems, each
of which had its own way of adding to the converfions, and each its
own canon of fyllogifm. In my firſt ſyſtem (which has now grown into
Chapter V) the permanent introduction ofthe contranominals is a com-
pletion of converſion : and the reduction, by the remarks in pages 96,
&c. ofall fyllogifms to univerſal affirmative premiſes, was the canon of
fyllogifm. In the ſecond, ſeen in A and C, which has grown into
Chapter VIII, there is the univerſality of ſimple converfion, and the
canon of the middle term . Sir W. Hamilton may deny (I believe he
does) that theſe are canons : let it be ſo; but I took them for canons,
and thought ofthem when I saw the word canon in his ſummary. And
then the queſtion was, had Sir W. Hamilton one of theſe ſyſtems, or a
third one ? I had been throughout our correfpondence well pleaſed with
the idea that I had hit upon ſomething in common with Sir W. Ha-
milton ; and in my anſwer to communication B I faid,
 Appendix. 305
' I am not at all clear that I shall not have to claim only ſecondary
'originality on ſeveral points. When I see " defects of the common
doctrine of converfion" and a " fupreme canon" of categorical fyllo-
'gifm, I must wait for further information. I think I may yet be
....

'able to flatter myſelf that I have followed you in ſome points unknow-
' ingly.'
The reader will obſerve that this inſtructive communication is fup-
poſed to tell me, that in my thoughts the predicate has all kinds of
quantity : though in truth both have their quantities is not Engliſh for
either may have any one of two species of quantity. Sir W. Hamilton
has expreſſed (perhaps) the dictum which is to have taught me new
quantification, in terms ofthat new quantification unknown. By both
have quantities he ſeems to aſſert that he meant both have all quantities.
That both have their quantities, is true in the common ſyſtem : theſe
words, which expreſs a truth of the common ſyſtem, Sir W. Hamilton
declares to be a fure mode of communicating the difference between his
ſyſtem and the common one. This may do in his own lecture room,
in which he has the arbitrium et jus et norma loquendi in his own
hands. A diſtinctively unmeaning phrafe may, in virtue of his expla-
nations, paſs current between him and his pupils : and a private bank,
of courſe, muſt receive its own notes. But they are not lawful tender
anywhere : nor good tender out of the neighbourhood.
I ſhall now proceed to the letters in the Athenaum (III and IV).
Theſe contain the iſſues raiſed by the pamphlets : my ſhort letter con-
tains the ſtrength of my cafe : I am to preſume that my opponent's
letter contains the ſtrength of his anſwer, and I think it does ſo. At
leaſt I can fee nothing ſtronger in his pamphlet.
MR. DE MORGAN. SIR W. HAMILTON.
I take this mode ofacknowledg- In reply to your letter in the laſt
ing the receipt of your printed number of the Atheneum :-you
letters to me. I promiſed you an were not wrong to abandon your
anſwer, if you would bring for- promise " oftrying the strength of
ward the grounds ofyour affertion my poſition;" for never was there
that I had acted with breach of a weaker pretenfion than that, by
confidence and falſe dealing. But you, ſo ſuicidally maintained. You
you now admit that your grounds would, indeed, have been quite
are no grounds ; you declare your right had you never hazarded a
conviction that (though chargeable ſecond word ; for every additional
with confufion, want of memory, ſentence you have written is ano-
&c. &c.) I have acted with good ther mif-ſtatement, calling, ſome-
faith ; and you offer a proper re- times, for another correction .
traction and apology. You ſtate in
various places and manners, that
though you are fatisfied of my in-
tegrity, all may not be ſo ; and,
thereupon, you call for an anſwer.
But I think that others will be
X
306 Appendix .
quite fatisfied with your own an-
fwer to your own charge.
There is nothing left which I
care to diſcuſs with you. Our
views of logic, their coincidences,
their differences, their firſt dates,
my memory, &c. I am content to
leave to those who will read my
ſtatement and your letters, with
two remarks.

There is no ſtrength in an abandoned poſition. My pamphlet was

publiſhed in defence of my own character : when Sir Wm. Hamilton
retracted his charge of breach ofconfidence and falfe dealing, there was
nothing to which I stood engaged, nothing I cared to write ſeparate
pamphlets on, eſpecially when the approach of this preſent publication
was confidered. Any one who reads page 9 of my pamphlet, in which
the promiſe was made, will fee that it has reference to what I there call
"the infamy which would attach to any one who had deſerved the
terms he used for the conduct he described." I certainly forgot to fay
"unless you retract : " but as he had already refused to retract (though
he had propoſed tofufpend the charge, provided I would then undergo
an examination) it did not enter into my head to provide for ſuch a con-
tingency. The aſſertions about weakness, misſtatement, &c. are for the
reader's judgment. I did not, in this letter, allude expreffly to Sir W.
Hamilton's various infinuations that the old charge might be true : both
becauſe, at the firſt hurried reading, I did not become aware of their
extent ; and alſo, becauſe I wiſhed to take time before I made up my
mind as to the way oftreating what I faw ofthem.
MR . DE MORGAN. SIR W. HAMILTON.
1. As foon as the queſtion of You do not deny, that your cor-
character was diſpoſed of, it was refpondence afferts a claim to the
your buſineſs to ſhow that my Ad- principle communicated to you by
dition,* written after I communi- me ; but you complain that I have
cated with you, contained ſome not ſhown that your Addition in-
principle not contained in my Me- volves a new doctrine, uncontained
moir,† written before I communi- in that part ! [from the overt con-
cated with you. This you do tradictions of its other parts I had]
not do. You affert, and you de- of your Memoir which you de-
ſcribe, andyou ſum up ; butyou do clared to contain the principles
not quote, except a few words, uſed in your Addition . And this
which are not in that part of my you can ſay, when I explicitly
Memoir which I declared to con- ſtated that " throughout the whole
tain the principles uſed in my paper (the Memoir) not only is
Addition . there much in contradiction-there

* Here given in C. † Here given in A, ſo far as relevant.

‡ Sir W. Hamilton's part of this is B.
 Appendix. 307
is abſolutely nothing in (more then
fortuitous) conformity with the
theory of a quantified predicate"
(L. p. 34). This, too, you can
fay whilft before your eyes, unan-
fwered, there was lying " myfor-
mal request, that you would point
out any paffage of your previous
writings in which this doctrine
(that afferted in your ' Statement,
of a quantification of the middle
term, be it fubject or predicate) is
contained" (Ibid)-for I couldfind
none ; and none has by you been
indicated.

I do deny, in one ſenſe, that my correſpondence afferts a claim to the

principle communicated by Sir W. Hamilton : for I deny that he com-
municated any principle. I prefume of courſe that the Profpectus and
letter fent on the 28th of December are out of the queſtion: fince I
gave the ſyſtem on which the charge was made by return ofpoft. Sir
W. Hamilton has very properly confined himself, in his pamphlet, to
his communication (B) of November 2, as containing the communica-
tion which he afferts me to have used. Let the reader look through it
and ask himself what new principle is communicated, and where.
Sir W. Hamilton afferts that he has shown my Addition to contain
a new doctrine, not contained in one definite part of my memoir, by the
contradictions of its other parts. Let P, Q , R, be parts of a memoir ;
and S an addition. By showing that P and Qcontradict one another,
Sir W. Hamilton thinks he ſhows that S contains a doctrine not in-
volved in R. The fact is, that all my memoir except ' Section iii. On
the quantity ofpropofitions' and one other paragraph (from both which
Ais taken) belongs to the ſyſtem of Chapter V. in this work : while
Section iii., the other paragraph, and the addition, belong to Chapter
VIII. Let the reader take notice that Sir W. Hamilton (who, by the
way, feems to confider I explicitly ſtated ' as a ſufficient anſwer to
'you have not ſhown ') does findsomething in my memoir in conformity
with the theory of a quantified predicate. He says it is fortuitous :
but it did not feem to him requiſite to bring it forward, and point
"Howout
itsfortuitousness. This point is for the reader to judge of.
dare you," he ſays, " rob me of my quantified predicate." " Good Sir,"
I anſwer, " I had it before I knew you." " What if you had," he
replies,
Sir W." itHamilton
is enoughcannot
if I inform you that
find either in it wasmemoir
the only by
or accident."
the addition
(he ſays here only in the previous writings, but in his pamphlet (p. 34)
he ſtates it of both memoir and addition), any thing about the doctrine
ofquantification of the middle term, whether it be ſubject or predicate,
which doctrine he ſays is repugnant to all that is there taught. It is
308 Appendix .
true thatin the next fentence he refers to previous writings, as cited. I
will therefore conclude that Sir William included the addition by mif-
take, and meant the memoir only. Whether my Section iii. (A) is or
is not full of quantification of the middle term, without reference to
whether that middle term be fubject or predicate, I am quite content to
leave to the reader. Sir W. Hamilton ſays he cannot find it. This I
believe, and wonder at : but it does not follow that it is not there. Let
the reader look .
Again, when Sir W. Hamilton afferted that C contains ſomething
which I got from him, and which is therefore not in A, I repeat that
he ought to have pointed out what it is. His aſſertion that he cannot
find it in A neither proves that it is not in A, nor that it is in C.
This is the pinch which obliged him to write forty-four pages ofac-
cufation in anſwer to fixteen of defence : and this is the point on which
the question will finally turn. I am tedioufly often obliged to bring
the whole matter to its ABC ; but what else can I do with an oppo-
nent who writes an ignoratio elenchi of forty-four pages long.
Sir W. Hamilton is not good at finding. Immediately after what
he has quoted from himself as above, comes the following paſſage ;-
' In regard to your third affertion, that " perfectly definite quantifica-
' tion destroys the necessity of distinguishing subject and predicate ;" this
' is altogether a mistake. It is not " definite quantification," (in what-
' ever ſenſe the word definite be employed), but the quantification ofboth
' the terms which " destroys the neceſſity ofdiftinguiſhing ſubject and
'predicate ;" and this by ſhowing, that propofitions are merely equations,
'and enabling us to convert them all-fimply.'
Inow quote from myſelf. Of the two fentences now coming, Sir
W. Hamilton quotes the firſt, omits the second, which ſhows that my
phrase perfectly definite ' means definite in both terms, and then makes
the preceding remark.
In fact, perfectly definite quantification deſtroys the neceffity ofdif-
'tinguiſhing ſubject and predicate. To say that ſome 20 Xs out of50,
' are all to be found among 70 Ys, or that 20 out of 50 Xs are 20 out of
'70 Ys, is preciſely the ſame thing as ſaying that 20 out of 70 Ys are
'20 out of 50 Χs.'
In a writer of whom diſhonest intention might be concluded, we
ſhould know how to explain the omiffion of the ſecond ſentence. But
there is no diſhonesty in Sir W. Hamilton : the omiſſion must be
referred to the fame diſpoſition which prevents him from ſeeing quan-
tification of the middle term in A. What I take that diſpoſition to be,
matters nothing to my reader. Perhaps this ſentence alone will enable
ſome to detect that I had not any idea ofthe ſecond ſyſtem ofquantifi-
tion independently of the third.
MR. DE Morgan . SIR W. HAMILTON.
2. All the alleged inconfiften- You ſay, that my expoſure of
cies which you find in my letters, your inconfiftencies is unavailing,
&c. will not help you till you have except " I ſhow that my commu
 Appendix . 309
done this : and even then, you nication was intelligible." You
will have to ſhow that your com- forget that it is for you to explain
munication was intelligible. how, having "Subscribed to," as
Inglancing over my letters and having " rightly understood," twen-
the maſs of notes which you have ty-two fentences of my profpectus
written on them, I ſee that I have (L. pp. 19, 16), you could ſubſe-
feveral times uſed inaccurate lan- quently declare that communica-
guage, as people do in hurried let- tion to be unintelligible !! (L. p.
ters. Still more often you have 59). I have now no doubt, how-
misunderstood me. If my occa- ever, that you then " ſubſcribed
fional inaccuracy and your ocса- to" more fentences than, by you,
fional miſunderſtanding ſhould be were " rightly understood. " In-
held to furniſh ſome excuſe for you deed,hadyou only betimes avowed
whenyou precipitately charged me that all you had " fubfcribed to, as
withdiſhonourable conduct, I ſhall rightly understood," was to you
be better pleaſed than not. really unintelligible, and that the
repetition of my doctrine was in
your mouth mere empty found,
two pamphlets might have easily
been ſpared.

First, the profpectus is not the " communication." The communica-

tion is that of November 2 (B). Let the reader look at it, and fee
whether it be intelligible communication of new principle.
In my pamphlet I have ſeveral times ſpoken of the communication,
though there were two. This was natural enough, inaſmuch as there
was one communication (that ofNov. 2), on which the charge was made
against which that pamphlet was a defence. Sir Wm. Hamilton has
never ventured to maintain that I derived anything from the communi-
cation ofDec. 28, containing the profpectus, to which I replied on the
evening I received it, as preſently mentioned. But he makes, in various
places of which the above is one, a mixture of the two communica-
tions.
Secondly, I have looked carefully at pages 19 and 16 of Sir Wm.
Hamilton's letter, and at all the rest ofour correfpondence, without find-
ing that I have ever admitted that I ſubſcribed to any part ofthe prof-
pectus as by me " rightly understood." Page 59 is no doubt a miſprint
for 39. I have neither found, nor have I the flightest remembrance of,
any ſubſcription of mine to any thing Sir Wm. Hamilton ever wrote as
" rightly understood."
Irepeat the account given in my pamphlet ofthe manner in which I
ſubſcribed to this profpectus ;-
The next communication is dated Dec. 28 , and conſiſted of 1. A
• letter. 2. A printed profpectus of Sir William Hamilton's intended
'work on logic. Nothing turns on this, for the fimple reason that my
'anſwer contained the moſt expreſs and formal proofthat, come by it
' how I might, I was then in the moſt complete written poſſeſſion of all
• I have ſince publiſhed.... The profpectus which accompanied this letter
310 Appendix.
' is very full on the results which Sir William Hamilton can produce
• from his principles ; but gives nothing, I think, certainly nothing intel-
' ligible to me, on thoſe principles themſelves.
As foon as I saw theſe reſults, I inſtantly saw that many of them
' agreed with my own. I had then no doubt that we poſſeſſed ſomething
' in common ; and I said ſo very distinctly in my reply. As the reader
' will preſently fee, this firſt impreſſion has not been confirmed. Feeling
' it now time to ſecure whatever of independent diſcovery might belong
' to me, I anſwered Sir William Hamilton in two letters, dated Decem-
' ber 31 and January 1. In theſe letters-
' I. I returned the printed profpectus with the reſults underlined
• which my ſyſtem would produce.
' 2. I ſtated that I had a ſyſtem written on certain ſheets of paper,
' which I deſcribed as to number, fize, &c., adding the head words of
' each page. I felt inclined to get the fignature of ſome good witneſs put
'upon theſe papers ; but at the ſame time I felt reluctant that Sir Wil-
6
liam Hamilton ſhould fee, if it ever became neceſſary to produce theſe
papers, that I had been taking precautions against him. I therefore de-
' termined to make himself my witneſs .
6

3. I ſtated distinctly the firſt principles of both my ſyſtems, and the

•fyllogiftic formulæ to which they lead.'
Thirdly, I ſubſtantiate the above, ſo far as the ſubſcription is con-
cerned, by quoting two paſſages from Sir W. Hamilton's publication
of my letter of December 31 .
' I received your obliging communication this morning and am now
• fully fatisfied that I have, in one of my views of fyllogifm, arrived at
'your views in ſubſtance, or ſomething ſo like them, that I could fub-
' ſcribe in my own ſenſe to a great part of your paper ...... This
' chapter [meaning the one on the ſheets of paper above referred to]
' I might expreſs in your words wherever they are underlined in the
6

'profpectus which I return, hoping you will send another.'

Where are thoſe words " rightly understood" which Sir W. Ha-
milton attributes to me three times in one paragraph ?
He must have been quoting from memory. Seeing his results, I
found they were alſo my refults; so I told him that I could " ſubſcribe"
(and I cannot find I have uſed this word more than once, and it is in
page 19 referred to by Sir William) " in my own ſenſe to a great part
of" his " paper." If words can ſpeak meaning, I here tell him that I
ſubſcribe in my own ſenſe, leaving it to the future to ſhow whether I
ſubſcribe in his, that is, whether I understand him rightly.
[I was reading this for the preſs, when I found out the words which,
applied in one ſenſe hypothetically to one of his refults, Sir W. Hamilton
has transferred in a different ſenſe to all. One of his refults, ſpeaking
of the moods, is the eſtabliſhment of Their numerical equality under
all the figures,' the Italics being his. I could not make out the Engliſh
of this. The others I understood in the grammatical ſenſe. For ex-
ample, The abrogation of the ſpecial laws of fyllogifm' is intelligible :
I did not know whether my ſenſe of theſe words, that is to ſay, my
 Appendix. 311
abrogation of thoſe laws, was the fame as Sir W. Hamilton's ; ſtill that
hedid abrogate certain laws was clear. But numerical equality of moods
I could only understand as referring to the numerical quantities which
I fuppofed (the reader will remember that I fent back the profpectus by
the next poft, and had little time to look at it) Sir W. Hamilton's ſyſtem
to contain. It means, I find, that there are the fame number of moods
in all figures : but to attribute numerical equality to different things is a
mode of faying that there is the same number of them in different fets to
which I was unaccustomed. Having however, as I thought, divined
what the Engliſh of this might mean, I underlined it, adding (as Sir W.
Hamilton ſtates in one of the foot-notes, which I never remarked till
now) theſe words, " If I underſtand this rightly I may underline it I
think ." I meant, " If I can make out the words." This understand
rightly, Sir W. Hamilton actually takes from this ſentence, joins it to
my " ſubſcription" mentioned in another document, and repreſents me
as declaring that I have “fubfcribed to as rightly understood" twenty-two
Sentences, &c., and himself as quoting from one paſſage.]
But, had I betimes avowed my non-underſtanding, two pamphlets
might have been ſpared. Where are we now ? I did avow my not un-
derſtanding the firſt communication, and my fſubſcribing to the ſecond
in my ownsense. To which Sir W. Hamilton ſubſequently anſwered
to the effect that I spoke falſe, that I did underſtand the first, for that
I had fent him, in letters written immediately after the ſecond was
received, his " fundamental doctrine " and " many of its most important
confequences." What have I been contending for all along, except
that the doctrine of hisfirst communication was to me mere empty found,
and that all I was able to produce when I received thesecond, was my
own ? But Sir. W. Hamilton actually gives me a right to ſay, with
reference to the second, the more developed and more intelligible com-
munication, that I did not underſtand it, inſiſts upon my ſaying it, and
reproaches me for not ſaying it. Well then, to uſe a Scottish phrafe,
the leſs I lie when I say I did not understand the first, which is the
point at iſſue. So that, as to the matter of our controversy, Sir W. Ham-
ilton admits that there was (fortuitous he calls it) entrance of the theory
of the quantified predicate in my writings prior to his communications ;
and as to the conduct of it, he admits that I did not understand his
communication ; and in the face of fact, reproaches me with maintain-
ing that I did till after the pamphlets were written : when it was of the
effence of my ſtatement, firſt, that I did not underſtand, ſecondly that
neither I nor any one else could have understood, ſave only the pupils
to whom the requifites were addreſſed.
MR. DE MORGAN. SIR W. HAMILTON .
Your copious and ſlaſhing cri- I diſregard your mifreprefenta-
ticiſms on my intellect (by which tion that " I avenge myſelf for the
you avenge yourself for the retrac- retraction of my afperfion on your
tion ofyour afperfion on my integ- integrity by my copious and flash-
rity), I will profit by ſo far as I ing criticisms on your intellect."
312 Appendix .
diſcover them to be true: the reſt When your (excuſable) irritation
ſhall amuse me;-and the whole has ſubſided, you will fee that I
will be good for the printer. Take could only ſecure you from a ver-
one retort from me on the ſame dict ofplagiarism by bringing you
terms . You have much ſkill in in as fuffering under an illufion.
forming new words ; and, as is What, however, is all in all ;-my
fair, you put your own image and criticisms will not, I think, be
ſuperſcription on your own coin- found untrue.
age. I think you have got into If guilty of leſe majesty by re-
the habit of aſſuming the ſame ference to the Queen's Engliſh,
authority over that already exiſting have I not my accuſer as abettor ?
portion of our language which is For you not only paſſed my min-
commonly faid to belong to the tages (quantify and quantification)
Queen-and that you need an in- as current coin ; but, in borrow-
terpreter. If I can arrive at your ing, actually " thanked me for the
meaning by the time I write the words" (L. p. 22). However,
preface to my work on logic, I myverbal innovations are, at least,
will ſtate your claim, accompanied not elementary blunders. I do
by your own words ; ifnot, I can not, for example, confound a term
ſtill ſtate your own words. Till with a propofition, the middle with
then, I have nothing more to ſay. the conclufion of a ſyllogiſm.
Sir W. Hamilton unconsciously adapts his language to a very true
ſuppoſition, namely, that he has, in his pamphlet, made himself the jury
in this cafe. He is unfortunate about the mintage. I ſay to him ' You
make new words well, but I am afraid you alter the old ones.' To
which he replies ' Why, you thanked me for my new words. So I did,
and ſo I do again: but what has that to do with the lefe majesty part
ofmy infinuation.
Sir W. Hamilton ſays that I have ſomewhere (where he does not
ſay) uſed term for propofition, middle for conclufion, collectively for dif-
tributively. This may be ; fuch flips of the pen are common enough.
He fets them down as blunders of ignorance. I am not afraid the
reader will follow him. He ought to have faid where they occur, that
is, when he firſt mentioned them, in his pamphlet. Till I put theſe
letters together, I was fatisfied, on Sir Wm. Hamilton's ſtatement, that
I had done all theſe enormities : but now, after the case of " rightly
understood" which I have just had to diſcuſs, I do not feel ſo well fatif-
fied.

SIR W. HAMILTON,

Finally, I beg leave to remind you.--There is now evidence in your

poffeffion that for seven years, at least, the doctrine of a quantified pre-
dicate has been puclickly taught by me ; whilft, on your part, there is
a counter afſfertion or innuendo, which, as you cannot prove, it concerns
your character formally to annul.
‫د‬ I never denied that Sir W. Hamilton had taught a doctrine of the
quantified predicate. By the time I wrote my pamphlet, I was pretty
 Appendix. 313
fure that it was not the fame as mine. Sir W. Hamilton's anſwer
confirmed me in this, as appears in page 300.
I now come to mention a part of the diſcuſſion which I ſhould per-
haps have omitted, if I had not pledged myself in my pamphlet to give
an account of a certain offer which I there made to Sir William Ham-
ilton, in the event of that offer not being accepted. It is a curious
inſtance of that diſpoſition to hold a correſpondent or an opponent
capable of folving enigmas, and bound to do it, which appears in his
prefuming that (fee B, paragraph 2°) an obfcure reference to what is
done in common language would enable me to gueſs at the uncommon
language of his ſyſtem and his lectures. I infert it, alſo, as a ſpecimen
of the various mifunderſtandings and misapprehenfions which Sir W.
Hamilton imputes to me, referring to a matter which readers will ſepa-
rately comprehend. Had I ſpace or inclination to deal with them all,
I believe I could ferve them all in the fame way.
Ολ. 7, 1846, I learnt from Sir Wm. Hamilton that his doctrine
had obtained confiderable publicity through the notes and eſſays of his
ſtudents. In my reply, referring to this ſyſtem, and to his offer of
communicating it, I asked if he had a pupil whom he could truſt with
the communication ; the answer was B, preſently given. But, Dec. 28, in
fending the profpectus, Sir W. Hamilton informed me that, before
forwarding it (the firſt communication in which that he had other than
Ariftotelian quantification was intelligibly announced) he had waited for
a reply from Mr. That gentleman ' continues Sir W.
Hamilton, in words fome of which I place in Italics ' was a pupil of
' mine fix years ago, and obtained one of the highest honours of the claſs ;
' he was therefore fully competent to afford you information, which I
• begged him to do, in regard to my logical doctrines as they were taught
' ſo far back. I knew him to be a graduate of your College, and he tells
' me that he was for three years a pupil of your own. If you are ſtill
'intereſted in the matter, you can therefore obtain from him as an
'acquaintance, what information you wish, more agreeably than from a
' ſtranger. When he attended me, besides the twofold wholes in which
•the fyllogifm proceeds, the quantification of the predicate, and the effect
'of that on the doctrine of converfion, on the doctrine of fyllogiftic
' moods, on the ſpecial ſyllogiſtic rules, &c., were topics diſcuſſed, and
' partly given out for exercises. They were, infact, then mere common-
'place.
Jan. 13, 1847, Mr. called on me at Univerſity College, after
an evening lecture of mine, put his notes into my hands, and has fince
ſtated (in which I have no doubt he is correct, though I do not remem-
ber it) that he informed me he was doubtful whether they contained
exactly what I wanted, and that he would gladly furnish any additional
information. Now I conceived, as I thought it was intended by Sir
W. Hamilton I should do, that the notes of one of the beſt ſtudents,
even if not exactly what I wanted, were fure to contain something of
the mere commonplace (by which I took to be meant the ordinary matter
of the lectures) which was diſcuſſed, and given out as exercises to thoſe
314 Appendix.
who attended. But in theſe notes I found nothing on quantification (I
had now this key word, which did not appear in the main communica-
tion B) differing from what is usual; and after expreffing this in my
pamphlet, I proceeded as follows :
'But if there really be anything in which Sir William Hamilton has
' preceded me, I shall be, of all men except himself, moſt intereſted in
'his having his full rights. And I make him this offer, and will take his
'acceptance of it as reparation in full for his fufpicions and affertions.
With the conſent of the gentleman to whom theſe notes belong, which
' I am fure will not be refused to our joint application, I will forward to
' him a copy of their table of contents, having more than a hundred and
'fifty headings. From theſe Sir William Hamilton ſhall ſelect thoſe
' which are, in his opinion, fure to contain proof ofhis priority on any
'point which I have inveſtigated. Of theſe I will have copies made and
'ſent to him : and will print in the work on Logic which I am preparing
'(and in fome one part of it) the parts which he ſhall ſelect as fit to
'prove (or to ſhow that he could prove, let him call it as he likes) his
' cafe, or the germs of his cafe (as he pleaſes, again). Provided always,
'that the matter ſhall not run beyond fome eight or a dozen octavo pages
'of ſmall print. And I on my part propoſe that I shall be allowed to
'print, to one-half the amount ſelected by Sir William Hamilton, of ad-
'ditional extract : but if this be refuſed I will not infift on it. With this
•I will put a heading fully defcriptive of the reaſon and meaning of the
' inſertion, and ſuch distinct reference and account at the beginning of
'the preface as ſhall be fure to call the reader's attention to it. So that
'my book ſhall eſtabliſh the claim, ifit can be eſtabliſhed from the notes
'of one ofthe beſt ſtudents. If this offer be not accepted, an account of
'it will take the place of any other reſult. If Sir William Hamilton, or
'any one elſe, can propoſe anything to make this offer fairer, I ſhall pro-
'bably not be found indiſpoſed to accept the addition. And though, I
'will frankly ſay, my preſent conviction is that the acceptance of the
'offer would alone cauſe my work to knock Sir William Hamilton's
'affertions to atoms, yet I will pledge myself, in any cafe, to abide by it.
Had our places in this diſcuſſion been changed, I ſhould have taken
care that no reader of my anſwer ſhould have been left in ignorance of
ſo fair an offer on the part of my opponent : more eſpecially if that
opponent had been accuſed by me of fraud and falſehood, in a manner
which I felt obliged formally to retract. But Sir Wm. Hamilton does
not notice the offer, even by an alluſion : and refers to the notes in the
following way : -

' In regard to Mr - and his Notes, I beg leave to ſay, that in

'my relative letters, neither to that gentleman nor to you, did I ever
' refer to his Notes of my lectures, but excluſively to his personal infor-
' mation in regard to them. And for a fufficient reaſon. The Paragraphs
'on Logic dictated to, and taken down by, my ſtudents, on which I after-
'wards prelect, were written ſo far back as the year 1837, and prior to
،

many ofmy new views, and to the whole doctrine ofa quantified predi-
' cate. Theſe views, as developed, were, and are, introduced in a great
 Appendix . 315
meaſure as corrections of the common doctrine ; in the older Notes
' eſpecially, they may, therefore, not appear in the dictated and numbered
Paragraphs at all ; whilft, frequently, (particularly at firſt,) they were
'given out as data, on which, previous to farther comment, the ſtudents
' were called on or excited to write expofitory Eſſays. I distinctly recol-
'lect, that in the Seffion during which Mr. attended my courſe of
' Logic ( 1840-1 ) it was required, on the hypothefis of a quantified pre-
' dicate,-to ſtate in detail, the valid moods of each fyllogiftic figure ; and
' I, further, diſtinctly recollect, that Mr. was one of those who
' efſſayed this problem. If wrong on this point, I shall admit that my
'memory is as treacherous as yours. It was, indeed, quite natural, that
Mr.- ſhould give, and that you should receive, his Notes ; but,
'of courſe, you could have fought or obtained no perſonal information
'from him, in reference to the point in queſtion, without mentioning the
'fact. Were it, however, requifite to give proof from Notes of fo
....

' manifeſt a fact, I doubt not that ſcores of ſtudents would be willing to
'place theirs at my diſpoſal.
On the appearance of Sir W. Hamilton's pamphlet, Mr.
wrote him a very straightforward letter, of which he fent me a copy,
with permiffion to both of us to uſe it. The general tenor is that Sir
W. Hamilton is correct in his ſtatements of what he had taught (which
ſtatements I never impugned as to fact ; I did not know what they
meant). On the point in queſtion Mr. fays (the Italics are
mine) ;-
During the Seſſion in which I attended your lectures ( 1840 and
' 1841) your new ſyſtem, baſed on the thorough going quantification of
'the predicate (the fecond of the three ſyſtems mentioned in page 31 of
'your publiſhed letter) and its conſequences in making all propofitions
'fimply convertible &c. was not developed by you in your ordinary feries
' of Lectures. I believe it was not touched upon in them, but it was partly
'explained to the class verbally, and then given out as a subjectfor Ef-
Says. When the Eſſays were given in they were read aloud in the claſs,
'and commented upon by you, and in ſo doing you fully explained the
' ſyſtem as " a full extenfion and thereby a complete ſimplification of the
' fyllogiftic theory."
Theſe facts which were ſtrongly fixed in my memory, becauſe I
'believe on that occaſion I happened to be the only Effayiſt who had
' rightly apprehended and worked out the theſis, will account for the
'circumſtance that my notes, which were originally taken in ſhorthand,
'although containing a full Report of all your ordinary Lectures, are
' completely filent on the ſubject."
The reader may find out, if he can, where Sir W. Hamilton re-
ferred to perfonal information as diftinguiſhed from notes, or to his
teaching of his new ſyſtem, as a matter diſtinct from that of his ordinary
lectures : and muſt judge what his ſucceſs is in ſaying what he means

* I think this ſhould be extempore : meaning that Sir W. Hamilton uſually reads his
lectures .
316 Appendix.
to ſay. And he may find out further, how I was to gueſs that the
mere commonplace of the topics diſcuſſed in Sir William's teaching was to
come, after an interval of fix years, from his old pupil's personal infor-
mation, and not from the full and (as I found them) excellent notes which
he made at the time.
I should add that Mr. -, ſubſequently to the printed contro-
verſy, anſwered every query which I put to him on Sir W. Hamilton's
ſyſtem, but did not feel juſtified (as in a like caſe I ſhould not either) in
anſwering poſitively as to the minute details of it, after laying it by for
years .
I have mentioned one or two inſtances in which, as ſeems to me,
Sir W. Hamilton has a ſtrange idea of the ſenſe of his own words : I
will now take one of the cafes in which he has dealt as ſtrangely with
mine. The way in which we uſe language, is one of the means which
the reader has, for forming his judgment on the whole of this difpute :
and he must decide which of us is incapable of giving to the phrases of
the other their proper ſignification.
When I returned to Sir W. Hamilton his profpectus, with thoſe
parts underlined which I could interpret in my own ſenſe, the more
important parts relating to logical mood and figure were not thus un-
derlined. In the accompanying letter, I uſed theſe words, ' To mood
and figure, I have attended but little ; what I get on theſe points will
bė from your hint, or from your book.' The whole letter was on what
I had done in the way of inveſtigation, not of elementary reading : and
I may fafely ſay that it is clear I meant that I had not made mood and
figure, as conſtituent parts of a theory of fyllogifm, ſubjects of investiga-
tion, with a view to new properties. But Sir W. Hamilton, in two
places, makes me avow ignorance of the ordinary ſyſtem of mood and
figure. In a foot-note to the above, he says, " And yet, though con-
fefſedly to ſeek in the very alphabet of the ſcience, Mr. De Morgan
' would be a logical inventor ! What is here acknowledged in terms, is
'fufficiently manifeſted from mistakes.” * And in his pamphlet ( II. p.
9), he reprefents me as ' no proficient no thorough ſtudent, in the
ſcience ; and refers to this paragraph of mine as the ground of the
afſſertion. It would have been ſtrange, if, avowing ignorance of the
ordinary doctrine of mood and figure, I had said that what I ſhould get
on theſe points must be from Sir W. Hamilton's hint or unpubliſhed
book, when any ordinary treatiſe would have given it: ſo ſtrange, that
this clauſe ought, I think, to have ſuggeſted the obvious meaning. Is
Sir W. Hamilton's interpretation a fair one ? I do not doubt that he
meant it to be fair. What I aſk is, has he the power to read fairly as
well as the will ?

The two preceding cafes (that of the notes and that of the avowed
ignorance) are ſpecimens of Sir W. Hamilton's give and take, of the

* Sir W. Hamilton ſhould have cited a few : but when he declares I have made
elementary blunders, he does not give ſo much as a reference. The plan is a ſafe one.
 Appendix . 317
manner in which he expects to be understood, and ofthat in which he
claims a right to understand. They are alſo, of courſe, ſpecimens of
my own.
In (A), the ſymbols A, E, I, O, are the A , E , I , O , of this work :
and a, e, i, o, are the A', E', Ι', Ο '.
(A) From the paper as sent to Cambridge before I had any communica-
tion whatsoeverfrom Sir William Hamilton (without any corrections).

SECTION III. On the quantity ofpropofitions.

" The logical uſe of the word some, as merely more than none,'
needs no further explanation. Exact knowledge of the extent ofa pro-
pofition would confift in knowing, for instance in ſome Xs are not Ys'
both what proportion of the Xs are ſpoken of, and what proportion
exists between the whole number of Xs and of Ys . The want of this
information compels us to divide the exponents of our proportion into
o, more than o not neceffarily 1, and 1. An algebraiſt learns to con-
ſider the diſtinction between o and quantity as identical, for many
purpoſes, with that between one quantity and another : the logician
muſt (all writers imply) keep the distinction between 0 and a, however
fmall a may be, as facred as that between 0 and 1-a ; there being but
the fame form for the two cafes . We shall now ſee that this matter
has not been fully examined.
" Inference must conſiſt in bringing each two things which are to be
compared into comparison with a third. Many compariſons may be
made at once, but there must be this proceſs in every one. When the
compariſon is that of identity, of is or is not, it can only be in its ulti-
mate or individual cafe, one of the two following :- This X is a Y,
this Z is the very fame Y, therefore this X is this Z ; or elfe ' This X
is a Y, this Z is not the very fame Y, therefore this X is not this Z.'
And collectively, it must be either Each of theſe Xs is a Y ; each of
theſe Ys is a Z ; therefore each of theſe Xs is a Z ; or elfe ' Each of
theſe Xs is a Y, no one of theſe Ys is a Z, therefore no one of theſe
Xs is a Z.'
"All that is eſſential then to a ſyllogiſm is that its premiſes ſhall
mention a number of Ys, of each of which they ſhall affirm either that
it is both X and Z, or that it is one and is not the other. The pre-
miſes may mention more : but it is enough that this much can be picked
out; and it is in this laſt proceſs that inference confifts .
" Aristotle noticed but one way of being ſure that the ſame Ys are
ſpoken of in both premiſes ; namely, by ſpeaking of all of them in one
at leaſt. But this is only a caſe of the rule : for all that is neceſſary is
that more Ys in number than there exiſt ſeparate Ys shall bespoken of in
bothpremises together. Having to make m+ n greater than unity, when
neither m nor n is ſo, he admitted only that caſe in which one of the
two m or n, is unity and the other is anything except o. Here then
are two fyllogifms which ought to have appeared, but do not,
318. Appendix .
Most of the Ys are Xs Most of the Ys are Xs
Most of the Ys are Zs Most of the Ys are not Zs
... Some Xs are Zs ... Some of the Xs are not Zs

And instead of moſt, or ++ a, of the Ys, may be ſubſtituted any two

fractions which have a ſum greater than unity. If theſe fractions be m
and n, then the middle term is at least the fraction m + n - 1 of the Ys .
It is not really even neceſſary that all the Ys ſhould enter in one pre-
miſs or the other : for more than the fraction m+n - 1 of the whole
may be repeated twice.
"And in truth it is this mode of fyllogifing that we are frequently
obliged to have recourſe to ; perhaps more often than not in our uni-
verſal fyllogifms. All men are capable of fome inſtruction ; all who
are capable of any inſtruction can learn to diftinguiſh their right and left
hands by name; therefore all men can learn to do ſo.' Let the word
all in theſe two caſes mean only all but one, and the books on logic tell
us with one voice that the fyllogifm has particular premiſes, and no con-
clufion can be drawn. But in fact idiots are capable of no inſtruction,
many are deaf and dumb, ſome are without hands : and yet a conclufion
is admiſſible. Here m and n are each very near to unity, and m +n- I
is therefore near to unity. Some will say that this is a probable con-
cluſion : that in the cafe ofany one perfon it means there is the chance
m that he can receive inſtruction, and n that one ſo gifted can be made
to name his right and left hand : therefore m× n (very near unity) is
the chance that this man can learn ſo much.
" But I cannot ſee how in this inſtance the probability is anything
but another fort of inference from the demonftrable concluſion of the
fyllogifm, which must exiſt under the premiſes given. Besides which,
even if we admit the ſyllogiſm as only probable with regard to any
one man, it is abſolute and demonftrative in regard to the propofition
with which it concludes.
" But this is not the only caſe in which the middle term need not
enter univerſally : this however is matter for the next Section. I now
go on to another point."

Extract II.

" I now take the two cases in which particular premiſes may give a
conclufion : namely
I, XY+XY=XZ XY+Y : ZX : Z Οτο

on the ſuppoſition that the Ys mentioned in both premiſes are in num-

ber more than all the Ys. If Y,, and Y, ſtand for the fractions of the
2

whole number of Ys mentioned or implied in the two premiſes, and y I

and y₂ for the fractions of the ys implied or mentioned, we ſhall by a

 Appendix. 319
repetition of the proceſs on YX +YZ = XZ (the other being obtained
in the course of the proceſs) arrive at the following reſults or their
counterparts : remembering that Y₁ +Y₂ is greater or less than 1 , ac-
2

cording as y ,ty, is leſs or greater.

2

Deſignation. Syllogifm. Condition of its exiſtence.

II YX + YZ = XZ Y1+Y2 greater than I

I

Ο1ο
Jo YX + Y: Z = XZ
100 Y : X + Y : Z = xz
Ooi X : Y + yz = X : Z Y +Y , less than I
I 2

yx + yz = XZ
Ooi X: Y + yz = X : Z ....

Ioo
00 X : Y + Z : Y = XZ

(B) Communication received on the 4th or 5th of November from Sir

William Hamilton, being the pretextfor his charge that I have, with
injurious breach of confidence towards himself, and falfe dealing to-
wards the public, appropriated his “ Fundamental Doctrine of Syllo-
gifm" privately communicated to me : and, after the retraction of that
charge, noticed in pages 297,8 , for the affertion that I have done the
Same thing unconsciously.
" 16, Great King Street,
November 2nd, 1846 .
" DEAR SIR, I have been longer than I anticipated in anſwering
your last letter. I now ſend you a copy of the requisites for the prize
Eſſay, which I gave out to my ſtudents at the cloſe of laſt ſeſſion. It
will show you the nature of my doctrine of fyllogifm, in one of its
halves. The other, which is not there touched on, regards the two
wholes, or quantities in which a ſyllogiſm is caſt. I had intended ſend-
ing you a copy of a more articulate ſtatement which I meant, at any
rate, to have drawn up ; but I have not as yet been able to write this.
I will send it when it is done. From what you ftate of your ſyſtem
having ' little in common with the old one,' and from the contents of
your Firſt Notions, we shall not, I find, at all interfere, for my doctrine
is fimply that of Aristotle, fully developed.
It will give me great pleaſure if I can be of any uſe, in your inveſti-
gations concerning the history of Logical doctrines. I have paid great
attention to this ſubject, on which I found, that I could obtain little or
no information from the profeſſed hiſtorians of Logic ; and my collec-
tion of Logical books is probably the most complete in this country.
But, as I mentioned to you in my former letter, it is only in fubordinate
matters that in abstract Logic there has been any progreſs .
" I remain, dear Sir, very truly yours,
" W. HAMILTON ."
 320 Appendix.
Essay on the new Analytic of Logical Forms.
Without wiſhing to preſcribe any definite order, it is required that
there ſhould be ſtated in the Effay,-
1 °. What Logic poftulates as a condition ofits applicability.
2°. The reasons why common language makes an ellipfis ofthe ex-
preffed quantity-frequently of the subject, and more frequently of the
predicate, though both have always their quantities in thought. [This
paragraph is the one on which Sir W. Hamilton principally relies].
3 ° . Conversion ofpropofitions on the common doctrine.
4° . Defects of this.
5 ° . Figure and Mood of Categorical ſyllogifm, and Reduction, on
common doctrine (General ſtatement).
6° . Defects of this (General ſtatement).
7 ° . The one fupreme Canon of Categorical Syllogiſms .
8°. The evolution, from this canon, of all thespecies of Syllogifm .
9°. The evolution, from this canon, of all the general laws of cate-
gorical Syllogifms .
10° . The error of the ſpecial laws for the ſeveral Figures of Catego-
rical Syllogifm.
11 ° . How many Figures are there.
12 ° . What are the Canons of the ſeveral Figures.
13°. How many moods are there in all the Figures : ſhowing in con-
crete examples, through all the Moods, the unessential variation which
Figure makes in a fyllogifm.
(Thoſe which follow 13 ° were wrong numbered. )
15° . What relation do the Figures hold to extenſion and comprehen-
fion.
16°. Why have the second and third Figures no determinate major
and minor premiſes and two indifferent conclufions : while the first Fi-
gure has a determinate major and minor premise, and a single proximate
conclufion.
17°. What relation do the Figures hold to Deduction and Induction.
N.B. This Eſſay open for competition to all ſtudents of the claſs of
Logic and Metaphyfics during the laſt or during the enſuing ſeſſion.
April 15th, 1846.

(C) Extract from the Addition to my Paper, taken, as can be shown,

from the papers which I gave the means of identifying in January
last, and which papers (though I hold it immaterial) I affert to have
been written before I received any logical communication from Sir
William Hamilton. (To be compared with the extracts given in A).
" Since this paper was written, I found that the whole theory of the
fyllogifm might be deduced from the confideration of propoſitions in a
form in which definite quantity of aſſertion is given both to the ſubject
and the predicate of a propofition. I had committed this view to
paper, when I learned from Sir William Hamilton of Edinburgh, that
 Appendix. 321
he had for fome time paſt publicly taught a theory of the fyllogifm
differing in detail and extent from that of Ariftotle. From the pro-
ſpectus of an intended work on logic, which Sir William Hamilton has
recently iſſued, at the end ofhis edition ofReid, as well as from infor-
mation conveyed to me by himself in general terms, I ſhould ſuppoſe it
will be found that I have been more or leſs anticipated in the view juſt
alluded to. To what extent this has been the cafe, I cannot now
afcertain ; but the book of which the profpectus juſt named is an
announcement, will fettle that queſtion. From the extraordinary extent
of its author's learning in the history ofphilofophy, and the acuteness of
his written articles on the ſubject, all who are intereſted in logic will
look for its appearance with more than common intereſt.
" The footing upon which we ſhould be glad to put propofitions, if
our knowledge were minute enough, is the following. We should ſtate
how many individuals there are under the names which are the ſubject
and predicate, and of how many of each we mean to fpeak. Thus,
instead of Some Xs are Ys,' it would be, Every one of a ſpecified Xs
is one or other of b ſpecified Ys.' And the negative form would be as
in No one of a ſpecified Xs is any one of b ſpecified Ys.' If propofi-
tions be ſtated in this way, the conditions of inference are as follows.
Let the effective number of a propoſition be the number of mentioned
cafes of thefubject, if it be an affirmative propoſition, or of the middle
term, if it be a negative propofition. Thus, in ' Each one of 50 Xs is
one or other of 70 Ys,' is a propofition, the effective number of which
is always 50. But No one of 50 Xs is any one of 70 Ys ' is a propo-
fition, the effective number of which is 50 or 70, according as X or Y
is the middle term of the ſyllogiſm in which it is to be used. Then
two propoſitions, each of two terms, and having one term in common,
admit an inference when I. They are not both negative. 2. The
ſum of the effective numbers of the two premiſes is greater than the
whole number of exifting cases of the middle term. And the exceſs of
that fum above the number of cafes of the middle term is the number
of the cafes in the affirmative premiſs which are the fubjects of inference.
Thus, if there be 100 Ys, and we can ſay that each of 50 Xs is one or
other of 80 Ys, and that no one of 20 Zs is any one of 60 Ys ;-the
effective numbers are 50 and 60. And 50+60 exceeding 100 by 10,
there are 10 Xs, of which we may affirm that no one of them is any
one of 20 Zs mentioned.
" The following brief ſummary will enable the reader to obſerve the
complete deduction ofall the Aristotelian forms, and the various modes
of inference from specific particulars, of which a ſhort account has
already been given.
" Let a be the whole number of Xs ; and t the number ſpecified in
the premiſs. Let be the whole number of Zs ; and w the number
ſpecified in the premiſs. Let b be the whole number of Ys ; and u and
v the numbers ſpecified in the premiſes of x and z. LetX,Y denote
that each of t Xs is affirmed to be one out of u Ys : and X, Y, that
:

each oft Xs is denied to be any one out of u Ys. Let X , ſignify m

Y
322 Appendix.
Xs taken out ofa larger ſpecified number n ; and ſo on. Then the five
poſſible fyllogifms, on the condition that no contraries are to enter either
premiſes or conclufion, are as follows :-
1. XY +ZY = Xt + w-bit Z20=21+20-6, 201w

2. XY +YZ = Xi+ v-b,t Zw=2 tv-

+ v6,b, wτοΧρ
3. YX+YZ = Xu +v-b,& Zw=Zu + v-b, wX
t t

4. XY +Z : Y = X + v-b, t : Zw
5. YX +Z : Y = Xu + vd : Zw
"The condition ofinference expreſſes itſelf; in the Xm, of the con-
t

cluſion, m must neither be o nor negative. The firſt caſe gives no

Ariftotelian fyllogifm ; the middle term never entering univerſally (of
neceffity) into any of its forms, under any degree of ſpecification which
the uſual modes ofſpeaking allow. The other caſes divide the old fyl-
logiſms among themselves in the following manner : they are written ſo
as to ſhow that there is ſometimes a little difference of amount of ſpeci-
fication between the reſults of different figures, which changes in the
reduction from one figure to another. The Roman numerals mark the
figures.
2.
t = a, v= bY)Z + X)Y =X)Ζ , το Barbara I.
t = a, v = bX)Y + Y)ZZ , X w Bramantip IV .
t < a, v= bY)Z + X,Y =XZ , 20 w
Darii I.
t < a, v = bXY +Y)Z =Z , wXt. Dimaris IV .

3. u = b, v = bY)X + Y)Z = Z6, 20X , Darapti III.

u < b, vbYX + Y)Z =ZwXu, t Diſamis III.
u = b, v < b | Y)X +Y )Zw=Zv, wXv, Datifi III.
4. t = a, v = b, w = cY.Z + X)Y = X.Z Celarent I.
t = a, v = b, w = ι Ζ.Y +X)Y = X.Z Cefare II.
t = a, v = b, w = cX) Y +Z.Y= Z.X Cameftres II.
t = a, v = b, w = ι Χ)Y +Y.Z= Z.X Camenes IV .
v = b, w = ι Y.Z =XY = X : Z Ferio I.
v = b, w = cZ.Y+ XY = X : Z Feftino II.
t = a, v = b, X) Y + Z : Y=Z : X Baroko II.

5. u =b, v = b, w = cY.Z + Y) X =X : Z Felapton III.

u = b, v = b, w = ι Ζ . Υ+Y)X =X , Z Fefapo IV .
v = b, w = ι Υ.Z+YXX , t: Z Ferifſo III.
v = b, w =ι Ζ.Y+YXX , t: Z Frefifon IV.
u = b, W= c
Y : Z+Y)X =X : Z Bokardo III .

I conclude by fubmitting to the reader what I began with, namely,

that until Sir William Hamilton produces ſomething from C, intelligi-
bly hinted at in B, and neither ſubſtantially contained in the matter, nor
 Appendix. 323
immediately deducible from the principles, ofA, he has no right what-
ever to affert that I have borrowed from him confcioufly or unconfci-
oufly. I have not found any perſon who thinks that ſuch a thing can
be produced : and I leave every reader to form his own opinion whether
it can be done or not.

APPENDIX II .

Onsomeforms ofinference differing from those of the Ariftotelians.

THINK it defirable to ſtate all I know of any attempt todeal with
I the forms of inference otherwise than in the Ariftotelian method .
Since the time of Wallis, three well known mathematicians have written
on the fubject, Euler, Lambert, and Gergonne : there may have been
others, but I have not met with them.
Euler's Lettres à une Princeſſe d'Allemagne ſur quelques ſujets de
Phyſique et de Philofophie' (3 vols. 8vo. Peterſburg 1768-1772, accord-
ing to Fuſs) contain the repreſentation of the ſyllogiſm byſenſible terms,
namely, areas. There was a Paris edition by Condorcet and Lacroix,
in 1787, as is ſtated by Dr Henry Hunter, who publiſhed an Engliſh
tranflation from it and from the original edition, London, 1795, 2 vols.
8vo. Euler makes use of circles to repreſent the terms. In a tract
publiſhed (or completed) in 1831, in the Library of Useful Knowledge,
under the name of the Study and Difficulties of Mathematics' I fell
upon this method before I knew what Euler had done, uſing, for dif--
tinction, ſquares, circles, and triangles, as in Chapter I. of this work.
The author of the " Outlines" preſently mentioned, has what I con-
fider a very happy improvement on Euler. The propoſition ' fome X
is Y,' is reprefented by the latter as the circle of X, partly inſide and
partly outſide the Y. The author of the "Outlines" puts a broken
fegment of the circle of X inſide the circle of Y, leaving it unfettled
whether the rest of the circle is united to the broken piece, or tranf-
ferred elfewhere. *
But Euler had been preceded in the publication of this idea by Lam-
bert, in his ' Neues Organon, &c.' Leipzig, 1764, 2 vols. 8vo. In
this work, the terms are reprefented by lines, and identical extents by
parts of the lines vertically under one another, as in page 79. The
whole notion is repreſented by continuous line, the part left indefinite
in particular propoſitions by dotted line. Some ofthe contranominal
forms are more diſtinctly mentioned than is usual, but there is no intro-
duction that I can find ofany form of inference which is not Ariftote-
lian.

* I ſhould say that Euler does not use the numerical, but the magnitudinal notion,
(fee page 48 of this work).
324 Appendix.
In the ſeventh volume of the Annales de Mathématiques (Niſmes,
1816 and 1817, 4to.) there is a paper by the editor, M. Gergonne,
entitled Effai de dialectique rationelle. I did not fee this paper, nor
Lambert's work, until after my memoir in the Transactions of the
Cambridge Society had been publiſhed. The ſecond would have given
me no hint : the firſt might have done ſo. There is the idea, and fome
formal uſe, of a complex propoſition : but the divifion is erroneous.
The fubidentical, identical, and ſuperidentical forms are there ; theſe
are not easily miſſed : the others whichGergonne uſes are, the complete
exclufion (the contrary orsubcontrary of my ſyſtem, which, disjunctively,
are only the common univerſal negative) and partial inclufion with par-
tial exclufion (the complex particular, or fupercontrary, of mine). The
ufe of contraries is expreſſly* forbidden, the old converfion by contra-
poſition formally declaredfalſe, and the particular propoſition aſſerted
to be incapable of being made univerſal. But M. Gergonne's complex
propofitions, fuch as they are, are uſed in a manner reſembling that in
chapter V, of this work, though requiring a ſeparate tâtonnement for
many things the analogues of which appear as connected reſults of my
ſyſtem. Accordingly, I am bound to attribute to M. Gergonne the firſt
publication of the idea of a complex fyllogifm, and of the comparison of
the fimple one with it. But numerical ſtatement is not hinted at.
Sir William Hamilton's ſyſtem dates, as to its publication in lectures,
from 1841 , as far as has yet appeared. What I have to ſay of it will
be found in another appendix.
In 1842, there was publiſhed anonymously ' Outline of the laws of
thought' ; London and Oxford (Pickering, and Graham) octavo in twos
(ſmall). The author is the Rev. Wm. Thomson, tutor of Queen's
College, Oxford. It is a very acute work, and learned. The ſyſtem of
propoſitions is extendedby the introduction ofboth the common quanti-
fications of the predicate into the affirmatives only, which introduces the
propofitions U and Y, as the author calls them, or " All Xs are all Ys,"
and " Some Xs are all Ys ."
The memoir in the Cambridge Tranſactions in which I gave the firſt
account of what has fince grown into Chapters IV, V, VIII, and X, of
this work, is deſcribed as to date in the preceding appendix. With re-
ference to the ſubject of chapter V, I may note the following defects
of that memoir: 1. That only one arrangement of X and Z as pre-
miſes being taken, only half the ſyſtem is given, and many correlative
arrangements are not obtained (fee page 140). 2. That owing to my
not feeing distinctly that each univerſal propoſition has two weakened
forms, the fyllogifms AA'I' and E'E'I are confidered as a claſs apart.
3. That much of the power of forming eaſy rules is not gained, by the
order of reference being made XY, ZY, XZ, instead of XY, YZ, XZ.
The former appears at firſt the more natural order, and is certainly

* I am told that ſome works on logic uſed in the Irish colleges formally announce
that thetruth of the [ordinary] laws offyllogifm depends upon the exclufion ofcontra-
ries : but I have not met with any of them.
 Appendix. 325
more easily deſcribed ; namely, to refer each of the concluding terms to
the middle term, with which both are compared. I obſerve, fince,
that M. Gergonne adopts this last order of reference : but the other is
by an immenfe deal more convenient in its reſults, as I think I have
ſhown .
With reſpect to the numerical quantification, what I did in the Me-
moir and Addition is given in full in the preceding appendix. Sir
William Hamilton, who distinctly renounces all claim to the " arithme-
tically articulate" ſyſtem, and doubts whether it afford any basis for a
logical developement, ſtates that he had formerly obtained the " ultra-
total quantification" (page 317) and thrown it away as a cumbrous and
uſeleſs ſubtlety, without publiſhing it, as I understand, in any way. To
his reply, he appends a note which I think it defirable to republiſh at
length, as a document in the hiſtory of this ſpeculation, and that I may
make that hiftory complete (II. p. 41) .
' I have avoided, in the previous letter and poſtſcript, all details in
' regard to the third ſcheme ofquantification (p. 32) ; becauſe that ſcheme
'except in ſo far as it is confounded with thefecond, has no bearing in
'the controverfy ; and I admit that whatever Mr. De Morgan has
' therein accompliſhed, he has accompliſhed independently of me. Fur-
' ther, I shall not deny him any claim of priority to whatever he may
'have ſtated in our correfpondence, in reference to this third ſcheme.
Finally, I shall acknowledge, for I think it not improbable, that his
' fyllogifm (p. 19) ſuggeſted a reconfideration, on my fickbed, of a cer-
'tain former ſpeculation, in regard to the ultratotal quantification of the
'middle term in both premiſes together ;-a ſpeculation determined by
' the vacillation of the logicians, touching the predeſignations more, most,
' &c. but which I had laid, afide, as a uſeleſs and cumbrous ſubtlety.
•Aristotle, followed by the logicians, did not introduce into his doc-
'trine of fyllogifm, any quantification between the abſolutely univerſal
'and the merely particular predeſignations, for valid reasons .-1 °, Such
quantifications were of no value or application in the one whole (the
' univerſal, potential, logical), or, as I would amplify it, in the two cor-
' relative and counter wholes (the logical, and the formal, actual,
'metaphyfical,) with which Logic is converſant. For all that is out of
'claſſification, all that has no reference to genus and ſpecies, is out of
Logic, indeed out of Philofophy ; for Philofophy tends always to the
'univerſal and neceſſary. Thus the highest canons ofdeductive reaſon-
' ing, the dieta de Omni et de Nullo, were founded on, and for, the
'procedure from the univerſal whole to the fubject parts ; whilſt, con-
' verſely, the principle ofinductive reaſoning was eſtabliſhed on, and for,
' the (real or prefumed) collection ofall the ſubject parts as conftituting
' the univerſal whole.-2°, The integrate or mathematical whole, on
'the contrary, (whether continuous or difcrete) the philoſophers con-
' temned. For whilſt, as Aristotle obſerves, in mathematics genus and
' ſpecies are of no account ; it is, almoſt excluſively, in the mathemati-
'cal whole, that quantities are compared together, through a middle
' term, in neither premiſe, equal to the whole. But this reaſoning, in
326 Appendix .
'which the middle term is never univerſal, and the conclufion always
' particular, is, as vague, partial, and contingent,-of little or no value
' in philoſophy. It was accordingly ignored in Logic; and the prede-
' fignations more most, &c. , as I have faid, referred, to univerfal, or,
(as was most common) to particular, or to neither, quantity. This
•difcrepancy among Logicians long ago attracted my attention ; and I
'faw, at once, that the poſſibility of inference confidered abſolutely,
'depended, excluſively on the quantifications of the middle term, in both
'premiſes,being, together, more than its poffible totality-its diftribution,
'in any one. At the fame time I was impreſſed-1 , with the almoft
' utter inutility of fuch reaſoning, in a philofophical relation : and 2°,
'alarmed with the load of valid moods which its recognition in Logic
'would introduce. The mere quantification of the predicate, under the
' two pure quantities of definite and indefinite, and the two qualities of
' affirmative and negative, gives (abſtractly) in each figure, thirty fix
' valid moods ; which, (if my preſent calculation be correct,) would be
'multiplied, by the introduction of the two hybrid or ambiguous quan-
' tifications of a majority and a half, to the fearful amount offour hun-
' dred and eighty valid moods for each figure. Though not, at the
'time, fully aware of the ſtrength of theſe objections, they however
6

،
prevented me from breaking down the old limitation ; but as my fu-
preme canon of Syllogifm proceeds on the mere formal poſſibility of
' reaſoning, it of courſe comprehends all the legitimate forms of quanti-
' fication. It is ;-What worst relation offubject and predicate,fubfifts
' between either of two terms and a common third term, with which one,
' at least, is positively related ; that relation fubfifts between the two
' terms themselves : in other words ;-In as far as two notions both
،

agree, or one agreeing, the other disagrees, with a common third notion :
'-in so far, thoſe notions agree or disagree with each other. This canon
' applies, and proximately, to all categorical fyllogifms,-in extenfion
'and comprehenfion,-affirmative and negative, and of any figure. It
'determines all the varieties of ſuch fyllogifms ; is developed into all
' their general, and ſuperſedes all their ſpecial, laws. In short, without
'violating this canon, no categorical reaſoning can, formally, be wrong.
'Now, this canon ſuppoſes that the two extremes are compared together,
'through the fame common middle; and this cannot but be, if the
'middle, whether, ſubject or predicate, in both its quantifications to-
' gether, exceed its totality, though not taken in that totality in either
'premife.
But, as I have ſtated, I was moved to the reconfideration of this
' whole matter ; and it may have been Mr. De Morgan's fyllogifm in
' our correfpondence (p. 19), which gave the ſuggeſtion. The reſult
'was the opinion, that theſe two quantifications ſhould be taken into
' account by Logic, as authentic forms, but then relegated, as of little
'uſe in practice, and cumbering the ſcience with a fuperfluous maſs of
' moods. As to Mr. De Morgan's ſtatement in our correfpondence (p.
'21 ) of the principle on which (by his later ſyſtem) fuch fyllogifms
' proceed, this, to uſe his own expreffion, " I did not comprehend at
 Appendix. 327
' all ;" nor do I now,* having, to ſpeak with the Rabbis, " referved it
' for the advent of Elias." I saw however, that, be it what it might,
' it had no analogy with mine ; indeed, even from the fuller expofition
' of his doctrines, contained in the body ofthe Cambridge Memoir and
' its Addition, which I afterwards received, I can find no indication
' of his having generaliſed either, 1º the comprehensive principle of all
' inference, that the two quantifications of the middle term, should, to-
' gether, exceed it as a single whole ; or, 2°, under a non-distributed
' middle, the two exclusive forms of its quantification. On receipt,
however, of Mr. De Morgan's Cambridge Memoir, I ſaw, or thought
' I faw, in the body of the paper, on his old view, ſome manifeftation of
*a leſs erroneous doctrine upon this point, than that afterwards contained
' in his Letters and Addition, upon his new. Accordingly, to obviate
• all miſconſtruction, I wrote immediately the following letter, † ofwhich
'an account has been previously given (p. 26, note).
EDINBURGH, 30th March, 1847 .
،
Your paper read to the Society I have curforily peruſed ; but though
'oppoſed to many of its doctrines, I admire the ingenuity which charac-
* The paſſages which Sir William Hamilton does not underſtand, are the following,
and alſo that relating to the effective terms, in C of the preceding appendix.
" Now ſuppoſe propofitions in which the quantitative part of the preceding is made
more definite . Say that
Xt Yu and Xt : Yu
mean

Every one of t Xs No one of t Xs

is one or other of u Ys is any one of u Ys
Let the effective number of caſes in a propofition be the number which it makes ef-
fective in inference. Then the effective number in a poſitive propofition is the num-
ber of cafes of thefubje&
t.
The effective number in a negative propoſition is the number of caſes of the middle
term.
And the criterion of inference being poſſible, is that the fum of the effective num-
bers of the two premiſes (not both negative) is greater than the whole number of cafes
of the middle term.
And the exceſs is the number of caſes involved in the inference, of all which are
mentioned in the conclufion-term (or terms) of the poſitive premiſs (or premiſes).
For instance, let b be the whole number of Ys in exiſtence : I ask whether we can
infer anything from
Xt Y u
effective number t

Zw : Yv V
V

Anſwer, if t + v be greater than b, we can infer

Xt+v-b : Zw
Or, if each of t Xs be one or other of u Ys, and no one of w Zs be any one of v Ys,
then ift and v together are more in number than there are Ys, we may infer that no
one of t + v - b Xs is any one of the w Zs juſt ſpoken of."
† This letter (the first paragraph of which is omitted, as not relevant to this appen-
dix,) was addreſſed to me, and was ſent open to my friend Dr. Sharpey, to be deli-
vered to me. Dr. Sharpey refuſed to deliver (and, as it happened, I was as much
prepared to refuſe to receive) any thing on the literary ſubject matter ofthe controverfy
which did not contain a retraction of Sir W. Hamilton's then fubfifting charge againft
me. Accordingly, I never saw it till it appeared in print.
328 Appendix.
'teriſes it throughout. On one point, I find we coincide, in principle,
' at least, against logicians in general. They have referred the quantify-
'ing predefignations plurimi, and the like, to the moſt oppoſite heads ;
' ſome making them univerſal,-fome, particular, and fome between
' both ; (for you are not correct in ſaying, (p. 6), that logicians are
'unanimous in regarding them as particular, [though moſt do]). This
'confliction attracted my attention ; and a little confideration ſhowed
،

me, that befides the quantification of the pure quantities, univerſal and
'particular, (which I call definite and indefinite,) there are two others of
' theſe, mixed and half developed, which ought to be taken into account
' by the logician, as affording valid inference ; but which, without ſcien-
' tific error, cannot be referred either to univerfal, (definite,) or to par-
' ticular, (indefinite) quantity, far leſs left to vacillate ambiguoufly be-
' tween theſe. I accordingly introduced them into my modification, in
Engliſh doggerel, of " Afferit A,” &c. , which [in the original caft] I
' formerly faid was at your ſervice ; and as it affords a brief view of my
'doctrine on this point, I may now quote it.

'A, it affirms ofthis, that, all,*

Whilft E denies of any ;
I, it affirms, whilst O denies,
Offome (or few or many).

Thus A affirms, as E denies,

And definitely either ;
Thus I affirms, as O denies,
And definitely neither.

A half, left ſemi-definite,

Is worthy of its ſcore ;
U, then, affirms, as Y denies,
This, neither leſs nor more.

' Indefinito-definites,
To UI, YO , laſt we come ;
And that affirms, and this denies,
Of more, most, (half plus ſome).

" The rule of the logicians, that the middle term ſhould be once at
" leaſt diſtributed [or indiſtributable,] (i.e. taken univerſally or fingu-
" larly, definitely,) is untrue. For it is ſufficient, if, in both the
"premiſes together, its quantification be more than its quantity as a
" definite whole. (Ultratotal)" " It is enough for a
" valid fyllogifm, that the two extreme notions ſhould (or ſhould not),
" of neceffity, partially coincide in the third or middle notion ; and
" this is neceſſarily ſhown to be the cafe, if the one extreme coincide

* Better : ' A, it affirms of this, these, all.

 Appendix . 329
" with the middle, to the extent of a half, (dimidiate quantification) ;
" and the other, to the extent of aught more than a half, (ultradimi-
" diate quantification).
" The firſt and highest quantification of the middle term (..) is
" fufficient not only in combination with itſelf, but with any of all the
" three inferior. The second (. ,) fuffices, in combination with the
" highest, with itſelf, and with the third, but not with the lowest.
" The third (.) fuffices, in combination with either of the higher, but
" not with itſelf, far leſs with the lowest. The fourth and lowest (,)
"ſuffices only in combination with the higheſt." [1. Definite ;
" 2. Indefinito-definite ; 3. Semi-definite ; 4. Indefinite.]" "
Of the effect of this new ſyſtem of quantification in amplifying the
'fyllogiftic moods, (which in all the figures remain the fame,) I ſay no-
'thing. It should be noted, however, that the letters A, E, &c. do not
'mark the quantification [and qualification] of propofitions, (as of old)
' but of propofitional terms . The ſentences within inverted commas are
'taken from notes for the " Eſſay towards," &c.
• Before concluding, I ought to apologiſe, in the circumstances, for
' the details, that have inſenſibly lengthened out, of a part of my doc-
'trine, which I have found, to a certain extent, coincident with what
6

appears in your paper. I was anxious, however, that you and others
'ſhould have no grounds for furmiſing, that I borrowed any thing from
'my predeceſſors without due acknowledgment.-On ſecond thoughts,
'however, I deem it more proper to make this communication through
' a third party.'
The diſcuſſion between Sir William Hamilton and myself called a
very able third party into the field, who addreſſed the following letter
to the editor of the Athenaum, in whichjournal it was publiſhed, June
19, 1847.
• Sir,-As two great logical innovations-the one due to Sir William
•Hamilton, the other due to Mr. De Morgan uſed in conjunction, have
'led me to the ſimpleſt and moſt general formulæ of ſyllogiſm that ever
'have been given (formule which correct a ſerious miſtake into which
' both Sir William Hamilton and Mr. De Morgan have fallen), I think
' it will gratify thoſe intereſted in logical ſcience if you would give them
'publicity through your columns.
' n', n , nII, &c. are any numbers. When placed before a term, as
' n "xs, n" marks the total number of the claſs x; placed before a pro-
'poſition, it marks the number of things of which we mean to ſpeak.
Thus, n', of n'xs are of nys, means that a number ofthings n¹ are
'alleged to have both the characteriſtics x and y ; and are to the whole
' claſs of xs as n' to n", and to the whole claſs ofys as n¹ to nIII : fimi-
' larly with the negative propoſition n' of n'xs are not of nys, n
' things being here ſaid to have the characteriſtic x, and to want the
'characteriſtic y. It is clear, from the nature of a propoſition, that in
affirmatives, n' can never be greater than the least extenſive of the
terms, and in negatives never greater than the number of the claſs
• whoſe characteriſtic it is ſaid to have. But within theſe limits the pro
330 Appendix.
• portion n' to n" may be wholly undetermined ; we then mark it with
' the wordfome, we call this, with Sir William Hamilton, indefinite
'quantity. It may be perfectly determined; as of equality when we mark
' it with all, every, or, following Mr. De Morgan, any other arithmeti-
'cal proportion-as a half. (Sir William Hamilton has erred in calling
' a half, femi-definite ; it is thoroughly definite) . All this we call defi-
' nite quantity. Laſtly, the indefinitude may be reduced within limits
-indefinito-definite, as most, &c.
The firſt formula contains all fyllogifms with an affirmative conclu-
'fion, without any exception.
II III
' I. n' of n'xs are of nys
V

n' of n' zs are of nys

(n + n - n¹11) of n"xs are of n'zs
As Sir William Hamilton's principle takes away all diſtinction of
' ſubject and predicate in affirmative propoſitions, it will be ſeen that, by
'varying the proportions of the ſymbols, n', &c., every poſſible affirma-
'tive logical inference, in whatever mood or figure, emerges .
The fyllogifms with negative queſtions or conclufions, are not fo
' ſimple. They fall into two diviſions, according as, in the negative
'premiſs, the things ſpoken of have the characteriſtic of the extreme, or
' ofthe middle ; and from each of theſe, two concluſions, not one, are
'drawn, according as the things to be ſpoken of in the conclufion have
•the characteriſtic of the extreme in the affirmative premiſs, or of that
* in the negative premiſs .
' II.
n' of n xs are of nys
' n " of n'zs are not ofntys concludes ;
' doubly 1° (n² + n" - n") of nuxs are not of n'zs
2º (n² + n - n") of n'zs are not ofnixs.
' It is to this formula I referred as correcting a ſerious error into which
• Sir William Hamilton and Mr. De Morgan have fallen-of holding, as
' a general principle of all inference, that the two quantifications of the
' middle term ſhould exceed it as a whole ; for this ſyllogiſm proceeds
' wholly irreſpective ofthe total quantity ofthe middle, which is excluded
• from our ſymbolic conclufion .
' III. n' of n²xs are of nys
' n of ntys are not of n'z's concludes ; alſo,
' doubly 1 ° (n + n n ) of nuxs are not of n'zs
2º (n + n² + n" — n — п¹¹) of n'zs are not of nxs.
Such are the three ſymbolical formulæ of every poſſible logical infe-
' rence. I have the demonſtrations that theſe are in all their extent valid,
' and are the only poſſible forms ; but it is ſufficient to give here the re-
'fults.
6

• It will ſurpriſe no one who confiders that the negative propofition is

'not converted in the ſame ſenſe as the affirmative, that the negative
* fyllogiftic formulæ are not reducible to one. For the rule ofnegative
 Appendix . 331
'converfion changes the things ſpoken of, and is as follows : n' of n'xs
' are not of nys; converts (n + n - nt) of nys are not of nixs. The
'conſequence ofa form univerſally true, (n -n") of ntlys are not of nxs.
' As to the two concluſions, they are but the converſe of each other.
' It will not be difficult to interpret theſe, by n² = nt as every or n³ :
' n" indefinite =fome, &c. The uſual Ariftotelic forms will be ſeen to
'be derived from them. Thus the mood Cefare, and the correſponding
' indirect mood (or, if you will, the mood of the fourth figure, call it at
' another time Celantes or Cadere at will, but let it be Celantes for the
'nonce), come forth from the third formula.
' n = n gives no y is z ... n " : n' indefinite
‘n = " every z is y ... n² : n ™ indefinite.
Hence in Cefare, no x is z from our firſt,
' and in Celantes, no z is x from our fecond conclufion, and ſo of all
' the others.
• I owe it to Sir William Hamilton and Mr. De Morgan to say that
' without their improvements I could not have advanced one ſtep. Mr. De
• Morgan has even attempted a like reduction to general formulæ, and has
'failed, chiefly through a miſapprehenfion ofSir William Hamilton's prin-
' ciple ofquantified predicate. He has introduced a fuperfluous quantity,
-one logically uſeleſs, or worſe than uſelefs, as the reſult has ſhown.
This confufion explains his errors. Had it not been for this circum-
' ſtance, I ſhould not have had the honour of preſenting theſe formulæ
' to logicians.
' Permit me to add what I think alſo of ſome value. I am not of thoſe
'who think with Sir William Hamilton that the ſyllogiſm always pro-
' ceeds in the two counter wholes of intenfion and extenfion-that it
'muſt always be an involution or evolution in reſpect of claſſification.
This is, no doubt, true in the moſt important reaſonings ofſcience ; but
' it is not ſcientifically accurate to affert this univerſally.
Quality, which is the comprehenfive element, is ofthree kinds-not
two, as heretofore affirmed ; for fince Kant, the diviſion of affirmatives
• into analytic and ſynthetic, or (as Sir William Hamilton wiſhes) expli-
⚫cative and ampliative, has been establiſhed. James Bernouilli has puz-
zled himself to reduce theſe two to the fame form, but without ſucceſs ;
⚫ for that contains an immediate relation of part to whole, and only a re-
mote one of part to part, while this contains an immediate relation of
'part to part, and remote of part to whole. Theſe, as diſtinct kinds of
quality, are erroneoufly elided in language. As the words ampliative
' and restrictive are generally oppoſed in logic, perhaps we might replace
• the old diviſion of propofitions, according to quality, into affirmative and
negative-by one into Explicative, Ampliative, and Restrictive.
Where, then, both premiſes are ampliative, the ſyllogiſm proceeds
'purely by force ofextenfion. There is neither involution nor evolution
-neither induction nor deduction-but a paſſage or tranſition from one
' mark to another, or from claſs to claſs. Of this kind are all fingular,
' or, as Ramus calls them, proper fyllogifms. Let us call this new
 332 Appendix.
' claſs of fyllogifms traductive, to contraſt it with the inductive and de-
'ductive.
•The use oftheſe in philoſophy as independent modes ofinference will
' easily appear. When we collect the ſcattered fragments of our know-
'ledge into unity of ſcience, we uſe induction and inductive fyllogifm ;
'when we apply the principles of ſcience to ſpecial events ofthings, we
' uſe deduction and deductivesyllogism; butwhen, abandoning one ſcheme
'of claſſification, we transfer our knowledge directly to another, we uſe
' traduction and traductive syllogism. Thus, in political ſcience, what
'has been predicated by hiſtorians of men claſſed geographically is tranf-
'ferred to men claſſed according to conftitutions of government by tra-
'duction. This last eſcapes Sir William Hamilton's rule, and never
' concludes through a comprehenfive containing and contained.
' I ſhall not add, at preſent, any attempt to prove à priori the excluſive
' validity of fyllogiftic inference.
' I admit that I ought not, without good ground, to diſſent from a ma-
'tured opinion of Sir William Hamilton in any part ofphilofophy, ftill
'more in logic ; but I obey the force ofdemonſtration, and, as Ludo-
' vicus Vives ſaid in reſpect to Aristotle, Verecundè diſſentio.
'Yours, &c .
JAMES BROUN .
' Temple, June 9, 1847. '

My reply to this confifted in forwarding, on the fame 19th ofJune,

to the editor of the Atheneum, a ſummary of the reſults ofchapter VIII,
then written. This fummary appeared on the 26th : I do not infert it,
becauſe the chapter in queſtion is a better anſwer ; and though the pub-
lication faved my rights, the republication is unneceſſary. Mr. Broun's
three forms are the firſt (without the contranominal), the ninth , and the
eleventh, ofpage 161. Mr. Broun was wrong in deducing from the
two latter forms that the principle of the middle term was erroneous :
for in theſe very forms the two quantifications exceed the whole : being
the whole (in premiſe one) plus ſome (in the other). As to the ſuper-
fluous quantity, it only becomes fuperfluous when ſuch quantifications
are introduced as diftinguiſh ſpurious from admiſſible propoſitions : fee
pages 145, 146, in which it is ſhown that the forms are correct.
Nothing but cloſe comparison, and that after practice, would detect
the accordance of the two ſymbolic modes of expreffion in pages 145
and 161. I am not therefore ſurpriſed that Mr. Broun ſhould, having
obtained cafes of that in page 161 , pronounce that in page 145 erro-
neous .

In the anſwer which I made, I promised to ſtate distinctly how much

of the chapter was written before Mr. Broun's letter appeared. This
I now do. With the exception ofpages 145, 146, the matter of which
is mostly from my Cambridge Memoir, the whole of it was then written,
excepting fuch verbal alterations and occaſional introduction ofſentences,
as take place at the preſs, or at the last reading ofthe manufcript. I had
 Appendix . 333

thought that there would be no neceſſity to introduce thoſe pages, ex-

cept flightly, and in anſwer to certain objections which ſeemed likely to
occur. The examination which the aſſertion that they are erroneous
made me give my previous forms, pointed out the defirableness of intro-
ducing them as they now ſtand.
September 17, 1847. I had finiſhed the preceding appendix, when I
became aware of the existence of the Commentationes Philoſophicæ
Selectiores ' of Godfrey Ploucquet, of Tubingen, Utrecht, 1781 , quarto.
The last title (p. 561) is ' De Arte Characteriſtica. Subjicitur Methodus
calculandi in logicis, ab auctore inventa. 1763.' I find by a catalogue*
that this methodus calculandi had been previously publiſhed in 1773, at
Tubingen, at the end of a work entitled ' Principia de Subſtantiis et
Phænomenis : alſo that the Methodus demonſtrandi directé omnes
fyllogifmorum ſpecies ' of the fame author (which is probably the thing
I am going to deſcribe) was publiſhed at Tubingen in 1763. From the
title of a work which, I am informed, exiſts, namely, Sammlung der
Schriften welche von logiſchen Calcul des Prof. Ploucquet betreffen'
Tubingen, 1773 , one would ſuppoſe that this ſyſtem had obtained great
local currency. I give a ſhort account ofit : premiſing that Ploucquet
appears to have been a well informed mathematician, much given to
pure ſpeculation on mental ſubjects .
The calculus (a term which Ploucquet uſes in as wide a ſenſe as I do
when I call the contents of Chapter V. a part of the calculus of infe-
rence) conſiſts in the invention of a ſimple notation, and the mechanical
ſubſtitution, in one premiſe, of an identical equivalent to the middle term
therein contained, taken from the other premiſe (this last being one in
which the middle term is univerſal). There is neither uſe of contraries,
nor numerical definition : but there is every variety of quantity of the
predicate which can be produced by fimple converfion of the ordinary
forms. A term uſed univerſally is denoted by the capital letter ; par-
ticularly, by the ſmall letter : affirmation by juxtapoſition ; negation, by
interpofing <. Thus X)Y is Xy ; X.Y is XY ; XY is xy ; X:Y is
xY. The following is a complete ſpecimen :
Sint præ- Pm
miffæ SM

Calculo : smP quoddam s non eft P

Omnis ducatus eſt aureus
Quædam moneta non eft aurea.
Da
m>A
Calc. maD. ſeum D, quædam moneta non eft ducatus.
As Ploucquet ſeems to think that this actual application of the calculus
to concrete inſtances, by aid of their initial letters, is a material part of
* The second edition of Mr. Blakey's ' Eſſay on Logic ' recently publiſhed, contains
a catalogue of upwards of a thousand works on logic, briefly titled.
334 Appendix.
his ſyſtem, I have inſerted the caſe entire. The rationale of the ſyſtem
conſiſts in that ſubſtitution of identicals for each other, which I under-
ſtand Sir William Hamilton (with perfect truth) to employ in every
cafe. Thus we have in the above Some of the Ss are not any Ms,
are not thoſe Ms which make up all the Ps, are not therefore any Ps.'
This demand for identical ſubſtitutes requires both kinds ofquantity for
every predicate, and Ploucquet uſes them accordingly, as far as wanted
to eſtabliſh the Ariftotelian fyllogifms. Sir W. Hamilton goes further,
and invents fyllogifms for all the kinds of quantity. Thus Ploucquet
ufes mP or ' fome Ms are all the Ps ' and Pm or ' all Ps are not fome
ofthe Ms ; but not MP or pm .
At the fame time with the knowledge of Ploucquet I obtained that
of the work of a follower and extender, M. W. Drobitſch, author of
Neue Darstellung der Logik ... Nebſt einen logiſch-mathematiſchen
Anhange,' Leipzig, 1836, octavo. As far as the ſymbolic part is con-
cerned, Mr. Drobitſch begins by a convention which would reconcile
any one to the found, not merely of Barbara and Celarent, but even of
Baroko and Freſiſon. He makes S and P the ſubject and predicate of
the conclufion and M the middle term ; and puts the Ariftotelian vowel
between them : thus S)P is SAP, and P:S is POS. Hence his pre-
miſes may be mapsam or mop ſam ; and one of his ſyllogiſms is mep-
Samſep. In the algebraical part, he uſes large and ſmall letters for the
univerſal and particular, or for the whole and part extent of a term.
He alſo introduces the figns = and to fignify identity and (what I
call) fubidentity. This uſe of the mathematical ſigns involves an ex-
tenfion, which is made by all thoſe who ſignify the identity of X and
Y by X=Y. The mathematician thinks of extent as quantity only :
the logician includes both quantity and poſition. Thus when the for-
mer ſays that five feet are less than ſeven feet, he means any five feet,
be they part of the ſeven feet or not : the latter, when he says that X
is a name of leſs extent than Y, means not only that the former can be
contained in the latter, but that it is. To make negative propoſitions,
Mr. Drobitſch takes a limited univerſe (call it U, as I have done) an
extent greater than the utmoſt extent of all the names, otherwiſe inde-
finite. And here he falls into ſome confufion : X and Y being the
names, he ſays U must be ofgreater extent than X+Y : now if we had
X)Y, U need only be of greater extent than Y. If from the genus Y
be taken all the ſpecies X, the remainder is denoted by Y-Х. Ас-
cordingly, the contrary of X is U-X.
Mr. Drobitſch then lays down eight forms of predication, of which,
however, he only uſes the ordinary ones. And I cannot find out that
the limited univerſe, or the contrary, has any uſe except to furnish means
of notation. The eight forms are ;-firſt, X=y, or my X)Y ; ſecondly,
X= Y, or X)Y+Y)X ; thirdly, x =y, or XY ; fourthly, u =Y, or
Y)X ; fifthly, X <U-Y, which tells us that X is all contained in
what is left of the univerſe after Y is removed, or is X.Y ; fixthly,
X= z < Z < U-Y, a very roundabout way of ſaying that X isfubcon-
trary of Y, or X.Y+xy ; ſeventhly, x=U-Y or X:Y ; eighthly,
 Appendix . 335
x=X-Y, which tells us that Y is a ſubidentical of X, or Y)X+X : Y.
This is in fact a mixture of two ſyſtems, both in principle and nota-
tion. The forms are A₁ , A', O₁ (and O'), E , I , D, D₁ (and D'), and
C. Alſo C is virtually given : but E', I', C', do not appear. The
ordinary rules under which the mathematicians uſe = and <, remain
true in this logical uſe of them and thus there is an elegant mode or
exhibiting the inference in ſyllogiſms. For instance, in Camestres we
have P = m , S < U — M … < U -m : < U-P ; or S < U- P .
Itwould have been more confiftent to have made = , <, and ▷ , (in-
troducing this laſt) ſerve all purpoſes. But it has happened very often
that a ſyſtem ofnotation, already exhibited, has been extended by a better
one, and mended only, instead of being reconſtructed. Ploucquet had
uſed the large and ſmall letters, and for denial : the latter ſymbol a
ſtrange one, if mathematical analogy were intended. Mr. Drobitſch
has ingenioufly contrived that < ſhould repreſent denial, and has been
led to what might have uſefully amended all he had to begin with. Tak-
ing little x to repreſent a part of the extent of X, &c. and U for the ex-
tent of the univerſe, the following notation might have been adopted :
Firſt when and ▷ both include their limit, = . We should have

A XAY or Y X A' YX or XY
O x < U - Y or U_Yx O'y < U-X or U-Xy
E XU-Y or U-YX E' XU-Y or U—Y < X
IxY or Y x I' is inexpreſſible.
To expreſs I', we must invent a ſymbol for a part of U-X.
Next, when < and ▷ do not include their limits, we have
DXY or Y X C XU- Y or U-YX
DX= Y or Y=X CX- U-Y or U-Y-X
D' XY or Y < X C' XU-Y or U -Y < X
Pis inexpreſſible.
I am inclined to think that the repreſentation of quantity and location
both under one ſymbol is objectionable, if that ſymbol be one already
appropriated in mathematics to quantity only. I would on no account
accuſtom myſelf to read A < B as A is less than (becauſe a part of) B.
Mr. Drobitſch is much more complete than his predeceſſors in his enu-
meration of the various kinds of forites .
October 29, 1847. While this ſheet was paſſing through the preſs,
I became acquainted with " A fyllabus of logic, in which the views of
Kant are generally adopted, and the laws of fyllogifm ſymbolically ex-
prefied. By Thomas Solly, Eſq." Cambridge, 1839, 8vo. The
ſymbolical expreſſion here given is of a peculiar character : the algebraic
figns are adopted in a ſenſe which preferves the rules ofſign, while the
ſymbols reprefent the terms of the fyllogifm, or elſe the notions ofpar-
ticular and univerſal. Thus, if p ſtand for particular, u for univerſal,
and m for one of the terms of a fyllogifm, m= u or m - u = 0 implies
336 Appendix.
that m is a univerſal term, and (m - u)(n -p)= 0 implies the alternative
that either m is univerſal, or n is particular. By means of ſuch alter-
native relations, the conditions of validity of the various figures are
expreſſed. Mr. Solly contends for fix forms in each figure, by intro-
ducing all forms which have weakened conclufions, and proves àpriori,
from his equations, that fix and no more are poffible in each figure. If
I had admitted weakened forms, there would have been fixteen more
fyllogifms, which might be deduced, either from the eight univerſals, or
from the fixteen particulars.

THE END .

C. WHITTINGHAM, CHISWICK .

3
n
APR 2.3 1929