The Myth of the Line: Ford's Production of

the Model T at Highland Park, 1909-16

KAREL WILLIAMS, COLIN HASLAM and

JOHN WILLIAMS, with ANDY ADCROFT
and SUKHDEV JOHAL
Universities of Central Lancashire, East London and Aberystwyth

Myth is constituted by the loss of the historical quality of things; in

it things lose the memory that they were once made. The world
enters language as a dialectical relation between activities, be-
tween human actions; it comes out of myth as a harmonious
display of essences.
R. Barthes, Mythologies
So it is with Henry Ford's production of the Model T at Highland Park,
which is mythologised as the productive innovation of the moving
assembly line (or MAL) and as the marketing initiative of 'any colour so
long as it's black'. The MAL is central to the popular iconography of
mass production; in Modern Times a MAL drives Charlie Chaplin mad.
The MAL is also central to the academic historiography of Ford; before
and after the Nevins l history of the company, the MAL is treated as an
essential sign and cause of Ford's price cuts and efficiency gains. This
article questions the myth that the MAL was Ford's key productive
innovation before 1918 and aims more positively to restore some of the
historical qualities of activity and action which have been overlaid and
lost in the historiographic myth.
The article is organised in a fairly straightforward way into three
sections. The first section de-constructs the myth of the line through the
analysis of two recent accounts of Ford by Hounshe1l 2 and Lewchuk. 3
These accounts disseminate and reinforce the myth because they are
based on research in the Ford archive and they are regarded as authori-
tative by popularisers such as Womack, Jones and ROOS.4 The second
section of the article offers a reconstruction of Ford's activities which
initially makes the negative case against the line; while the third section
goes on to propose a positive reconceptualisation of Ford. This recon-
ceptualisation is based on a break with the traditional concept of mass
production and a re-definition of Ford's activity as high-volume repeti-
tive manufacturing.

Business History. Vo1.35. No.3. (1993). pp.~7

PUBLISHED BY FRANK CASSo LONDON
FORD AT HIGHLAND PARK 67
I
Hounshell and Lewchuk's main subject is something other than Ford,
because each of these authors treats Highland Park as a single chapter
episode in a longer book-length story. Thus Hounshell5 writes the
history of American manufacturing technology 'from the American
system to mass production, 1800-1932'; and Lewchuk 6 tells a transatlan-
tic story of American and British attempts to control the outcome of the
'effort bargain' in the twentieth century. Although their archive re-
search generates some revisions about the chronology of the MAL
innovation and its precursors,7 Hounshell and Lewchuk essentially con-
firm the old privilege of the line within a new framework. For
Hounshell, the innovation of the MAL at Highland Park in 1913 adds
the necessary ingredient which constitutes mass production, while for
Lewchuk the line is a uniquely efficacious method of labour control. In
each case the new framework is crucial because it provides them with a
conceptual point of purchase at the same time as it narrows the range of
relevant empirical evidence.
Hounshell's framework is defined by two interrelated borrowed con-
cepts. The first is the concept of an 'American system' of manufactur-
ing using single purpose machine tools to make interchangeable parts
with semi-skilled workers and no hand fitters. 8 As Hounshe1l 9 demon-
strates, the American system in this sense largely originated in the
1914-16 work of Roe and represents a kind of ex-post historiographic
construction of nineteenth-century American manufacturing practice.
Hounshell's second concept of mass production denotes the twentieth-
century combination of the American system plus MALs to deliver
large quantities of low cost product. As Hounshell lO recognises, this
concept can be traced back to Cameron's article on mass production
which appeared (under Henry Ford's name) in the 1925 Britannica. l1
This defines mass production variously: in terms of results as the provi-
sion of 'great quantities of a single standardised product'; in terms of
precision with absence of fitters; in terms of the application of the
mechanisation principles of 'power, accuracy, economy, system, conti-
nuity and speed'; and in terms of techniques such as sequential pro-
cessing, the transfer of work to the worker and the use of instruments
such as the MAL.
This concept of mass production is problematic because it offers not
analysis but description which jumbles together results, principles and
techniques. As an historical organising concept, mass production in this
sense is dangerous because it offers a covert generalisation based on
Ford's 1920s manufacturing practice which projects specific and
68 BUSINESS HISTORY

contingent features of Ford's practice as universal and necessary charac-

teristics of a coherent system. When the concept is applied to other
times and places, the confusions and difficulties multiply. Hounshell
accommodates General Motors and the annual model change in the
1920s under the heading of 'flexible mass production'Y According to
Ohno 13 and Monden,14 Toyota after 1950 simultaneously combined
variety with low volume and breaks with the principle of moving the
work to the stationary worker. Should Toyota then be accommodated
under the heading of 'really different mass production'? These problems
are not registered by Hounshell, who does not evaluate the concepts of
the American system and mass production but merely uses them as the
basis for a grid reading of firms like Singer, McCormick and Ford.
Hounshell offers a kind of 'tick the box' history where the historian's
task is to trace the development of the American system across different
sites by registering the presence or absence of particular technological
elements in specific enterprises. His main purpose is to establish the
nineteenth-century absence of some essential features of the American
system outside the subsidised armoury sector; Hounshell demonstrates
that, from 1850 to 1880, key technical elements of the system such as
interchangeability and specialised machine tools were absent at
McCormick and Singer.IS Against this background Highland Park rep-
resents the twentieth-century presence of the old armoury elements plus
the addition ofFord's 'own production techniques'. 16 At Highland Park.
the American system of interchangeability without fitters was a reality 17
and Ford's practice also incorporated the techniques of pressing devel-
oped in the bicycle industry. 18 Ford's key achievement was to add a new
assembly technique 19 through 'the adoption of the revolutionary assem-
bly line in 1913'.20
The interplay between the technical and the social is a minor theme in
Hounshell's book which presents Ford's $5 day in 1914 as a necessary
investment in labour compliance and 'the final element in the develop-
ment of mass production at Ford' .21 This same interplay is developed as
the major theme in Lewchuk's book. His point of departure is lived
experience and hostility to economics. Lewchuk is a former line worker
at Ford of Canada, and his book is a kind of intellectualisation of the
line worker's experience. Intellectually, he is hostile to the orthodox
economic account of production as factor combination which invokes
differences in factor cost to explain choice of technology and variable
productive outcomes. Against this Lewchuk emphasises conflict over
the 'effort bargain' between employers and workers. This is mediated by
(productive) techniques such as the MAL and by social institutions.
especially the regime of labour control. These instruments and insti-
FORD AT HIGHLAND PARK 69
tutions jointly influence the terms and conditions on which 'labour time'
is converted into 'labour effort'.
Within this framework, Lewchuk's main purpose is to explain the
decline of the British car industry (or, more precisely, of the British-
owned car firms) in the 1960s and 1970s. Lewchuk claims that these
firms failed to realise the productive potential of the MAL because their
managers never adapted the British regime of indirect labour control to
the requirements of the imported American technology;22 the 'British
system of mass production,23 combines Ford's technology with dele-
gated control and piece rate incentives. 24 Ford's original success with
the line at Highland Park is presented as a kind of inverted image of
later British failure. The MAL was a technique which allowed manage-
ment to exploit American labour's weakness and loss of control over the
effort bargain.25 As such, the MAL was 'a new mechanism for control-
ling effort norms,26 which could not otherwise be raised by foremen,
even in a non-unionised factory like Highland Park whose immigrant
workforce operated in a city of open shops.27
Self contradiction is the problem with this history because Lewchuk's
own discourse does not escape from the unreal assumptions of the
microeconomics which he criticises. Generally, Lewchuk operates in a
production-centred, supply-side problematic where demand-side expla-
nations for the success and failure of specific firms and industries are not
considered. Thus, British problems about the level and composition of
demand in the 1960s and 1970s are relegated to a footnote in Lewchuk's
text. 28 As for the particular significance which Lewchuk attaches to the
line as a form of labour discipline, this argument rests on the assumed
general absence of market constraints so that any firm may speed up the
line and reduce labour content per unit. Logically, any slowdown in
response to market constraints would have directly opposite effects on
labour. In this respect, microeconomics is more realistic because only
those firms operating under conditions of perfect competition can dis-
pose of an unlimited output at existing prices.
If these conceptual weaknesses have not been registered, that is partly
because the Hounshell and Lewchuk accounts are coherent and
mutually supportive. Their intellectual division of labour is such that, in
the history of technology and labour control, the one authority plays the
leading role and the other the supporter. Thus Lewchuk 29 respectfully
accepts Hounshell's history of American technology and presents Ford
by 1914 as 'a mass employer of unskilled workers using the American
system' which had gone a stage beyond by 'linking tasks together into
uninterrupted and in some cases, mechanically paced lines of work,.3o
Hounshell is much less labour and conflict centred but he attaches a
70 BUSINESS HISTORY

similar significance to machine pacing which is presented as Ford's

substitute for, and alternative to, Taylorism. 31 Thus, 'with the installa-
tion of the assembly line and the extension of its dynamism to all the
phases of factory operations the Ford production engineers wrought
true mass production'. 32
If the conceptual basis for this kind of claim is dubious, its empirical
support is equally problematic, as we can demonstrate by briefly and in
turn considering Hounshell and Lewchuk's treatment of two key issues.
The first issue is their productivity evidence, which is striking but
incomplete; this raises the question of whether the MAL innovation will
bear the weight explicitly assigned to it in Hounshell and Lewchuk's
account. The second issue is their failure to define the relation of
assembly to the earlier processes; this raises the question of whether the
processes of assembly and sub-assembly (where the MAL was applied)
will bear the weight implicitly assigned to them in Hounshell and
Lewchuk's account.
The historiographic presumption that the MAL was Ford's decisive
innovation at Highland Park can be traced back to one major source;
Arnold and Faurote's33 book-length, process by process, description of
Model T production based on factory visits in 1914 just after the MAL
had been developed for a variety of assembly tasks. Without analysing
the overwhelming detail of their account or reflecting on the (acciden-
tal) timing of the visit, generations of historians have focused on the
most immediately intelligible item in the book and reproduced Arnold
and Faurote's estimates of the (direct) labour savings attributable to the
introduction of the line in particular assembly tasks. As Lewchuk 34
observes, Arnold and Faurote provide 'the most widely quoted estimate
of the MAL's impact on labour productivity'. Historians seldom omit
Arnold and Faurote's35 claim that, in chassis assembly, the introduction
of the MAL between September 1913 and April 1914 reduced the direct
labour requirement from 12 hours 28 minutes to 93 minutes.
Hounshell takes the argument no further than citing various Arnold
and Faurote estimates of labour time savings directly attributable to the
line?6 He notes the richness of the Ford company's records 3? but does
not exploit them, except to present the basic explicandum in a table 3il
which summarises the phenomenal increase in Model T sales and re-
duction of prices between 1909 and 1916. The causes of this achieve-
ment are never analysed, and Hounshell equally does not consider the
determinants of cost reduction such as the composition of internally
controlled costs or the trend of bought-in purchases. On these issues,
Hounshell is inhibited by the anti-Humean logic of his methods; grid
reading can be used to register the presence or absence of technical
FORD AT HIGHLAND PARK 71
elements at specific sites. It cannot be used to explore empirical re-
lations of succession and attribution involving cause and effect.
Lewchuk takes the argument further by attempting to calculate pro-
ductivity in terms of cars per man and labour hours per car.
Unfortunately, the results are ambiguous because each of his two
productivity series covers only part of the period or part of the labour
input. Lewchuk's39 first productivity series on cars per man year covers
the years 1906 to 1913 and is based on a simple division of output by
numbers employed, without correction for variation in purchase con-
tent. In interpreting this series, Lewchuk makes much of an 'effort
crisis,40 which shows up as a pause in productivity growth in 1911 and
1912. This was 'perhaps,41 the stimulus to the innovation of the line; or,
more assertively, 'it was the MAL which resolved these problems by
introducing a new mechanism for controlling effort norms'.42 This in-
terpretation is ingenious but insecure because it could only be vindi-
cated by a longer, continuous series which showed a large (or larger)
jump in productivity around 1913 to 1914 that could be definitely
attributed to the line.
The latter period after 1913 is covered, in a discontinuous way, by
Lewchuk's43 second productivity series on labour hours per car. This is
based on the Model T cost books, which list a labour cost for the chassis
and body of the Ford Touring car. Lewchuk's calculation of labour
hours is obtained by dividing an hourly rate into cost book totals for
labour costs and making an adjustment for changes in purchase content.
Because the cost books begin in December 1913, it is again not possible
to construct a continuous series which would allow a comparison of
early and late productivity gains. More fundamentally the cost book
category of 'labour' does not include all the labour in the car because it
was part of a standard costing calculation whereby Ford listed (direct)
'labour' and 'materials' costs at each process stage and then applied a
multiplicand to absorb the 'overhead' cost which included all the in-
direct labour. The omission of indirect labour is a major problem,
because the Ford company's multiplicand of 1.3 or 1.5 implies that there
was more indirect than direct labour in the T. This must raise questions
about the way in which, since Arnold and Faurote, the MAL has been
generally presented as a way of saving direct labour. If Lewchuk
unconsciously reproduces the traditional preoccupation with direct
labour, this is a symptomatic error which reflects the abstraction of
economic discussion of productivity and the concrete limits of a line
worker's consciousness. Throughout Lewchuk's book it is assumed that
productivity is determined by the 'effort' of the direct worker; Lewchuk
never recognises that, in manufacturing operations, there are usually
72 BUSINESS HISTORY

more indirects than directs, and improving the ratio of indirects to

directs is usually an important way of raising productivity.
If Hounshell and Lewchuk's treatment of empirical evidence on
productivity is unsatisfactory, our misgivings about the privilege of the
line are reinforced by the absence of any systematic account of the
factory. Neither Hounshell nor Lewchuk provide a coherent description
of the span of the processes at Highland Park, the factory layout or the
methods of materials handling. These central issues are dealt with as
peripherals. Thus Hounshell, for example, notices that Highland Park
included a foundry which seems to have attracted his attention mainly
because it includes a pioneering application of the conveyor. 44
Hounshell also mentions a monorail system for materials handling
which he casually (and misleadingly) claims delivered goods 'throughout
the factory'. 45
Neither Hounshell nor Lewchuk can avoid the Highland Park
machine shop which cut as well as assembled the metal which went into
the Model T engine, gearbox and axles. This machine shop attracted
almost as much attention, before 1914, as the MAL did afterwards;
Hounshell 46 rightly emphasises the importance of Colvin's 1913-15
account of Ford's machine shop practice47 as a necessary supplement to
that of Arnold and Faurote. Both Hounshell and Lewchuk recognise
that the machine shop was a major site of innovation using techniques
which were independent of the MAL. Hounshell 48 emphasises the
significance of machines in order of use with workpiece transfer by
gravity slides and roller beds. Lewchuk 49 observes that the single pur-
pose machine tools were fitted with increasingly sophisticated fixtures to
avoid worker idle time. The two authors differ mainly about the chron-
ology of machine shop innovation: Hounshell 5o argues that gravity slides
were part of a cluster of 1913 to 1914 innovations; while Lewchuk
presents machines in order of use as an earlier discovery of 1910. 51 But
neither author assigns a productive or economic weight to machine shop
innovation or explains the relation between this innovation and the
MAL. Hounshell observes cryptically that gravity slides were a source
of 'equally surprising gains' ,52 while failure to give due weight to the
machine shop is built into the organisation of Lewchuk's chapter on
Highland Park: the machine shop is treated in an add-on section at the
end of the chapter-"3 after Lewchuk has made his misleading claims
about productivity in the section on the MAL.
FORD AT HIGHLAND PARK 73
TABLE 1
PHYSICAL PRODUCTIVITY SERIES A, 1909-16

Number of Cars Cars per Labour Hours

Employees Shipped Man Year per Car
(i) (ii) (iii) (iv)
1909 1,655 13,941 8.4 357
1910 2,773 20,739 7.5 400
1911 3,976 53,800 13.5 222
1912 6,876 82,500 12.0 250
1913 14,166 199,100 14.1 216
1914 12,880 240,700 18.7 127
1915 18,892 368,599 19.5 123
1916 32,702 585,400 17.9 134
Sources and Notes: Column i: Number of employees taken from A. Nevins, Ford: The
Times, the Man, the Company (New York, 1954), p. 648.
Column ii: Number of cars shipped from Highland Park in built up or kit form for
assembly in branches. The total number of cars shipped is taken from the company's
shipping ledgers in Ford Archive, Accession 463, Box 1.
Column iii: Cars per man year totals are calculated by dividing total number of cars
shipped by the number of employees at Highland Park and in branch factories which
assembled kits from Highland Park. The total number of American employees is taken
from Nevins, Ford, p. 648.
Column iv: Labour hours per car are calculated on the assumption of 3,000 man hours up
to and including 1913, when factory workers worked six ten-hour days in each of 50 weeks.
From early 1914, after the introduction of the eight-hour day, we assume 2,400 hours
when factory workers worked six eight-hour days for each of 50 weeks. This last assump-
tion improves the 1914-16 performance slightly because branch assembly operations
worked a nine-hour day after 1914.

II
In this second section we displace myth. The aim here is to rework the
relevant evidence on productivity and span of processes so as to con-
struct a refutation of the myth of the line - a myth which has already
been weakened by deconstruction. Negative work of this kind satisfies
the protocols of positivist testing but it cannot reinstate history and
restore the historical quality of things. The third section takes up the
more positive task of recovering the historical relation between activi-
ties and actions through a reconceptualisation of Ford's manufacturing
activity.
On productivity, the relevant evidence is company-wide productivity,
and the first requirement is a continuous series on physical productivity
in terms of cars per man year and available labour hours per car. Table 1
presents basic calculations based on the Nevins54 employment series and
the best available output series of cars shipped from Highland Park (in
built-up and kit form). Tables 2 and 3 register two complications: the
purchase content of each Model T fell between 1909 and 1916, while
74 BUSINESS HISTORY

TABLE 2
PHYSICAL PRODUCTIVITY, SERIES B, 1909-13

Number of Cars Cars per Labour Hours

Employees Shipped Man Year per Car
(i) (ii) (iii) (iv)
1909 2,190 13,941 6.4 469
1910 3,572 20,739 5.8 517
1911 4,100 53,800 13.1 229
1912 7,042 82,500 11.7 256
1913 16,000 199,100 12.4 242
Sources: As Table 1, except for Column i: Number of employees taken from Ford Times,
June 1913.

TABLE 3
PHYSICAL PRODUCTIVITY AT CONSTANT PURCHASE CONTENT, 1909-16

Cars per Man Purchases as a Cars per Man

Year Percentage of (at 1909
Gross Output content)
(i) (ii) (iii)
1909 8.4 68.2 8.4
1910 7.5 62.8 8.9
1911 13.5 61.1 16.5
1912 12.0 61.1 14.6
1913 13.9 60.0 17.5
1914 18.8 58.0 24.8
1915 19.5 52.4 29.3
1916 17.9 55.5 25.1
Sources and Notes: Column i; From Table 1, Column iii.
Column ii. Material purchases are taken from Ford Archive. Accession 157; and gross
output from Ford Archive, Accession 96, Boxes 8 and 10.
Column iii. Cars per man hour at 1909 content are calculated by dividing the purchase
content of 1909 gross output by the purchase content of gross output in later years. This
factor is then applied to the cars per man total in Column: for each year after 1909.

various sources gave slightly different employment totals for the early
years. The three tables all tell variants on the same basic story. Table 3
shows a phenomenal threefold increase in cars per man (at constant
content levels), while Table 1 shows a broadly similar reduction in
labour hours. The tables all show the same uneven pattern of producti-
vity increase with two remarkable productivity leaps in 1910-11 and
1913-14, separated by a short two to three year plateau; Table 1, for
example, shows a cars per man average of 8.0 in 1909 and 1910 which
leaps to 13.5 in 1911, runs on a plateau to 14.1 in 1913 before leaping to
18.7 in 1914.
For the myth of the line, the immediate problem is the first producti-
FORD AT HIGHLAND PARK 75

vity leap of 1910-11, which coincides with full production at Highland

Park and predates the moving assembly line (MAL) by more than two
years because (as all agree) the MAL was first used in flywheel magneto
production in the spring of 1913. The consequences of this first leap for
the myth of the line emerge most graphically in the Table 1 series on
labour hours per car. This shows that Henry Ford's pre-war achieve-
ment was to take two-thirds of the labour content out of the Model T at
the same time as he was building more of each car; altogether Ford took
out around 250 labour hours between the 1909-10 plateau of 350-400
labour hours per car and the 1914-16 plateau of 125-135 hours. But of
the overall reduction of 250 hours, no less than 150 hours was taken out
in the first leap of productivity between 1910 and 1911. It is also worth
pointing out that the Table 2 productivity series, based on slightly higher
employment totals in the early years, shows a substantially higher
labour hour input of over 500 hours per car in 1910 which would make
the leap of productivity between 1910 and 1911 all the more dramatic. In
any case, the myth of the line is surely refuted if two-thirds of Ford's
total labour hour reduction was achieved at least two years before the
MAL.
Furthermore, it is not clear that the second productivity leap of
1913-14 can be securely attributed to the MAL which (all agree) was
applied to a variety of sub-assembly tasks as well as final chassis assem-
bly between spring 1913 and summer 1914. Causal attribution is imposs-
ible because of the absence of any agreed chronology for other major
innovations, like gravity slides, which mayor may not have been
introduced at the same time as the MAL. Until other Goint) causes are
ruled out, it is impossible to attribute (all) the aggregate effect to the
(one) identified cause. Our epistemological scepticism on this point is
reinforced by the observation that Arnold and Faurote's estimates of
the process by process labour savings attributable to the MAL appear to
add up to something substantially less than the 100 labour hours saved in
the leap between 1913 and 1914 according to Table 1. In the most
famous example of chassis assembly, Arnold and Faurote55 estimated
the unit labour saving at approximately 11 hours. The savings were
much smaller in the many other sub-assembly tasks; in front axle
assembly, for example, Arnold and Faurote56 estimated the MAL saved
only 1112 hours.
If the physical causes of Ford's productivity increases have never been
identified, the financial effects are generally misunderstood. This is
because much discussion of the MAL assumes that it was the reduction
in internally controlled labour hours which allowed Ford to cut the price
of the Model T. It is essential to correct this misunderstanding if we wish
76 BUSINESS HISTORY

TABLE 4
PRICE CUTS, PAYROLL AND MATERIAL COSTS, \909-16

List Price Sales Payroll Material

of Touring Revenue per Cost per Cost per
Model T Model T Model T ModelT
$ $ $ $
(i) (ii) (iii) (iv)
1909 850 865 64 590
1910 950 875 100 544
1911 780 570 65 348
1912 690 660 6S 407
1913 600 532 61 319
1914 550 637 65 371
1915 440 487 64 255
1916 360 472 70 262
Sources: Column i: Price of the Touring T in dollars, This is the list price of the Touring
(rag top, four seat) Model T which was by far the most popular model variant in the early
years. In most years there were autumn price cuts and we have taken the usually higher
price which prevailed for most of the year. List prices are taken from Ford Archive,
Accession 157.
Columns ii and iii: Sales revenue per T is calculated by dividing total sales revenue by
Model production volume. Sales revenue is taken from Ford archive, Accession 463, Box
1 and payroll from Ford archive, Accession 843.
Column iv: Materials purchases are taken from Ford archives, Accession 157; and gross
output from Ford archive, Accession 96, Box 8 and 10.

to understand Ford's achievement and the importance of manufacturing

span at Highland Park. Table 4 lays out the relevant evidence and shows
how the expansion of Model T sales was stimulated by Ford's policy of
continuous deep price cuts; the price of a fully equipped Touring Twas
reduced from $950 to $360 between 1909 and 1916. However, the price
cuts were not, and could not have been, directly covered by Ford's
success in taking labour hours and costs out of the car. Table 4 shows
that there was never more than $100 of internal labour costs (direct,
indirect and managerial) in the Model T; before and after the $5 day,
Ford usually managed with payroll costs per car in the region of $65. It
was the costs of bought-in materials and components which initially
blocked the possibility of price cuts. Table 4 also shows that, when the T
cost $950 in 1909-10, the bought-in materials and components cost $550.
The subsequent trend of the bought-in materials bill is equally interest-
ing because it shows two definite steps down - between 1910 and 1911
and then between 1914 and 1915 - which roughly coincide with the leaps
in productivity. It is equally significant that, over the whole period
1909-16, the overall $328 reduction in materials and components costs
explains 55 per cent of the total $590 cut in the list price (Table 4).
The myth of the (assembly) line can finally be displaced by showing
FORD AT HIGHLAND PARK 77
that the crucial reduction in materials and components cost was driven
by Ford's deliberate extension of the span of the company's manufactur-
ing activity which was designed to exploit a conversion efficiency advan-
tage that pre-dated the MAL. Like most other companies in the infant
American auto industry, Ford was an assembler which in 1909 bought in
most of the car from components suppliers who provided chassis frames,
wheels, springs, finished bodies, soft tops and some of the engines. The
parts bill of $550 included several tiers of embodied supplier labour and
conversion profit. By this date, Ford already had a substantial conver-
sion efficiency advantage because the company could realise a dollar of
manufacturing value added at substantially less labour cost than its
suppliers. From 1909 to 1916 labour's share of value added at Ford is
consistently low at 27-32 per cent; much less than the 50-55 per cent
ratio then current in Michigan or US manufacturing as a whole. 57 Given
the composition of Model T costs and Ford's relative conversion
efficiency, the prize of cost reduction could only be gained if Ford built
and operated a plant with the manufacturing capacity to substitute in-
house production for bought-in components. Logically, any subsequent
gain in assembly efficiency was secondary.
The all-new factory at Highland Park was designed to meet the
primary manufacturing requirement. Production of Model T com-
ponents began when the first phase of the building was completed in the
spring of 1909. 58 The new buildings, which were extended to fill the 60-
acre site over the next seven years, represented an explicit commitment
to build more of each car in-house. Ford Times 59 claimed that 'in this
plant everything from screws to upholstery that enters into Ford cars
will be manufactured'. The span of manufacturing operations was deci-
sively extended by the construction of a large, state of the art machine
shop which cut, pressed and assembled metal for the four high value
added sub-assemblies (engine, gearbox, front and rear axle). The Model
T could not be sold cheaply as long as these key components were
expensive to manufacture or buy in. By the summer of 190960 the
company was erecting the steelwork for a 117,000 square foot, single
storey shop, and by winter 191261 Ford was adding an 187,000 square
foot extension to this machine shop. By April 191562 the single storey
machine shop covered ten acres and contained 5,500 machine tools.
From autumn 191063 the machine shop was fed by an in-house foundry
which supplied engine blocks and cylinder head castings. Other early
developments included a heat treatment facility which was necessary
because Ford's 'strength with lightness' philosophy required the heat
treatment of every metal component.
The effects of Ford's extension of the manufacturing span are
78 BUSINESS HISTORY

frustratingly difficult to measure, but the financial ratio of purchases to

sales or gross output does provide a financial proxy. As Table 3 shows,
this ratio falls from 68 per cent in 1909 to a low of 52 per cent in 1915;
the implication is that Ford's manufacturing accounted for less than one-
third of the value of the car in 1909 and nearly half by 1915. This
financial measure understates the physical change because Ford was
brilliantly successful in reducing the costs of components whose pro-
duction was brought in-house. In most cases, in-house production
quickly halved the cost of the component and, significantly, Ford's
suppliers were generally unable to show similar reductions on those
components which continued to be out-sourced. These points can be
illustrated by some random comparisons drawn from the April 1914 and
April 1916 Model T cost books which list bought-in prices and estimate
in-house costs. 64 Amongst other components, top iron, (soft) top, hood
and coil were brought in-house during this period and their 1916 cost of
production was estimated at 54 per cent of their 1914 purchase price. By
way of contrast, the wheel set, carburettor and springs, which were still
out-sourced in 1916, show a purchase reduction of just under 20 per cent
over two years. The cost reducing leverage was strongest in the case of
the high value added sub-assemblies, especially the engine and gearbox,
which by 1912-13 were being manufactured in Ford's machine shop. By
1914, Ford was casting, heat treating, cutting and assembling a complete
engine and gearbox for a total cost of $61 in direct labour, materials and
overheads. This was a remarkable achievement when in 1914 Ford was
paying almost as much to its body supplier, Briggs, which charged $55
for a bare open body minus soft top and screens.

III
It is unconstructive to refute myths about innovation through positivistic
measurements unless this work is complemented by a more positive
effort to understand the historical place of particular innovations in a
sequence of actions and activities. But, if it is easy to refute myth, it is
more difficult to reconstitute historical knowledge: the latter exercise
requires a reconceptualisation of specific activities. Because generalisa-
tions like mass production provide no secure basis for reconceptualisa-
tion, we begin this section by proposing our own concept of Repetitive
Manufacturing (ReM).
All manufacturing involves the physical conversion of raw materials
and components into finished products. In a market economy the physi-
cal conversion work also doubles as the financial activity of adding
value. ReM is manufacturing which involves repetition of the basic
FORD AT HIGHLAND PARK 79

process operations so that successive workpieces travel along the same

work path. ReM may involve, but does not necessarily require, stan-
dardised products. Families of related products with different dimen-
sions and specifications can be processed provided process equipment
can be rapidly changed over to handle differently sized workpieces.
Equally, ReM may involve, but does not require, high-volume, short-
cycle production. ReM has been applied to the production of railway
rolling stock and supertankers where individual finished units are pro-
duced in cycles measured in weeks rather than minutes. ReM is univer-
sally used in the production of complex consumer goods for volume
markets whether the goods are brown, white or automotive. If the
classic twentieth-century exponents of ReM are car firms like Ford and
Toyota, the nineteenth-century pre-history of ReM runs back from
companies like Singer, through the armouries, and all the way to
BruneI's system for manufacturing naval blocks during the Napoleonic
Wars.
Successful ReM involves a systematic relation between results, goals,
instruments and techniques of productive intervention. As in all manu-
facturing, the result of effective productive intervention is cost re-
duction. Again, as in all manufacturing, the basic goal is to reduce costs
by taking labour out because wages and salaries typically account for
more than half the cost incurred in adding value. This point is misunder-
stood in orthodox economics, which tries to explain the effects of adding
capital in a production function framework on the false assumption that
a given quantum of labour and capital input can only generate a fixed
quantum of output. Effective ReM uses specific instruments and tech-
niques which negate that prediction. The standardisation of the work-
path creates possibilities of increasing throughput and reducing costs by
taking labour out without large capital expenditures. The general instru-
ment is improved flow within and between the different stages of the
conversion process. Smooth, continuous flow is the basis for ReM cost
reduction because flow improves the utilisation of direct labour and
eliminates unnecessary indirect employees engaged in materials hand-
ling. Techniques for improving flow include low-cost modifications to
the existing process equipment, changes in machine layout and improve-
ments in materials handling. The writings of ReM practitioners like
Woollard 65 and Ohno66 - the production gurus of BMC and Toyota
after World War II - provide a guide to some of these techniques. They
should, however, be treated with some caution because the gurus tend
to claim transcendent, universal privilege for their techniques of auto-
matic transfer (Woollard) and cellular manufacturing (Ohno) which
represent specific solutions to particular problems.
80 BUSINESS HISTORY

The history of ReM has yet to be written. In that history Ford at

Highland Park will figure as a brilliantly successful pioneer. The com-
pany's success can be measured in terms of the low level of stocks and
the favourable ratio of indirect to direct workers because the realisation
of flow reduces the accumulation of work-in-progress within and be-
tween processes and partly in consequence eliminates the employment
of indirect materials handlers. By 1915, Highland Park was running with
stock cover of 3-5 days for major components like chassis frames and
engines;67 as early as 1913, buffer stocks between individual production
departments were down to a few hours, with zero stocks inside the
departments. 68 This stands comparison with current Japanese achieve-
ments. When we visited Toyota in 1988, panels were being made in
batches equal to one and half days requirement while buffer stocks
between assembly and satellite factories of Toyota and its suppliers were
down to a few hours. Using the ratio of direct to indirect labour hours,
Ford's performance is actually superior to that of its Japanese suc-
cessors. After 1913 the Ford company's ratio of indirect to direct labour
hours was 2:5:1, whereas the comparable ratio in Japanese manufactur-
ing is now 4: 1 and no less than 8: 1 in modem American manufacturing.
If the principles of flow ReM are simple their application is extraordi-
narily difficult. This was particularly so in the production of the Model
T, where Ford faced particular problems about weight, complication
and rapidly shortening cycle times. The Model T had been designed for
low-cost manufacture as a lightweight car; each Touring T weighed
1,200 pounds which is significantly less than the 1,400 pound weight of
Europe's lightest current production car, the Citroen AX.69 But, when
the Ford company was producing 1,000 cars a day in 1914, it was
shipping 625 tons of metal out of the factory each day and that weight
had to be handled many times between production stages. At the same
time, each Model T was a complex product which contained around
10,000 separately numbered T parts. Around half these parts were
manufactured in-house at Highland Park in hundreds of thousands of
separate process operations. To take just one low technology example -
a pressed steel cover at the bottom of the engine carried the crankshaft
and doubled up as the oil sump - no less than 78 separate operations
were required to tum steel sheet metal into a finished cover. 70 Ford also
had to cope with the rapid decline in the cycle times as volume increased
from 14,000 cars in 1909 to 585,000 in 1916. When the company was
making 1,000 cars a day in the summer of 1914 and most of the factory
was working two shifts of eight hours, the cycle time for the completion
of finished cars and sub-assemblies like the engines and axles was
already under one minute.
FORD AT HIGHLAND PARK 81
The achievement at Highland Park was epoch making because flow
had never before been realised in the high-volume production of a
complex engineering product which was heavy and awkwardly shaped.
The nineteenth-century originators of flow ReM had processed liquid or
sheet products like flour or textiles. Mid-century engineering appli-
cations of ReM only covered a short span of processes, as in naval
blocks, or small and easily handled components, as in clocks, or easily
worked materials such as wooden rifle stocks. ReM products of the late
nineteenth century were either, like sewing machines, much simpler
than the Model T, or, like railway locomotives, produced over much
longer cycle times. Against this conceptual and empirical background
we can more precisely pose the question about how the MAL contribu-
ted to flow both as a transfer device and as an instrument for reorganis-
ing production at the point of assembly. This is a positive historical
question which can take us beyond the negative refutation of myth.
Any discussion of the significance of the line as transfer device must
start from the more general distinction between linear and cellular
principles of factory organisation. In all manufacturing operations Ford
favoured the linear principle of moving the workpiece past the station-
ary worker. 71 Thus, in the Highland Park machine shop of 1914, as
described by Arnold and Faurote, the basic organisation for production
of each machined component was one man (or crew) per machine or
fixed workstation, with workpiece transfer via gravity slides and rollers
in straight lines between machines in order of use.72 Occasional face-to-
face positioning of two machines with one minder was Ford's only
concession to the kind of cellular manufacturing which Toyota later
used mainly in sub-assembly production: after 1950 Ohno systematically
exploited the individual worker's ability to walk and manually transfer
work between the stations of a U-shaped cell. The advantages of cellular
manufacturing were discovered in Japanese firms whose non-Fordist
task was to produce a varied output on long cycles at low volume. 73
Toyota subsequently added the concept of variably sized work teams in
interconnected cells, and met the requirement for increasing volume by
replicating the whole cell rather than duplicating the bottleneck
machine as Ford did. 74
Within Ford's linear organisation at Highland Park, the moving line
figures as one instrument for realising the movement of the workpiece
past the stationary worker: all Ford's moving lines had a workpiece
transfer function, and by 1915 roughly half of them were simple con-
veyors without any assembly or manufacturing function. But, at
Highland Park, the moving line was only one amongst several equally
important instruments for moving the workpieces past the stationary
82 BUSINESS HISTOR Y

worker: in this respect Highland Park before World War I was different
from the later Rouge factory which Ford developed in the 1920s. At
Rouge, the moving assembly line was used as a universal transfer device
within the factory, so that by 1926 Rouge had no less than 27 miles of
conveyor. 75 By way of contrast, in the spring of 1915, when the moving
line had been applied extensively in final and sub-assembly production,
Highland Park had only 1112 miles of moving line which divided into half
manufacturing conveyor and half pure transfer conveyor. 76 In the
Highland Park machine shop, the main mechanised transfer system was
not a conveyor but a 11/4 mile long overhead monorail, which had been
constructed somewhat earlier.77 By early 1914 this monorail shifted
1,400 tons of work-in-progress each day into, through, and out of the
machine shop to the final assembly building H.78 Strategically sited
overhead craneways also played an increasingly important role in the
delivery of bought-in materials and components to many of the depart-
ments. From the autumn of 1910 a 56 foot wide craneway bisected the
main machine shop79 and narrower 40 foot craneways were placed
between the six-storey buildings (W, X and Y) constructed in 1913-14.
The importance of all these transfer devices, including the MAL, was
diminished because Ford's manufacturing philosophy at Highland Park
was to reduce or eliminate transfer through layout change rather than to
install mechanically powered handling systems. In this respect, again,
Highland Park was a proto-Japanese factory whose techniques were
very different from those of American manufacturing after World War
II, which used mechanised handling systems as an easy way of eliminat-
ing indirects and incidentally fixed layouts around these systems. 80 At
Highland Park, layouts were continually changed so that work-in-
progress travelled shorter, more direct routes and required less hand-
ling. The conversion of Highland Park to a short-travel factory required
sustained attention to three issues: first, the sequence of shops; second,
the positioning of machines in order of use; and third, the development
of simple, direct methods of intra-process transfer. By 1912, Highland
Park's shops (or departments) were laid out on a west to east axis along
which work-in-progress travelled to final assembly.81 Within each shop,
Ford's practice was to arrange machines, regardless of function, accord-
ing to 'sequence of use,.82 By persevering with this principle, Ford was
able to reduce the distance which the engine block travelled between
finishing operations from 4,000 to 334 feet. 83 With machines placed
close together and in order of use it was also possible to introduce
simple direct intra-process transfer devices like gravity slides, roUer
beds and racks of pipe and angle iron.
If the historical answer is that the MAL played a limited role as
FORD AT HIGHLAND PARK 83

transfer device at Highland Park, the remaining question is whether the

MAL played a more important role as an instrument of productive
reorganisation. Any discussion of the MAL and productive reorganisa-
tion must begin by making the general point that the concept of manu-
facturing flow undermines the idea of key or crucial innovation at a
single point: in long span heterogenous manufacturing operations, fac-
tory flow can only be realised through a variety of productive inter-
ventions of different kinds at a multiplicity of points. Thus, at Highland
Park, the realisation of factory flow depended on solving problems in
the metal cutting stage as well as final assembly, and the two stages of
manufacture presented different problems and required different sol-
utions. Within metal cutting, large-scale investment in single-purpose
machine tools was not enough to secure a steady flow of machined
components. Innovative fixture design was also necessary because as
cycle times fell to, and below, one minute, the key limit on machine
throughput and direct labour utilisation was the cycle set-up time which
was wasted on clamping and unclamping the workpiece. By 1915, Ford
was installing multi-fixtures which allowed machines and their minder to
work continuously. Colvin describes the machine shop before these
multi-fixtures were installed and details the many ingenious ways in
which cycle set-up had already been reduced. Ford used quick acting
'one touch' clamps84 and generally favoured drop-in gravity location by
pegs on steel tables. 85 The company also developed 'multiple tooling' so
that many cuts in several planes could be made at each clamping86 and,
where this was not possible, Ford designed fixtures that could finish
many workpieces in a single pass. 87
From a flow perspective, the relative importance of specific inno-
vations at particular process stages, like cutting or final assembly,
depends on the amount of labour saved within the process and the
impact on flow outside the process. By these criteria, the MAL made a
very modest contribution at the final assembly stage. The amount of
labour saved in process could not be large because 'chassis' (final)
assembly before the line was a relatively small department which
employed many fewer workers than a large department like casting in
the foundry. Before the introduction of the MAL, in September 1913,
chassis assembly employed 500 assemblers on two shifts. By way of
contrast, the foundry then employed 1,450 men on three shifts.88 The
final assembly labour requirement was modest before the MAL because
the company had been able to raise assembly productivity by recompos-
ing manual tasks through simple division of labour without moving
lines. Arnold and Faurote's example of piston and connecting rod
assembly demonstrates the potential of this kind of reorganisation.
84 BUSINESS HISTORY

Here, the number of workers was halved from 28 to 14, and throughput
was increased simply by the division of the task into three separate
operations with no change in tools except for the installation of a gravity
slide. 89 A small department can, of course, still be a serious bottleneck
constraint on throughput factory flow; the paint shop, for example, is
often a bottleneck in modern car assembly plants. But any final assem-
bly bottleneck was removed before the introduction of the MAL when,
in the autumn of 1912, Ford announced a new policy of decentralising
final assembly to branch plants in 'all the principal cities,.90 Thus, the
consumer could avoid delivery charges and Highland Park could con-
centrate on the vital cost-reducing task of increased component
production. 91
In sub-assembly production, the effects of the MAL were more
important, because assembly innovation here allowed Ford to realise
the full benefits of earlier innovation in the cutting stage of the machine
shop which had freed up output that went forward into bottleneck sub-
assembly departments. This restriction of flow in sub-assembly pro-
duction helps to explain why the MAL was first developed for sub-
assembly work, like the flywheel magneto. Front axle production, as
described by Arnold and Faurote, provides a good illustration of the
general pre-MAL problem with sub-assembly work. The built up front
axle was a major sub-assembly because it included the front suspension
in the form of a transverse leaf spring as well as the outboard steering
gear. Although Ford just about kept up with growing demand by
reorganisation of sub-assembly work there was a large imbalance be-
tween the potential throughput of front axle machining and the actual
output of the front axle assemblies, and this imbalance compromised
labour utilisation of the cutting workforce which was three times larger
than the assembly workforce. After pre-line improvements, the front
axle assembly capacity in January 1913 was 450 axles per shift assembled
by 125 men. The front axle cutting or 'finishing department' employed
350 on machines which were mostly capable of 800 cycles per shift. 92
Significantly, when the MAL was introduced in front axle assembly, the
assembly capacity was raised to 800 per shift so that the full potential of
the cutting stage could be realised. As throughput increased, the MAL's
impact on out of process labour utilisation in cutting was much more
important than the in-process reduction of the front axle assembly crew
from 125 to 44. 93
Within this framework we can finally recover the historical signifi-
cance of Ford's MAL. In orthodox production and operations manage-
ment texts, the line is inserted into a problematic of static optimisation;
these texts present algorithms which model optimal solutions to the
FORD AT HIGHLAND PARK 85
division of assembly tasks between a given number of stations. Ford's
realisation of flow within ReM production was very different because
Ford deliberately created dynamic imbalance between the departments
of a rapidly expanding factory as a way of levering continuous challenge
and response improvement. By 1912 large-scale investment and
ingenious fixture design had freed up the output potential of the
machine shop and created dynamic imbalance because the company
was, as Klann 94 observed, 'making parts a hell of lot faster than we
could put them together on cars'. Ford's response was the MAL. This
reversed the dynamic imbalance because the MAL delivered an assem-
bly capacity which was well beyond Ford's foundry and cutting capabili-
ties: in February 1914, the three MALs at Highland Park assembled
1,212 chassis in one ten-hour shift which implies a two-shift capacity of
more than 2,000 chassis. This was more than three times the immediate
assembly requirement and around twice the throughput capacity of the
machine shop: in June 1914 the maximum two-shift capacity of
Highland Park was 1,000 sets of components a day and by this date 400
chassis were assembled in branch factories. 95 Mythically, the line is
Ford's crucial innovation; historically it is a perfect example of the
dialectical relation between activities and actions.

NOTES

1. H. Nevins and F. Hill, Ford: the Times, the Man, the Company (New York, 1954).
2. D. Hounshell, From the American System to Mass Production (Baltimore, 1984).
3. W. Lewchuk, American Technology and the British Vehicle Industry (Cambridge,
1987).
4. 1. Womack, D.1ones and D. Roos, The Machine that Changed the World (New York,
1990).
5. Hounshell, American System.
6. Lewchuk, American Technology.
7. For example, Hounshell, American System, p. 244; Lewchuk, American Technology,
p.48.
8. Hounshell, American System, p. 227.
9. Ibid., p. 334.
10. Ibid., p. 7.
11. H. Ford, 'Mass Production', Encyclopaedia Britannica, 13th edition, Vol. 30, pp.
821-3 (Chicago, 1926).
12. Ibid., p. 267.
13. T. Ohno. The Toyota Production System (Tokyo, 1978).
14. Y. Monden, Toyota Production System (:"<orcross, Georgia, 1983).
15. Hounshell, American System, p. 328.
16. Ibid., p. 220.
17. Ibid., p. 233.
18. Ibid., pp. 215, 255.
19. Ibid., pp. 244-7.
20. Ibid., p. 10.
21. Ibid., p. 329.
86 BUSINESS HISTORY

22. Lewchuk, American Technology, p. 185.

23. Ibid., p. 5.
24. Ibid., p. 222.
25. Ibid., p. 36.
26. Ibid., p. 61.
27. Ibid., p. 37.
28. Ibid., p. 270; also see K. Williams, J. Williams and D. Thomas, Why are the British
Bad at Manufacturing? (1983).
29. Lewchuk, American Technology, p. 33.
30. Ibid., p. 42.
31. Hounshell, American System, p. 325.
32. Ibid., p. 237.
33. H. Arnold and F. Faurote, Ford's Methods and the Ford Shops (New York. 1915).
34. Lewchuk, American Technology, p. 49.
35. Arnold and Faurote, Ford's Methods, p. 139.
36. For example. Hounshell, American System, pp. 249, 255.
37. Ibid.,p.271.
38. Ibid., p. 224.
39. Lewchuk, American Technology. p. 46. Table 3.5.
40. Ibid., p. 61.
41. Ibid., p. 47.
42. Ibid., p. 6.
43. Ibid., p. 45. Table 3.4.
44. Hounshell, American System. p. 240.
45. Ibid .. p. 38.
46. Ibid., p. 224.
47. American Machinist. various issues. 1913-15.
48. Hounshell, American System, p. 229.
49. Lewchuk, American Technology, p. 59.
SO. Hounshell, American System, pp. 237-8.
51. Lewchuk. American Technology, p. 54.
52. Hounshell, American System. p. 256.
53. Ibid., pp. 52-7.
54. Nevins, Ford.
55. Arnold and Faurotc, Ford's Methods. p. 258,
56. Ibid .. p. 193.
57. Lee. E.S .• Population Redistribution and Economic Growth in the United States,
1870-1950 (Philadelphia. 1957). Tables M6 and M8.
58. Ford Times, April 1909.
59. Ibid .. July 1910.
60. Ibid., Aug. 1909.
61. Ibid., Jan. 1912.
62. Ibid., April 1915.
63. Ibid., Oct. 1910.
64. Ford Archive, Detroit, Accession 125, Boxes 1-20.
65. F.G. Woollard, Principles of Mass and Flow Production (1954).
66. Ohno, Toyota Production, 1978.
67. Arnold and Faurote. Ford's Methods, p. 63.
68. Ford Times. Jan. 1913.
69. Autocar, 18 Aug. 1988.
70. Arnold and Faurote. Ford's Methods. p. 82.
71. Ibid., p. 102.
72. Ibid., pp. 272-3.
73. Ohno. Toyota Production.
74. Monden, Toyota Production.
75. A. Nevins and F. Hill, Ford: Expansion and Challenge (New York. 1957).
FORD AT HIGHLAND PARK 87
76. Ford Times, April 1915.
77. Ibid., April 1912.
78. Ibid., April 1915.
79. Ibid., Nov. 1910.
SO. Anglo American Council on Productivity, Reports, Materials Handling in Industry,
May 1950.
81. Ford Times, Oct. 1912.
82. Iron Age, Dec. 1915, pp. 1276-7. Article written by Bornholt of the Ford Company.
83. Arnold and Faurote, Ford's Methods, p. 38.
84. American Machinist (1915), p. 971.
85. Ibid., p. 1039.
86. Ibid. (1913), p. 843.
87. Ibid., p. 227.
88. Arnold and Faurote, Ford's Methods, pp. 327, 350.
89. Ibid., pp. 105-7.
90. Ford Times, Sept. 1912.
91. Ibid.
92. Arnold and Faurote, Ford's Methods, pp. 193,174-9.
93. Ibid., p. 185.
94. Ford Archive, Detroit, W. Klann, 'Oral Reminiscences', p. 55.
95. Arnold and Faurote, Ford's Methods, p. 142.