Professional Documents
Culture Documents
and Culture
The International Quarterly of the
Society for the History of Technology
Computing Technology (Englewood Cliffs, N.J., 1985); about five pages are
devoted to tabulators in Herman H. Goldstine, The Computer from Pascal to
von Neumann (Princeton, N.J., 1972); similar brief discussions can be seen
in Brian Randell, ed., The Origins of Digital Computers: Selected Papers, 1st, 2d,
and 3d eds. (New York, 1973, 1975, 1982); and Charles J. Bashe et al.,
IBM's Early Computers (Cambridge, Mass., 1986).
4 J a m e s Cortada, A History o f Data Processing in the United States (Westport,
Conn., forthcoming).
Punched Card Machinery in Business and Government 755
History (New York, 1969), and Leon E. Truesdell, The Development of Punch Card
Tabulation in the Bureau of the Census 1890—1940 (Washington, D.C., 1965).
Details about the events surrounding the beginnings of the Hollerith
company can be found in Geoffrey D. Austrian, Herman Hollerith: Forgotten
Giant of Information Processing (New York, 1982).
6 A minor part in tabulating was played by the Burroughs Adding
Berkeley, Giant Brains or Machines That Think (New York, 1949). Perhaps the best available
account of how the punched card systems actually functioned can be found in Bashe et al.,
IBM's Early Computers (n. 3 above).
Punched Card Machinery in Business and Government 757
1989), which discusses at length the punched card era in England and the
international connections of British and American firms.
10 Eric J. Hobsbawm, The Age of Empire 1875—1914 (New York, 1987), p. 62.
11 Hobsbawm, p. 53, offers several English examples; there were 500 Lipton Tea
branch stores in England in 1899, whereas there had been none in 1870.
12 Michael Chatfield, A History of Accounting Thought (Hinsdale, Ill., 1974), p.
125. 13Gary John Previts and Barbara Dubis Merino, A History of Accounting in
America:An Historical Interpretation of the Cultural Significance of Accounting
(New York, 1979),chap. 4.
758 Arthur L. Norberg
statements on a regular basis." To prepare more than perfunctory
profit and loss statements, companies needed substantially more
information about their own operations.
How this came about can be seen in recent elegant studies by Judith
McGaw and Alfred D. Chandler, Jr. McGaw shows how the paper
industry changed its accounting methods over the course of the 19th
century to meet its needs for better knowledge of operations.15 During
the first half of the century, the emphasis was on financial accounting,
used to understand the overall financial condition of the firm. As firms
became larger and more complex after the 1860s, cost accounting was
added to enable better decision making. The desire within the
companies themselves for better information about internal costs was
more significant in this case than the need for public disclosure.
Breaking down costs and income into finer and finer increments
allowed the type of control that produced greater profit. Profitability
and increased sophistication in reporting seemed to go hand in hand.
As elsewhere in the development of American business in the second
half of the 19th century, the railroads showed the way. In The Visible
Hand, Chandler writes that, in meeting "the needs of managing the
first modern business enterprise, managers of large American
rail-roads during the 1850s and 1860s invented nearly all of the basic
techniques of modern accounting."16 Chandler provides evidence for
denominating the railroad as "the first modern business enterprise" by
reviewing the types of data assembled by operating companies. Data
about daily financial transactions were used to compile operating and
financial statements, the balance sheets. These balance sheets,
including capital accounting figures, could then be used to evaluate a
company's performance. This type of reporting was becoming
common-place by the 1870s, and in June 1879 a convention of railroad
commissioners recommended the adoption of a standardized form of
accounts.
In 1875 Albert Fink had published a book on accounting for
railroads, a system he had developed primarily as general superin-
tendent of the Louisville and Nashville in the 1860s.'7 Fink divided
maintenance of the roadway and general superintendence into thirty
accounts, with eleven more accounts for station operating expenses
14 Robert Sobel, The Big Board (New York, 1965), p. 177. See also Previts
the entire form from Albert Fink, Cost of Railroad Transportation, Railroad
Accounts, and Government Regulation of Railroad Tariffs (Louisville, Ky.,
1875).
Punched Card Machinery in Business and Government 759
21 Q u o t e d in ibid., p. 28.
22 Fielding H. Garrison, John Shaw Billings: A Memoir (New York, 1915), p. 342.
23 Tr ue sde ll, The Development of Punch Card Tabulation (n. 5 above), p. 29.
24 Quoted in Garrison, John Shaw Billings (n. 22 above), p. 343.
Punched Card Machinery in Business and Government 761
Herman Hollerith (n. 5 above), pp. 13—16. A slightly different and abbreviated version
can be found in Goldstine, The Computer from Pascal to von Neumann (n. 3 above), pp.
68-70.
28In the use of the machine in the census, the data would be read from the clock dials
and copied onto large computing forms, and the clocks all set back to zero. For
descriptions of the system, see Herman Hollerith, "An Electric Tabulating System. "
School of Mines Quarterly 10 (April 1889): 238—55 (reprinted in Randell, The Origins
of Digital Computers [n. 3 above], pp. 129—39); T. C. Martin, "Counting a Nation by
Electricity," Electrical Engineer 12 (November 11, 1891): 521—30; Austrian, Herman
762 Arthur L. Norberg
The most remarkable feature of this new system was its versatility.
Through the use of relays on the sectors of the sorting box, it was
possible to connect the system so that a section opened only when a
certain set of conditions were met. The system made simple counting
and complex tabulation faster and, by contemporary accounts, more
accurate.29 The range of successes in the U.S. Census, the health
departments of Baltimore and New York, and various agencies in the
states of Massachusetts and New Jersey seemed to assure the future of
the system and the fame of its designer. After final completion of these
various counting operations, however, the machines were re-turned. In
1894, as government contracts ended, Hollerith suddenly found
himself without income. He concluded that only the ongoing nature of
commercial business would keep his company active and growing.30 So
he turned his attention to railroads and insurance companies.
Tabulating business data, Hollerith found, was a far cry from the
counting done in the Census Bureau and local agencies. Nevertheless,
he had already encountered complicated statistical addition in the
agricultural census of 1890. Instead of counting individual items as
units, workers had to add different numbers from successive cards.
This involved taking the data from each farm number of acres, number
of livestock by type, implements, produce and accumulating totals. A
similar challenge existed with certain health statistics. For example, it
might be important to know when people were on sick leave: how many
people were absent one day, two days, one month. Hollerith developed
new machines for tabulating these types of data. He replaced the
"clock-face" counters with several "integrating" devices, which
contained up to eight decimal digits of capacity; each position was
represented by a ten-position wheel with visible numerals31 Toward the
end of September 1896, Hollerith signed a contract with the New York
Central Railroad for processing up to 4 million waybills anticipated in
the coming year. Hollerith decided to form a new company, which
would hold all his patents and would manufacture machines as
well.32 ." This was the Tabulating Machine Company (TMC).
Rodgers, THINK: A Biography of the Watsons and IBM (n. 6 above). In 1924, the
directors of C-T-R authorized a name change to International Business
Machines (IBM).
36 Truesdell, The Development of Punch Card Tabulation (n. 5 above), pp.
120-21.
37 One reliable source of biographical information on Powers can be
company at the time of the merger in 1927. The death date of 1915 found in
some sources seems incorrect.
38In fact, the Tabulating Machine Company took the Census Bureau to court in
firm the upper hand for a time. In 1914 Powers and his principal
associate, William W. Lasker, Sr., perfected the selection box for
changing connections to do another type of calculation, the running
total mechanism,42 and the automatic zero device on a seven-unit
tabulator. In 1915 Powers introduced a printing tabulator.
The Powers printing tabulator was an important machine.43 It listed as
well as tabulated, whereas the available Hollerith tabulator was a
nonlisting machine having a capacity of five adding sections. Now
tabulated totals could be printed with the designations of codes or
groups of numbers directly on the report or record, or one could list in
detail all the items represented on the cards. The tabulator selected and
added mechanically. Steel pins dropped through the card holes into
designated positions in a "selection box," later called a connection box,
making mechanical connections with the adding head directly above
the box. Connection boxes were made to be interchangeable in order to
facilitate different calculations.44 The Powers machine printed
designations and tabulated totals, with or without detail, on paper
strips or record sheets. In addition, it was equipped with one to seven
designating or adding units that operated simultaneously, and it made
five legible carbon copies.45
The competition posed by Powers's new designs played a major role in
Thomas Watson's strategy as he took over the new firm of C-T-R.
During World War I, Watson also hired a group to work on new
products. Some of the arrangements that engineers and mechanics
had made with TMC for R&D were more in the character of a modern
consultancy, and they did not spend all their time on Hollerith
business. When Watson took over the company, he brought
42 B e f o r e the invention of the "running total mechanism," it was
necessary for the operator to stop the machine and press a key in order
to "take " a total that reset the adding units. The new running total
mechanism was operated by two control cards, a "space" card and a "total"
card, which could be sorted in on top of each group of cards for which a
total was required or could be manually inserted in the deck by the
operator. "This increased the output of the machine, particularly where the
groups of cards to be totalled were small and numerous. This enabled the
operator to perform other duties, such as sorting, thereby reducing the
number of personnel necessary to operate a Punched-Card Department. "
("History of Remington Rand Punched-Card Accounting Machines, " n.d.
[internal evidence suggests mid-1930s], Vertical File, Unisys Archives,
Unisys Corp., Detroit.)
43See W. E. Freeman, "Improvements and Advantages Offered by the
Hollerith equipment see Austrian, Herman Hollerith (n. 5 above), chap. 18.
49 G. A. Shattuck and E. B. Kapp, "Summaries of Business Research:
10-12, see p. 12. These tasks were supervised by Walter V Davidson, who
a few years later developed the Unit Inventory Control system used by
market chains such as A&P, Safeway Stores, and First National Stores.
55SKoon, "Hollerith Tabulating Machinery" (n. 51 above).
56 F o r example, see Freeman, "Improvements and Advantages" (n. 43
p. 5.
770 Arthur L. Norberg
67See The American Digest of Business Machines 1924 (Chicago, 1924) and
69For a comparison of the two systems at this time, see Leslie J. Comrie, The
in 1934. The 405 remained the principal machine throughout the next
decade and a half.71 Remington Rand tried to match these
developments with its own machine (also named the 285) in 1937, and
with a Model 3 tabulator that appeared in mid-1940. The major
improvements over the Model 2 were in the accumulating and sector
printing mechanisms, and its use of the ninety-column card; there
were also some minor changes, such as the carriage.72 But this
machine differed little from the Model 2 and continued to be
mechanically driven.
The aim of this brief descriptive section has been to suggest the trend
in machine design, the nature of the competition between IBM and
Remington Rand, the range of machines used in data processing, and
the development of electronic design. Many changes seem to have been
a matter of competitive strategy rather than for new applications.
Sorting speed increased, but calculating speed improved only
marginally, if at all. The greater density cards were a distinct
advantage, especially for large firms. What is most apparent from
contemporary reports is the continuous spread in usage of the
punched card machinery.
Powers was reorganized in 1922, bringing together the Wales Adding
Machine Company and Powers Accounting Machine Company. Powers
was then acquired by James Rand and amalgamated into the new
Remington Rand Company, along with the Dalton Adding Machine
Company, makers of one of the first ten-key adding machines, in
1927.73 Nevertheless, by 1931 IBM was the leading manufacturer of
equipment and cards by a ratio of greater than eight to one; ten years
later IBM produced 16,956 units to Remington Rand's 1,583.74 IBM's
success in the marketplace was already clear.75
71 pp. 5 9 - 6 8 .
72 American Office Machine Research Service 1940 (n. 67 above), July
1940, sec. 1.2.
73"Remington Rand Inc., History," MS, n.d., Unisys Archives.
74"History of the Administration of Regulations of the Office Machine and
42 above), p. 14
77 "Rock Island Goes Modern in Material Accounting" (n. 50 above), pp.
976-77.
78 Information Systems at Aetna Life & Casualty: From Punched Cards to
tory, and the cards were delivered to the inventory file office. These
cards, sent through the printing tabulator, provided totals for each
day's production, changes in stock, and credits to the factory. The
companies followed similar accounting procedures before production
for planning of production runs and at shipment time for control of
inventory and billing.
The automobile industry is almost a mirror image of the rubber
footwear industry. Automobiles and trucks vary in style, contain many
parts, and require shipment to many locations. As with footwear, the
increasing variety of styles, colors, and options began to slow produc-
tion and delay delivery. Throughout the 1920s and 1930s, automobile
companies turned to the use of tabulator machinery to overcome these
delays. Here it is interesting to focus on the ordering procedure
developed by Chrysler Corporation.82 Upon receipt of a dealer's order,
two cards were punched with the essential information sup-plied by
the dealer: routing, region, district, dealer's name, order number, item
number, model, body type, paint, trim, wheels, trans-mission, radio,
heater, and other options. One card went to the equivalent of a
production planning office where a "Daily Master Building Schedule"
was prepared, and one went to the car distribution department where it
was filed according to region and dealer. Multiple copies of the
production card went to the various inventory-control points for parts,
while several copies stayed with the car as it was constructed. When
the car reached the shipping department, one of the last cards
remaining was checked to see that the order was correct. If so, the car
was shipped and the dealer was notified when to expect it. The cards
could then be used for many different types of statistical analysis, such
as trends in purchasing by color, size, and type of car.
Other examples of the use of punched card equipment between 1910
and 1940 could be drawn from the shipbuilding industry, electric light
and power, pharmaceuticals, and banking.83 A number of examples
could also be drawn from the public sector. The East Bay Municipal
Utility District, on the eastern shore of San Francisco Bay,
82R. G. French, "Quick Figures Cut Shipping Delays at Chrysler Plant, "
Machine," The Engineer 131 (May 20, 1921): 532—35. Businesses in the
United Kingdom, such as Dunlop (tires), Lever Brothers (pharmaceuticals
and soap products), Metropolitan-Vickers Electrical Company, and
Humbers Electrical Ltd. (electrical equipment) were all using equipment
supplied by British Tabulating Company (manufactured under Hollerith or
IBM patents). See E. W. Workman, "Cost Accounting by Machinery,"
Engineering and Industrial Management 6 (September 22, 1921): 314—18.
Similar developments were taking place in France at Compagnie des
Chemin de Fer du Nord (locomotives) and Roger Gallet (perfumes) and in
Russia at the Stalina Automobile Plant. See James Connolly, History of
Computing in Europe (IBM World Trade Corporation, n.d.), pp. 20, 35.
84Kimball and Sedgwick, "Tabulating Machines in Customer Accounting" (n.
records within the Social Security office is taken from a film produced by
IBM in 1937 called "Social Security at Baltimore: World's Biggest
Bookkeeping Job, " IBM Archives, Valhalla, N.Y.
90Ibid.
91John J. Corson, "From Tax Return to Wage Record, " 1938 address,
IBM Archives.
778 Arthur L. Norberg
an analysis of the U.S. law "in order to take full advantage of all the
facilities at our command in entering this new American field of
accounting."
Calculations for the Old Age Benefits tax to be paid by both employee
and employer, and the employer's contribution to state unemployment
compensation, could be performed on an IBM Automatic Multiplying
Punch. Cards of exempt persons, treated as a separate group, could be
used to prepare the periodic reports due both the state and federal
government.92 New cards were designed to contain information such as
gross earnings, Social Security deductions, and net earnings. Optional
report forms permitted employers to keep records for their own peculiar
needs as well as for govern-mental requirements. Still, the basic design
for the IBM tabulating system was the same as that developed in the
early 1930s. The advent of Social Security simply provided a greater
incentive for employers to adopt a tabulator system, whether by
acquiring their own machines or by using an IBM service bureau.
Tabulator Twilight
New demands for regular reporting of Social Security data firmly
cemented ties between business and government through the medium
of punched card machinery. To adhere to the system, business needed
the punched card, which had multiple uses in addition to compiling
data for reporting to government agencies such as Internal Revenue
and Social Security. The range of uses to which such machinery had
been put by business in the twenty years before 1936 keeping track of
inventory, sales, and employee hours facilitated the transition to wage
and tax reporting.
During World War II, as companies expanded to meet defense demands,
they had to do so with tighter inventories. Tabulators now became a
necessity, and tabulator companies appear to have had no difficulty
obtaining materials; annual production numbers exceeded prewar
totals by as much as double.93 Naturally, not all of these machines
went to business firms; between June 1942 and the end of 1943
approximately half went to the military. While some of these were
employed in traditional sorts of record keeping such as personnel and
purchasing, others were used for intelligence analysis and for
92Social Security Payrolls Reports Statistics (IBM, 1936), IBM Archives.
93See, e.g., "Controlled Materials Plan using punched cards to control the
flow of critical materials for the War Production Board, " Bulletin 1-0007,
Remington Rand, Tabulating Machines Division, 1943; "History of the
Administration ... ," War Production Board (n. 74 above), p. 43.
Punched Card Machinery in Business and Government 779
calculating artillery tables.94 But in those realms it was obvious that
they were slow and labor-intensive, and the natural consequence was
the pursuit of alternatives such as the Differential Analyzer and
ENIAC.95 The latter was a peculiar hybrid of punched card machine
and electronic circuitry. Yet it clearly suggested the feasibility of a
newer, faster system.
When one thought about it, the tabulator's shortcomings were clear.
First, cards had to be punched, verified, sorted, collated, and tabulated,
each in a separate machine. Second, operators had to carry volumes of
cards from one machine to another to accomplish the various tasks of
sorting, collating, and tabulating. Third, the machines were at their best
in analyzing large volumes of identical work; variations required
manual attention. Nevertheless, tabulators were significant tools in the
establishment of effective procedures for managing records. Most
important, they fostered a better under-standing of problems entailed in
calculation and, beyond that, a vision of how those problems could be
addressed even more effectively. The IBM 603 and ENIAC marked the
beginning of a new epoch. The epoch of punched card machinery was
essentially over by the early 1960s, by which time the virtues of two
generations of electronic digital computers had been demonstrated in
tabulator work as well as other forms of computation. Even so, for
another decade tabulators were kept in service for specialized,
repetitive applications. When electronic computers finally proved
sufficiently versatile for any sort of computation, business and
government had been thoroughly primed for their use through fifty
years' experience with mechanical tabulators. Now, as venerable
artifacts, tabulators may be found in exhibits of the tools that
facilitated the growth of American business and government in the
20th century.
94Kahn, The Codebreakers: The Story of Secret Writing (n. 7 above), pp.
563—64: Goldstine, The Computer from Pascal to von Neumann (n. 3 above),
pp. 129—30.
95 Nancy Stern, From Eniac to Univac: An Appraisal of the Eckert-Mauchly