|
| Early Adding and Calculating Machines |
Antique Adding and Calculating Machines
Scheutz Difference Engine, 1855
A
wide variety of calculating machines were invented beginning in the early 17th
century. Stepped-drum calculating machines solved lengthy multiplication
problems by successive addition and solved lengthy division problems by
successive subtraction. The image to the 
left shows the Pascaline
calculating machine of Blaise Pascal (1623-62), which was not a commercial
success. The first commercially successful machine of this type was the Arithmometer
made by Charles Xavier
Thomas de Colmar (1785-1870). The photograph to the right shows a Thomas Arithmometer
in the Smithsonian Institution's collection. Thomas introduced his Arithmometer in
1820 and spent the remaining half century of his life manufacturing and
improving it. Thomas was French, and in
1856 the Moniteur
reported that "Mr. Thomas has lately made the finishing improvements in the
arithmometer, at which he has been working for upwards of thirty years. It
is already used in many great financial establishments." Thomas's
Arithmometer was a commercial success in Europe after the Paris Exposition of 1867.
At some
point between 1866 and 1870, the Director of the Bureau of Statistics at the U.S.
Treasury Department purchased a Thomas Arithmometer for use in the Bureau. (Alexander
Delmar, "Recollections of the Civil Service: VI," Appletons'
Journal, Sept. 5, 1874; Manufacturer and Builder, August
1870, p. 1) Also, it appears that the St. Louis Mutual Life Insurance Co. had a
Thomas Arithmometer around 1872. (P. Kidwell, "The Adding Machine Fraternity
at St. Louis: Creating a Center of Invention, 1880-1920," IEEE Annals of
the History of Computing, Apr. 2000). In 1874, when the Thomas
Arithmometer was first offered for sale in the US, its price ranged from a low
of $75 for a machine capable of giving a product of 10 figures without quotient
or effacer to a high of $275 for a machine capable of giving a product of 20
figures with quotient and effacer. The notice of the machine's availability in
the US stated that "It is already used by the leading insurance companies
and mathematicians of England and the United States, and by many of the bankers
of London. For any lengthy calculations its services are invaluable, and in any
large institution will replay its cost in a very short time." (The
Bankers' Magazine, Apr. 1874)
As late
as April 1889, an article in Manufacturer and Builder (p. 75) stated that
"Calculating machines, as a class, have proved themselves to be more
remarkable for the ingenuity displayed in their construction than for utility.
As marvelous pieces of mechanical construction, they excite surprise and
admiration, but it may be doubted if thus far any one of them has shown itself
to be adapted to practical service in the daily business routine of government
bureaux, banks and other financial institutions, and the like." One
limitation of calculating machines introduced prior 1885 was that numbers were
entered with levers (rather than keyboards). While lever-set machines were useful for scientific,
engineering, and other calculations requiring multiplication and division of long
numbers, data entry using levers was too slow for routine
accounting and banking applications, which involved large amounts of addition.
An entirely different type of calculating machines was designed for use in construction of the many types of
large tables -- logarithmic, trigonometric, nautical, astronomical, actuarial, civil
engineering, etc.-- that were published and used extensively in the 19th
century. (Manufacturer and Builder, Aug. 1870, p. 225. On early table-making, see Martin Campbell-Kelly and William
Aspray, Computer:
A History of the Information Machine, 1996, pp. 10-15.) These machines
were called difference engines because of their large size and because their
mode of calculation was based on the method of differences. The Scheutz Difference Engine,
shown above in an 1855 illustration, was made by Georg and Edvard Scheutz of Stockholm, Sweden. This
was a "machine, similar to Babbage's in principle, for calculating tables
by means of ratios or common 
differences, which not only made the calculations
but automatically cast stereotypes of its results, so that it avoided possible
printer's errors." (Henry Lucian Arnold, The Complete Cost-Keeper,
New York, 1907, p. 375) Grant's Difference Engine, pictured to the left and described further to
the right, was exhibited at
the 1876 Centennial Exposition in Philadelphia. (Courtesy of The
Free Library ) "George B. Grant [1849-1917] of Boston's
wonderfully ingenious difference machine constructs intricate logarithmic
tables, and solves all the problems of the differential calculus, preparing also
a waxen mould from which electrotype plates can be taken." (Appletons'
Annual Cyclopaedia, 1876, New York, NY, 1877, p. 276) Another contemporary
description states that Grant's difference engine "is a large machine built
for the University of Pennsylvania, and designed for the construction of large
mathematical tables, such as tables of logarithms, sines, tangents, reciprocals,
square and cube roots, etc. It stands five feet in height by eight in
length, weighs 2000 lbs., contains, when in full working order, some 15,000
pieces, and is worth about $10,000." (Spencer F. Baird, ed., Annual
Record of Science and Technology for 1876, Harper & Brothers, New York,
1876, p. 43) As far as we are aware, no more than a handful of difference
engines were completed by all the manufacturers put together.
The first calculating machine that was practical for routine office work, the Comptometer, was introduced commercially in 1887 by the Felt & Tarrant Manufacturing Co. The Comptometer did not have a printer. In 1889, Felt & Tarrant began production of the Comptograph, a calculating machine with a visible printer that has been described as "without question the first practical recording-adding machine ever sold that would produce legible printed records of items and total under the variable conditions that have to be met in such a class of recording" (J.A.V. Turck, Origin of Modern Calculating Machines, 1921, p. 116). The Burroughs Registering Accountant, which was the first practical adding-listing machine, was introduced in 1892 by the American Arithmometer Co., which became the Burroughs Adding Machine Co. in 1905.
Both Felt &
Tarrant and Burroughs machines used keypads for entry of numbers, but there were significant differences between
the machines made by the two companies. Burroughs produced adding
machines that printed entries and totals on paper
tapes or forms. The operator of a Burroughs entered the number on the keyboard and then pulled
and released a lever on the side of the machine, causing the entry to print and
to be added to the running total. In order to print the total, the operator
depressed the "Total" key and pulled and released the same lever. While the operator could read the running
total on a register behind the glass front of the machine, the printer was on the rear of the machine. The operator had to lift a carriage on the rear of the machine to
see what was printed. In short, Burroughs machines were "blind"
printers. Competing
adding machines that printed in view of the operator were
promoted as "visible."
With
the exception of the Comptograph, calculating
machines did not print inputs or
outputs. Comptometers (rhymes with "thermometers") and Comptographs were key-driven
calculating machines. To add a number to the running total, the operator entered
the number by pushing down on the keys and releasing. The lever on the right
side was used to clear the machine.
An office equipment book explains that "The adding machine prints the
items as it is operated while the calculating machine only indicates the items
without printing them. When a printed record is needed for checking the work,
for statements, or for other purposes, the adding machine best serves the
purpose. Adding-listing machines are usually preferable for pure addition and
subtraction, especially when equipped with a direct-subtraction device.
[Direct-subtraction was added to
Burroughs machines in 1911.] But for computations involving multiplication or
division, speed and economy usually demand that this work be performed on
calculating machines, unless the importance of preserving printed lists that
will serve as permanent records overbalances the desirability of speed."
(Practical Business Administration, American Technical Society, 1930)
Both Burroughs and Felt & Tarrant were founded in the
mid-1880s, but business was initially slow. Neither sold
significantly more than 5,000 machines by the end of
1900. Turck (1921, pp. 144-45) states that adding and calculating machines
"met with very strong opposition for the first few years. The efforts of
book-keepers and counting-house clerks to prevent these machines entering their
department were inspired by the fear that it would displace their services. Of
the two classes of machines, the listing machines were the first to sell in
quantities that may be called large. This was probably due to the fact that they
were largely sold to banks." While Burroughs adding machines were marketed
heavily to banks, photographs from the 1910s demonstrate that Burroughs machines
were used in a wide variety of office settings. Comptometer advertising targeted
accounting departments, while Turck (1921, p. 111) reports that Comptographs
were targeted at banks.
Sales of Burroughs machines grew rapidly after 1900, with Burroughs outselling Felt & Tarrant by
about five to one during 1901-1910 and, with an expanded product line, ten to
one during 1911-1920. The machines produced by these companies and a number of
smaller competitors revolutionized the way offices carried out numerical calculations, accounting, and billing in the early 1900s.
A large number of higher-salaried men with pens who added columns of four-digit numbers rapidly in their heads
were replaced by lower-salaried office workers, many of them women, with machines.
(Lisa M. Fine, The Souls of the Skyscraper, 1990, p. 92)
The earliest photograph we have found that shows an adding or calculating machine
in an office dates from 1903 and shows a Burroughs registering accountant in a
bank. (System, 1903) [Insert photo from System 1903] The second earliest shows Burroughs adding machines in a Sears, Roebuck & Co. office in 1906.
In 1907, Albert B. Barrett stated with regard to bank operations, "The
registering account, or adding machine,...has proved itself one of the most
useful instruments ever introduced to the banks." He also stated that
"Lists upon what are called exchange slips are made of the checks in each
file. This was formerly done with pen and ink, but now the arithmometer or
adding machine is almost universally used." (Modern Banking Methods,
5th ed., Bankers Publishing Co., New York, NY, pp. 50, 249.) Beginning in 1910,
office photographs commonly show adding and calculating machines.. Burroughs
is the most common brand in photos taken in the 1910s. We have also identified Comptometer, Dalton, Universal, and Wales
machines in photos taken during the 1910s. For a 1912 photograph of an
office with a Universal, see Lisa M. Fine, The Souls of the Skyscraper,
1990, p. 110. Burroughs adding machines are pictured in the 1911 catalog for
Hesser Business College, Manchester, NH, and the 1912 catalog for King's
Business College, Raleigh, NC.
During the first decade of the 20th century, adding and calculating machines
produced by Burroughs, Felt & Tarrant, and their closest competitors ranged in price from
$60-$200 for a Mechanical Accountant key-driven calculator, $150-$315 for a
Comptometer, and $160-$250 for a Standard Adding Machine to $350 for a
Comptograph and $375 for a Burroughs Adding and Listing Machine. Because
these
machines were expensive, there was a market for the numerous simple adders
that were advertised
in business publications at $5 to $35.
Different types of calculating machines, including stepped-drum,
pinwheel, and direct multiplication calculators, were faster than adding machines or
key-driven calculators for multiplication and division involving long numbers.
As a result, these other types of calculators were more suitable for scientific
and engineering applications. However, stepped-drum and pinwheel calculators, as
well as the available direct multiplication calculators,
were not efficient for high volumes
of addition and subtraction, and hence they did not play a major role in the
typical early business office, at least in the US.
For photographs of a number of adding and calculating machines that are not
included in the Early Office Museum's exhibits, see Jay
M. Goldman's collection.