Unisys History Newsletter

Volume 4, Number 1
May 2000

Remington Rand Tabulating Machines

by George Gray

Tabulating Machines

One of the largest data tabulation tasks of the nineteenth century was the U.S. Census. The U.S. Constitution simply required an enumeration of free persons and other persons (which meant slaves) in each state every ten years, but within a few decades the census had become much more complex. A temporary staff was put together to do each census and then dissolved when the task was completed. In the 1880s, the government faced a serious problem. It had taken a staff of nearly 1500 people in Washington seven years to process the census of 1880 using manual methods, and there was great concern that the census of 1890 would not be finished before it was time to do another in 1900.

Herman Hollerith

Francis Walker, superintendent of the 1870 and 1880 census, and one of his assistants, John Billings, encouraged the efforts of the inventor Herman Hollerith (1860-1929) to find a faster way to handle the 1890 census. At the suggestion of Billings, Hollerith devised a system where the information for each person would be represented by punching holes in a card. This concept arose from a technique used by railroad conductors to ensure that each passenger had his or her own ticket: holes were punched around the edges of the ticket to indicate the sex and physical appearance of the passenger. Hollerith filed for patents during the mid-1880s and tried out his equipment in 1886 by doing a tabulation of death statistics for the city of Baltimore's health department. Robert Porter, superintendent of the 1890 census, appointed a three-man committee (one of whom was John Billings) to evaluate Hollerith's approach and two others. It recommended Hollerith's system for the 1890 census. Hollerith used cards that were approximately 6 5/8 inches by 3 1/4 inches in size, with 12 rows and 24 columns. Round holes punched in various locations indicated the characteristics of a person (sex, race, state of residence, etc.), and the sorters and tabulators used electricity to sense the presence or absence of a hole. His sorting box could process up to 80 cards per minute. Using Hollerith's methods, tabulation of the 1890 census was completed in a little over two years.

After completing the census, Hollerith founded the Tabulating Machine Company in 1896. It made punched card machines for railroad accounting. The New York Central Railroad, his first customer, processed over four million freight waybills per year. The tabulating machines enabled the railroad to keep up with increasing volumes and provide more frequent summary reports. By 1900, the company sold a tabulator, a sorting machine that handled 300 cards per minute, and an improved card punch. To do the 1900 census, the government used 311 tabulating machines, 20 automatic sorters, and 1021 punches, for which the Tabulating Machine Company was paid $428,239. Hollerith redesigned his cards in 1906 to have ten rows (for the values zero through nine) and 37 columns on a card that measured 7 3/8 by 3 1/4 inches. The number of columns was later increased to 45 on the same card size. In 1911, Hollerith's company merged with two others to form the Computing-Tabulating-Recording Company (CTR), and he took a position as a consulting engineer, leaving the direction of the business to others. Thomas J. Watson became general manager of the company in 1914 and moved up to president the following year. In 1924 he changed the company name to International Business Machines Corporation (IBM).

James Powers

Congress created a permanent Bureau of the Census in 1902. Simeon North, who became director in 1903, felt that Hollerith's company was charging too much money for use of its machines. He secured an appropriation of $40,000 to fund the development of new machines. In 1907, the Bureau hired James Powers, a Russian-born mechanical engineer who had emigrated to the U.S. in 1889. Powers developed a different type of card punch, one on which pins were set as each key was pressed, and then all columns were punched simultaneously when the card was finished. This reduced the wastage of cards due to typing errors. His sorters and tabulators used a mechanical approach to sensing holes, as opposed to Hollerith's electrical sensing, thereby avoiding patent infringement. The Powers machines, which used the same card format as Hollerith's, were used to process most of the 1910 census, although about one-third of the work was still done on old Hollerith equipment.

The government gave Powers the right to patent his machines, and he left the Census Bureau in 1911 to establish the Powers Accounting Machine Company in Newark, New Jersey. The company sold a punch, a sorter that was somewhat easier to use than Hollerith's, and a printing tabulator developed by W.W. Lasker. The printing feature was an improvement over Hollerith's tabulator, where the totals had to be written down by hand. In 1914, the Powers company moved to Brooklyn and also began selling in Europe. It asked CTR to sell it licenses to some of Hollerith's patents. Thomas Watson, the president of CTR, was very sensitive on issues relating to U.S. antitrust laws and agreed to the sale. However, the terms were very stiff: a payment to CTR of 25 percent of the gross rental of Powers machines for a license which covered only mechanical sensing of the punched card holes. Electrical sensing was not licensed. This left Powers with both cost and technical handicaps, a position so tenuous that the company was on the verge of closing during the recession of 1921. In 1922, CTR consented to cut the license fees in half, since Watson still feared a government antitrust suit should Powers go out of business. Powers merged with Remington Typewriter Company and Rand Cardex to form Remington-Rand Corporation in 1927.

IBM and Remington Rand

IBM and Remington Rand shared the punched card data processing market in the United States, in what economists call a duopoly. It is similar to a monopoly, but there are two companies instead of just one in an industry. At the time it was formed, Remington Rand was twice the size of IBM, but its principal products were typewriters and office equipment. The Powers operations became the Tabulating Machine Division, which accounted for roughly one fourth of corporate revenue. By the mid-1920s, IBM products were generally regarded as technologically superior to those of Powers, and IBM's market share increased to 80 percent by the end of the decade. In 1928 IBM redesigned its punched card to use narrow rectangular holes, permitting an increase from 45 to 80 columns on the same size card. This meant that IBM and Powers machines were no longer compatible: a punched card customer could not use a combination of machines from the two companies.

Remington Rand responded in 1930 by changing to a 90-column format, continuing to use round holes. The card was the same size as the new IBM card, but the layout was quite different. The Remington Rand card was composed of an upper half (containing columns 1-45) and a lower half (columns 46-90). Each card half had six rows, numbered 0, 1, 3, 5, 7, and 9. Zero and odd numbers were represented by a single punch in the appropriate position. Even numbers were represented by two punches: row 9 was always punched and the other punch went in the row of the number one less than the even number desired. Thus the value 4 would be represented by punches in the 3 and 9 rows. The letters of the alphabet and various other characters (such as the comma, percent sign, and ampersand) were represented by combinations of anywhere from two to five punches in a column.

IBM was in a relatively strong position when the Great Depression of 1929 hit. Its tabulating equipment was rented, not sold, and rental income held up well enough that company revenue fell less than ten percent from 1929 to 1932. This enabled IBM to continue with product research and development. It introduced the Type 600 multiplying punch in 1931, which had the ability to read two numbers from a card, multiply them, and punch the result in other columns of the card. In order to avoid layoffs, IBM continued to produce tabulating equipment and store it in warehouses, waiting for the economy to recover. These stored machines were needed after the passage of the Social Security Act of 1935 created new record-keeping requirements. All Social Security checks were printed on IBM punched cards. Although Remington Rand's tabulating equipment was also rented, the corporation as a whole was hit hard by the Depression. It was unable to catch up to IBM in reputation or sales. In 1940 IBM's profit was $21.7 million as compared to $4.9 million for Remington Rand, and IBM was estimated to have 90 percent of the tabulating market. This dominance continued during World War II when IBM supplied nearly all of the machines for military use.