The BINAC. A product of the Eckert-Mauchly Computer Corp.

N.P. Eckert-Mauchly Computer Corp. 1949.

Very rare sales brochure for Eckert and Mauchly’s BINAC, the first operational stored-programme computer in the United States. J. Presper Eckert, together with his partner John Mauchly, invented and constructed the first general-purpose digital computer (the ENIAC) during World War II. After the war he and Mauchly founded the first commercial computer company in the United States, the Electronic Control Co., soon renamed the Eckert-Mauchly Computer Corporation. While developing the UNIVAC for the U. S. Census Bureau, Eckert and Mauchly contracted with the Northrop Aircraft Company in southern California to develop and construct a BINary Automatic Computer (BINAC). The contract was signed in October 1947, with Northrop providing $80,000 up front; another $20,000 was due upon delivery of the machine. “Had it been finished on time [i.e., by May 15, 1948] it would have been in contention with the British computers at Manchester and Cambridge as the first working electronic stored-program computer. In reality it ran its first program in the late summer of 1949 and, as a consequence, became the first operational computer in America but not in the world” (Williams, pp. 359-360). With both input and output in base 8 (a compact way of representing binary values), the BINAC’s design made it very difficult to use; in addition BINAC’s complex and delicate machinery suffered in the delivery from Philadelphia to California, so much so that it was never able to function effectively as a production machine. Published the year BINAC was delivered, this flyer contains the computer’s statistics, a brief outline of its elements and general characteristics, coding instructions, and a conversion table comparing decimal, coded decimal, binary, and octal numbers. A full-page illustration shows the various components of the system. The document was likely produced for distribution when the machine was being demonstrated to interested parties. It may be the only document extant that gives the complete instruction set (sixteen instructions) for BINAC. Few copies of this brochure would have been distributed, as only one machine was sold, and Eckert-Mauchly rapidly turned their attention to building UNIVAC. ABPC/RBH list only the OOC copy (Christie’s 2005).

“In 1948 the NBS [National Bureau of Standards] agreed to go ahead with the development work for UNIVAC and signed a contract. By that time the Eckert-Mauchly people were in a very awkward position. Because of the very limited income they had from the NBS contract over the last 18 months, they had searched around to find some other firm or agency that would be willing to keep the company alive until the UNIVAC development contracts would be awarded. Their search brought them into contact with the Northrop Aircraft Company which agreed to fund the development of a Binary Automatic Computer (BINAC). In October 1947 Northrop signed a contract for the construction of a computer to be finished by May 15, 1948 …

“Northrop had envisaged the BINAC as an experimental machine which might become the forerunner of a much smaller airborne guidance system for the Snark missile they had under development. In the end it saw very little use by Northrop; some employees even claimed that although it functioned in the Eckert-Mauchly workshop, it never ran when finally delivered. This has been disputed by a number of Northrop employees who indicate that it did run several small problems after it had been delivered to their California location, but that was never in good enough shape to be used as a production machine. It would appear that a large part of the problem stemmed from the fact that Northrop took delivery in the Eckert-Mauchly workshop in Philadelphia and did not appreciate the fact that shipping a very sophisticated, highly experimental, machine required more than simply building crates and calling in the moving men” (ibid., pp. 359-360).

The BINAC consisted of two identical serial computers operating in parallel, with mercury delay-line memories, and magnetic tape as secondary memories and auxiliary input devices. On September 9, 1948 the second module of the BINAC was completed in Philadelphia. Among its numerous innovations were germanium diodes in the logic processing hardware – probably the first application of semiconductors in computers. Until its delivery to Northrop Aircraft in September 1949, the BINAC remained in Philadelphia for use in numerous sales demonstrations.

“In February 1949 Albert A. Auerbach, one of the designers of the BINAC CPU, ran a small test routine for filling memory from the A register. This was the first program run on the first stored-program electronic computer produced in the United States. 

“On August 22, 1949 Eckert-Mauchly Computer Corporation of Philadelphia issued a press release describing the sale of the BINAC. This was the first press release ever issued for the sale of an electronic computer …

“Eckert and Mauchly’s BINAC was the first stored-program computer ever fully operational, since the Moore School’s EDVAC, which was designed to be the first stored-program computer, did not become operational until 1952. The BINAC was also the first stored-program computer that was ever sold.

“The BINAC was extremely advanced from a design standpoint: It was a binary computer with two serial CPUs, each with its own 512-word acoustic delay line memory. The CPUs were designed to continuously compare results to check for errors caused by hardware failures. It used approximately 1500 vacuum tubes, making it virtually a mini-computer compared to its predecessor, the large-room-sized ENIAC, which used approximately 18,000 vacuum tubes. The two 512-word acoustic mercury delay line memories were divided into 16 channels each holding 32 words of 31bits, with an additional 11-bit space between words to allow for circuit delays in switching. The clock rate was 4.25 MHz (1 MHz according to one source) which yielded a word time of about 10 microseconds. The addition time was 800 microseconds and the multiplication time was 1200 microseconds. New programs or data had to be entered manually in octal using an eight-key keypad. BINAC was significant for being able to perform high-speed arithmetic on binary numbers, although it had no provisions for storing characters or decimal digits.

“In 1946, after developing and building the ENIAC (the first general-purpose electronic computer) for the U. S. Army during World War II, J. Presper Eckert and John Mauchly founded their own company for the purpose of designing and manufacturing electronic stored-program computers on a commercial basis. In October 1947, needing money to keep their business afloat while working on their UNIVAC machine for the U.S. Census Bureau, Eckert and Mauchly entered into a contract with Northrop Aircraft to build the Binary Automatic Computer (BINAC). Northrop, based in Hawthorne, California, was then engaged in a project to build a long-range guided missile for the U.S. Air Force, and had the idea of using electronic computers for airborne navigation; the BINAC, while not designed to work in flight, would perhaps be an initial step toward that eventual goal. Airborne computers did not become feasible until the 1960s, when miniaturized solid-state transistorized components became available.

“The BINAC was completed in August, 1949, $178,000 over budget; Eckert and Mauchly absorbed the loss themselves. Built with two serial processors, the BINAC functioned more like two computers than one, with the goal of providing a safety back-up for airplanes. Each part of the device was built as a pair of systems that would check each step. All instructions were carried out once by each unit, and then the result would be compared between the units. If they matched, the next instruction would be carried out; but if there was a discrepancy between the two parts of the machine, it stopped. The processors were only five feet tall, four feet long and a foot wide, tiny for those days. The machine could only do 3,500 additions per second compared to 5,000 on the ENIAC, but it could do 1,000 multiplications per second, compared to only 333 on the ENIAC.

“Many histories of computing state that the BINAC never operated successfully; however, Northrop’s ‘Description of Northrop Computing Center,’ an internal company document dated September 16, 1950, which I also handled in 2014, listed the BINAC as one of its three main pieces of computing equipment, and even though the machine was currently ‘being revised and improved for more reliable operation,’ it was still functioning at least somewhat satisfactorily a year after its delivery.

‘This machine has solved in seven minutes a problem on the effect of a certain wind pressure on a rubber diaphragm that would have occupied a mathematician for a year. It has solved Poisson’s Equation and obtained a network of 26 solutions in only two hours. For each of these solutions, the BINAC performed 500,000 additions, 200,000 multiplications, and 300,000 transfers of control, all in the space of five minutes … This machine, which is a general purpose computer calculating in the binary system but receiving and emitting its instructions in the octal system, will be demonstrated today on a short test problem (‘Description of Northrop Computing Center,’ p. 2)” (historyofinformation.com).

“Besides the innovative approach of using two parallel machines to check each other and the one large tank design of the mercury delay lines, the BINAC was the first to employ magnetic tape as a secondary memory and auxiliary input device. The magnetic tape industry had just begun to develop plastic-based recording tapes for audio use and these were incorporated into a unit they called a converter for use on BINAC. The experiment was important to the Eckert Mauchly firm because the design of UNIVAC called for magnetic tape storage of data. The small plastic-based audio tapes proved to be unsatisfactory so Eckert-Mauchly redesigned the system to work with nickel coated bronze tapes for the UNIVAC.

“When Northrop took delivery of the BINAC in September of 1949 and paid Eckert-Mauchly the remainder of the $100,000 contract price, an accounting showed that the total cost of the machine had been $278,000. This, of course, left the Eckert-Mauchly firm in an almost bankrupt state, but it did indicate that they were capable of producing an electronic digital computer. The fact that they were able to show prospective clients some large sections of computer hardware resulted in their obtaining several orders for the still speculative UNIVAC. Three of the machines were destined for use by government agencies, two for use by the A. C. Nielsen Company in its accounting departments, and the final machine was for the Prudential Insurance Company which was having difficulties both in accounting and in the computing of new sets of actuarial tables” (ibid., pp. 359-361).

Origins of Cyberspace 1145. Williams, A History of Computing Technology, second edition, 1997.



Reproduced typescript (282 x 218mm), stapled, 8 sheets, including full-page illustration.

Item #5513

Price: $6,500.00

See all items in Computers, Numerical Methods
See all items by