A typical command, of which the drum will hold 4000 in sequence, might read: "select the Number in A-1 and add to it the Number in B-2 and put the sum in C-3." Such a command can be given and the work accomplished by the computing section of the machine every four thousandth of a second.
Magnetic Tape Used
As Mark III solves problems, the answers are recorded on another magnetic tape similar to the one on which the information was put into the machine. A magnetic tape is used because Mark III turns out answers faster than a typewriter can print them. To overcome this difficulty, the tape is then fed to a tape reader which relays the answers to a battery of five typewriters.
Computations of the machine are checked at several points for possible errors. During the mathematical operations in the arithmetic section of Mark III, answers are double-checked at every stage in the problem before going on to the next stage. To insure that the final typed results are correct, all numbers are recorded twice on the magnetic tape by two parallel and independent systems. Unless both numbers on the tape are identical, the striking action of the type bars sets off an alarm and stops the typewriters.
To eliminate errors that might creep into answer sheets during type-setting and printing of permanent records, the results are printed directly to pages which can be photographed and printed by offset lithography for publication by the University Press.
Built for Navy
Mark III, which was begun in May, 1946, has been built for the Bureau of the Ordnance of the U. S. Navy to be used at the Naval Proving Ground Command at Dahlgren, Virginia. It is anticipated that testing operations will have been completed by the first of the year and the machine will go to the Navy at that time.
Mark III, a bakelite and steel instrument, is about 30 feet long and fifteen feet wide and weighs close to ten tons. It contains 100 miles of wire, about 4500 vacuum tubes for the electronic operations, 3000 relays, 2500 magnetic heads and play-backs to carry the information to and from the storage drums, and 400,000 solder connections. A staff of about 40 worked on the development and construction of the machine.
Benjamin L. Moore, assistant director of the Computation Laboratory, and Way Doug Woo, assistant professor of Applied Mathematics, were directly in charge of its development, design, and construction. Mechanical design and construction of the internal high speed magnetic drum storage system, one of the major components of Mark III, was the work of Robert Wilkins, assisted by Dexter Smith.
The adder and multiplier were largely the work of Charles Coolidge. Marshall Kinkaid was mainly responsible for the over-all design of the sequencing circuits. The input and output circuits were constructed by Richard Hofheimer.
Charles Richards, who had previously worked on Mark II and will go to Dahigren with Mark III, also worked on design problems throughout the construction period.
Professor Howard Aiken, co-inventor of the original calculating machine, and Director of the Harvard Computation Laboratory had general supervision of the project.