Aiken completes the course.
January 31 -- February 3, 1938
Aiken visits IBM's Endicott Laboratory and meets with Clair Lake and Frank Hamilton. Aiken wants to discuss "the possibilities of designing a machine of a very complex nature for solving equations pertaining to the ionosphere to work on television and also equations pertaining to astronomy."* Four functions of adding, subtracting, multiplying and dividing are discussed, with more complicated functions, such as logarithms, sines, tangents and interpolation to be discussed later. Aiken suggests that tape be used for control of machine.
Also discussed are means for storing constants. Hamilton suggests use of the ITR cross-connecting program panel but that is abandoned because excessive time is required to change set-up between problems. Hamilton then suggests use of rotary switches because they can be readily changed and easily read. This idea is adopted and employed in the ASCC.
Aiken suggests the possibility of dividing by multiplication of the reciprocals of the divisor, with the reciprocals obtained by a mathematical process called "iteration." Hamilton and Lake believe this method requires too many machine cycles and favor instead Bryce's method, which is adopted and employed in the ASCC.
February 7, 1938
Aiken writes to Hamilton with notes from the four-day meeting. Among the features outlined in the notes are the use of tape to control the machine, and a decision made by Hamilton and Benjamin Durfee that "all adding counters, switches, card feeds, etc., be connected to a single bus of 24 columns for purposes of transferring values. This decision reduced the amount of connections and wiring in the machine to a very great extent."*
The notes contain suggestions that photocells be used to sense the code punching in the control tape and that teletype tape be considered. (Lake and Hamilton decide instead that the tape to be used would be IBM card stock and that the value of the perforations would be sensed by contact fingers, both of which are incorporated into the machine.)
The notes refer to the control of the algebraic sign of products and quotients. The participants agree that this will be indicated by the presence or absence of a "9" in the highest order of the product or quotient counter. Subtraction is to be accomplished by the addition of complements. Aiken establishes the principle that all values throughout the solution of a problem should have the decimal point in the same position. Hamilton and Durfee design a means for attaining this principle in the machine circuit.
February 17, 1938
Hamilton indicates "the total number of permutations which might be obtained by combinations of five holes. This was the nucleus of the final code punching although the final coding was on a basis of three groups of eight holes each."*
March 22, 1938
Bryce asks Lake if sufficient information is on hand to provide for an appropriation to begin work on the machine.
March 23, 1938
W. Wallace McDowell** informs Bryce that cost estimates are impossible pending preparation of actual wiring diagrams. He states that IBM's understanding with Aiken is that IBM "would furnish all standard units required, such as counters, relays, switches, etc. and that the mounting of these standard units and the design of the circuits employing their use would be done by Mr. Aiken."*
(The estimated cost of the special equipment to be supplied to Harvard on the basis of the original understanding is estimated at approximately $15,000.)
April 27-28, 1938
Bryce, Lake and Hamilton meet. Bryce informs his colleagues that IBM has acquired a patent license allowing IBM to "alter the units position of a counter when balancing dependent upon certain conditions in the highest order of the counter."*
May 16, 1938
Bryce provides Hamilton with a tentative arrangement of the machine's panels and the approximate quantity of required counter positions. This serves as the basis for the preliminary layout.
May 19, 1938
A work order provides funds for a temporary layout showing the arrangement of the units on a panel of the machine.
August 17, 1938
Work begins on the layout of the panel arrangement.
September 12, 1938
The temporary layout is completed.
Sept. 26-28, 1938
Bryce, Lake and Hamilton estimate the cost of the machine to be$100,000.
* Extracts from an August 1944 internal IBM report.
** McDowell joined IBM in 1930, and after a brief sales career, was assigned to the Endicott plant as a designer in the engineering laboratory. He was one of the top advisors to Thomas J. Watson, Jr., on the development of the important IBM 701 computer. McDowell became an IBM vice president in July 1954.