The Mathematics Department and the Mark 1

Note: Alexander Ovsov has provided a translation of this page into Romanian. [ Toby Howard, 3 February 2012 ].

The Mathematics Department at Manchester made a relatively small but significant contribution to the progress of the Mark 1 machines. The two key figures were Professor M. H. A. (Max) Newman and Dr Alan M. Turing.

Max Newman lectured at Cambridge (to Tom Kilburn, Geoff Tootill and Alan Turing!) before he became one of the mathematicians drafted in the war into the Government Code and Cypher School at Bletchley Park, in 1942. (In fact it was his lectures that sparked Turing's famous early paper On Computable Numbers, and it was to Newman Turing first showed his paper.)

At Bletchley Park Newman directed the operation to break the second generation (Lorenz) encoding devices used by the Germans. This required very sophisticated mathematical and statistical techniques. He realised that it might be possible to build electronic machinery to assist in the processes, thereby bringing the time taken to break the codes from weeks to hours. After a few experimental attempts, with the help of TRE and then Tommy Flowers of the Post Office Research Laboratories at Dollis Hill, they produced a specification of the Colossus, which was effectively a fast electronic computer that was pre-programmed to carry out a small set of specific tasks. The first machine was operational around January 1944, and in all 10 were built. The Lorenz machines were used for transmitting high level messages, and the breaking of the codes gave the allies a great advantage from the weeks before D-day to the end of the war. The existence of Colossus was kept secret for 30 years.

After the war, Newman moved to the University of Manchester as Fielden Professor of Pure Mathematics, bringing two mathematicians from Bletchley Park with him, I.J. Good and D. Rees. Having seen the power and potential of the electronic computer with Colossus, he planned that they should investigate what kinds of problem would be suitable as applications for a computer. At the time, no stored-program computer had been built, and in 1946 he acquired a substantial grant from the Royal Society, funded by the Treasury, for "a projected calculating machine laboratory at Manchester University". There was a capital sum of £20,000 and an annual sum of £3,000 for staff salaries. He planned to build a computer as quickly and easily as possible, so that he could investigate the uses of computers, particularly for Pure mathematics. (This was partly to overcome resistance against providing government funding for a second computer on top of the NPL computer, designed by Turing, which was planned to provide a general national computing capability, where known requirements were dominated by Applied mathematics.) He immediately sent Rees to the Moore School course, and in the Autumn had a 3-month sabbatical to investigate possible sources for the design and construction of the computer, during which he visited the States himself.

Max Newman decided to base his computer on von Neumann's planned computer at the Institute of Advanced Studies in Princeton, USA. The IAS machine was designed around a storage device being developed by the Radio Corporation of America, the Selectron. In effect Newman was planning a very similar operation to Maurice Wilkes at Cambridge, who was basing his machine on the EDVAC and Mercury Acoustic Delay Line storage. However Newman's plans were severely delayed (as were von Neumann's) because the Selectron could not be made to work. (Ironically, in the end (see Selectron Progress Report), the IAS machine was forced to switch to using a Williams-Kilburn CRT store).

Meanwhile, Freddie Williams and Tom Kilburn had arrived at the Electrical Engineering department and they got their different design for a store, using a Cathode Ray Tube, to work within a year, by the autumn of 1947. So Newman abandoned his plan to build his own computer and waited for the outcome of Williams' and Kilburn's attempts to build one. In practice, he had relatively little influence on the design and building of the Baby and its evolution to the Manchester Mark 1. However he did have discussions with the team and acted as consultant for the potential uses of computers. He had a particular knowledge of the world scene, having always kept in regular contact with Turing, and so knowing about his ideas and ACE design, and being aware of the scene in the States from his and Rees' visits. He certainly had at least an indirect influence on the design of the Manchester Mark 1, mostly completed by November 1948, in that he had identified Mersenne Primes as a suitable subject for computer usage in pure mathematics; he and Turing had started working towards such usage during the summer of 1948, on the Baby, and it was clearly going to require multilength integer arithmetic. This requirement probably meant that a more comprehensive set of instructions was provided for the double-length accumulator than might otherwise have been the case. Newman did contribute to one of the 34 Mark 1 patents, sharing with Williams, Kilburn and Tootill the famous patent for instruction modification registers ('B-lines') -- this came out of an hour's brainstorming session, and no one in retrospect could remember who had suggested it.

Newman's grant was not required to fund the Baby, as the CRT store research and the building of the Baby were funded in effect by TRE (providing nearly all the electrical components and the two full-time workers, Tom Kilburn and Geoff Tootill). The first major call on the Royal Society grant was for the academic year 1948/49, to fund the appointment of Alan Turing to the Maths department and pay some of the wage bill for the extra staff being taken on by Electrical Engineering. TRE still supplied the components (and Ferranti two new drums). In the end the capital sum and the remaining half of the salaries component was spent in 1950 on a new building to house the Ferranti Mark 1.

Good and Rees appear to have been disillusioned by the amount of work involved in programming in practice and left in 1948 and 1949 respectively.

For a short biography on his general career see here, though note that the comment that Newman worked with Turing at Bletchley Park is misleading, in that they did not work on the same project.

Note that I am a bit worried that I am being unfair to Newman, and I am still getting flak on the subject. If you want further justification of my viewpoint . . .

Alan Turing was a brilliant original thinker. Formally a mathematician, in his lifetime he studied and wrote papers over a whole spectrum of subjects, from philosophy and psychology through to physics, chemistry and biology. He graduated from Cambridge in Mathematics in 1934, was a fellow at Kings College for two years, during which he wrote his famous paper which introduced the Turing Machine, went to Princeton for two years to do a Ph.D., and returned to Kings for a year. At the outbreak of war in September 1939 he was drafted to the Government Code and Cypher School at Bletchley Park as a cryptanalyst. Here he made a major contribution to the battle to decode the German Enigma encodings, designing the "Bombe", though he was not directly involved with the later Colossus project. After the war he went to NPL to design a stored-program computer for them, the ACE. But after delays in starting to build ACE he went back to Kings for a year, before being invited by Max Newman to come to Manchester.

Turing joined the Department of Mathematics as a Reader in September 1948, with the nominal title of "Deputy Director of the Royal Society Computing Machine Laboratory". (The Royal Society Computing Machine Laboratory was the room the Baby occupied; there was no known "Director"!) It is not clear what his official duties were initially with respect to the Baby/Mark 1 project. Before Turing started work in Manchester he asked for the Baby order code and sent up a routine for long division, which was corrected and got working by Tootill. As soon as the Manchester Mark 1 was generally available for use in April 1949, he enthusiastically set about using it, especially to investigate Mersenne Primes, in collaboration with Newman. In the summer of 1949 he was instrumental in acquiring paper tape equipment and assisted Dai Edwards in attaching it to the Mark 1. Meanwhile, he was continuing his theoretical work and in 1950 published another famous paper "Computing Machinery and Intelligence", which anticipated the subject of Artificial Intelligence.

Turing made contributions to the extra orders added in the Ferranti Mark 1, notably the random number generator. It is also likely that he and Newman influenced the comprehensive set of instructions provided on the Manchester Mark 1 in connection with the double length accumulator, since they required multi-length arithmetic for their Mersenne Primes work. (In practice there was little subsequent usage of the Mark 1s for anything longer than double-length arithmetic).

However, the main formal contribution Turing made to the Mark 1 project was that he worked on providing the early software requirements for the Manchester Mark 1, with the full time help of Cicely Popplewell, and wrote the first programming manual for the Ferranti Mark 1. So it was Alan Turing who was mainly responsible for the decision to use the Base-32 Numerical System and he devised with Cicely the Scheme A method of program organisation. He would therefore have been involved in writing the Scheme A Input routines and standard subroutines for general use.

By 1951, Newman and Turing (and indeed Freddie Williams) had withdrawn from active involvement in the Mark 1 Project and subsequent computer development, leaving the newly-arrived Tony Brooker to look after the interests of the programmer.

However Turing was still a keen user of the computer as a tool for his research interests, which now turned to "morphogenesis", the theory of growth and form in biology. And he was was always ready informally to help out other programmers of the Mark 1 with their problems.

Alan Turing remained at Manchester till his untimely death in June 1954.