15A Page 33
15 - SOME HISTORICAL NOTES
15A FIRST STAGES IN MACHINE DEVELOPMENT.
(a) Early development of Statistical methods.
The idea of breaking single Tunny messages without depth by statistical methods was first propounded in the autumn of 1942. The '1+2 - break in' was invented by W. Tutte in November, 1942, and tested out with success by paper stencils. He also suggested at this time the breaking of chi-wheel patterns by means of the rectangle, and succeeded in finding the chis from a message 15,000 letters long.
Methods for setting motors and psi-wheels (by 'contracting' de-chis) and the rectangle-method for breaking motors, were suggested by others working in the Research section at that time.
(b) Proposals for the use of machinery.
The idea of using electronic counters to carry out these processes at a practically useful speed was put forward by M.H.A. Newman and in December, 1942 he was given the task of developing machine methods of setting TUNNY.
A number of schemes were considered, including that of sliding photographic plates over each other, a method later perfected in U.S. It was soon settled that the best machine for the early experimental stages was one which read a 'message-tape' and a 'wheel-tape' photo-electrically, and combined them electrically before counting. Emphasis was laid from the start on the need for flexibility, in order that the routines designed in abstracto might be able to be modified in the light of experience without changing the machine.
(c) Heath Robinson.
The result of many discussions was the two-tape machine later called 'Robinson'. It consisted of a valve and relay counter, designed by Dr. Wynn-Williams, coupled to a tape-rack ('bedstead') and a "combining" unit, designed by Mr. Flowers of the Post Office Research Station, Dollis Hill. The Pilot model, Heath Robinson, was commissioned in January, 1943 and began working in June of the same year.
'Heath Robinson' amply satisfied the demands for flexibility, and there can be little doubt that the opportunities it gave for trying new techniques at this crucial stage played a decisive part in the later successes of Colossus.
(d) The first 'Tunny'.
The 'Robinson' machine for making counts was accompanied by what was called in the section the 'Tunny' machine, for preparing tapes. This was essentially a reproduction of the German machine in terms of relays and uniselectors, but with facilities for switching in only a selection of the wheels and impulses.
It is an important feature of all apparatus used in the section that it uses standard five impulse tape, without any special preliminary processing. Although this led to a good deal of trouble both in designing the apparatus and in the early days of operation, through stretching tapes, it was well worth while surmounting these troubles in order to be able to use ordinary commercially produced tapes and tape-making plants, (later including American (I.B.M.) Machinery).