15C Page 34
(e) Automatic Recording.
In the Robinson as originally designed the selected readings (those above the 'set-total') were shown on a screen, to be copied down by operators, who were then to cancel the reading by a switch. Shortly before the machine was finished Mr. Gifford, of TRE, suggested that he should design a printer which would print the settings and totals. The automatic recording to which this led proved to be an indispensable part of the process. For operations in which certain initial scores form the basis of complicated later runs, the extra hazards introduced by mistakes and fatigue of copying, and lack of uniformity in hand written dossiers, are great enough to reduce the proportion of success substantially. A rack for automatic recording was therefore made a part of Colossus, even though this entailed some weeks' delay in the arrival of Colossus I.
15B EARLY ORGANISATION AND DIFFICULTIES.
(a) The Initial Staff.
The initial staff of Mr. Newman's Section consisted of M.H.A. Newman, soon joined by D. Michie, with 16 Wren operators and two engineers, working first two shifts and then three, in a two-roomed hut (Hut 11.)
(b) Development of the system of checks.
The early difficulties were sufficiently severe to prevent more than three messages from being set in any week in the first three months of operation. They arose partly from machine faults, (incorrect tapes from Tunny and incorrect counts on Robinson), partly from operator's errors. The standard of accuracy needed before there was any possibility of success was very much higher than would ordinarily be required of this kind of apparatus, or of operators. A single letter omitted in a tape destroyed the value of the run and the ordinary length of a tape was about 3000 letters. A count missed at the beginning of a run on Robinson gave wheel settings bearing no simple relations to the true ones. In addition there were numerous opportunities for wrong plugging, switching, and tape-setting on both machines. An error which passed undetected through several stages of the work could take hours or even days to track down.
To remedy this state of affairs a system of checks was gradually evolved which made it a rare occurence for a mistake to persist through several operations. To achieve this very elaborate checks were necessary, and about half the operational time was occupied in carrying them out. It was made a principle that the design of a new routine must include all the checks required, and in estimating the merits of a proposed routine the nature of the checks required had always to be take into account. It is for this reason that checks are described so fully in the chapters that follow.
15C PERIOD OF EXPANSION.
(a) Mass production of Robinsons.
Towards the end of 1943 the pressure for a large production by machine methods had grown, for two main reasons. The Tunny network had grown, the value of the contents had raised the traffic to the highest level, and the tightening up of German precautions against 'depths' had caused production by 'hand' setting methods to sink almost to zero. The introduction of the P5 limitation of the end of 1943 made depth-reading impossible. A large programme of machine construction was therefore embarked on. Twelve Robinsons were ordered in the late summer of 1943, and the first factory model