General Report on Tunny
25B Page 156
When the reading is i + j, the cipher, and therefore the true wheel also, has moved forward j places relative to the trigger, i.e. the trigger with its cross has moved backwards j places relative to the true wheel, and the score is that for the W - jth character of the wheel.
 |
This argument depends only on relative settings and is not invalidated by the fact that deltaed wheels are set up on Colossus. Because Colossus deltas the cipher backwards, the settings recorded must be increased by one to obtain true settings, [53E]. To avoid confusion, settings appropriate to deltaed wheels are used throughout wheel-breaking, and converted only when sending messages to Ops.
(d) Practical Procedure on Colossus.
Count R the number of places looked at: i.e. /'s and T's
Count Norm: i.e. /'s assuming that ΔΧ5 is all crosses, measured on Colossus, as T's with ΔΧ5 all dots.
Both R and norm are unaffected by stepping Χ5, and as a check are each measured at least twice, whilst Χ5steps.
Reset to the correct message setting: insert a cross in the last position of the Χ5 trigger, step Χ5 and start.
The frequent changes in the trigger were originally effected by pushing pins into the back of Colossus, but finally all machines used seriously for wheel-breaking were equipped with a wheel-breaking panel, on which each has a three-way switch, the three switch positions being
| (i) single cross in last position | |
| (ii) all dots | |
| (iii) patterns as set up on panel. |
(Early wheel-breaking on Robinson see R1, pp 51, 56, 86; On Colossus R1 p 96. Some suggestions not adopted R2 p 84, R4 p 26.)
25B WEIGHING THE EVIDENCE
(a) Significance test.
After a wheel-breaking run has been completed it is necessary to know
Back to General Report on Tunny. Contents.