General Report on Tunny
23H Page 94
To set Χ4Χ5, having set Χ1Χ2 the useful runs (not all independent) are
 |
i.e. | 4.5./1.2. | ||
 |
i.e. | 4x5x/1x2x | ||
 |
i.e. | 4=5=/1=2 | ||
 |
i.e. | 4+5/1x2x | ||
 |
i.e. or |
 |
4+5/1x2x 4.5./1.2. |
All these can easily be run using multiple testing.
To set Χ3Χ5 having set Χ1Χ2 the only new useful run is 3.5./1.2.
The reader may be interested to work out all possible runs supposing that Χ3Χ4 are set first [R3 pp. 95, 124; R5 p. 106]
(f) Runs for the last wheel.
These may be expressed compactly
| For Χ3 | U / 5 J 3 F X 0 G P Q Y S H I R
A 9 8 K N D B M V L W Z E T 4 C | ||
| For Χ5 | / U 5 8 D F G Z 9 P 3 A Y M
T Q J K B X R E H Y 0 W S N |
The letters above are good letters in order of merit, the letter below is that which differs from it on the impulse to be set.
For hard flogging the letters may be run separately but simultaneously on the five counters of Colossus.
Otherwise the letters may be run in batches e.g. for Χ3 U/5; J3FIOG; PQYSH. [R4 pp. 42, 82.]
(g) ΔD Bigram Runs.
Because Colossus looks only at place and remembers one other, the use of these is limited. The multiple test memory circuits are unsuitable because they remember only a single wheel.
The ΔD bigram U5 since Colossus Δ's backwards, is equivalent to ΔD = 5 Δ2D = M. If a ΔZ tape and ΔΧ wheels are used the Colossus plugging and switching required is
| Z1 + Χ1 = x, | Z2 + Χ2 = x, | Z3 + Χ3 = ., | Z4 + Χ4 = x, | Z5 + Χ5 = x. | |
| ΔZ1 + ΔΧ1 = ., | ΔZ2 + ΔΧ2 = ., | ΔZ3 + ΔΧ3 = x, | ΔZ4 + ΔΧ4 = x, | ΔZ5 + ΔΧ5 = x. | |
(h) Use of evidence other than ΔD.
If a message can be set on some but not all Χ's it may yet be possible to
| (i) set the motors [23L] | |
| (ii) send a de-chi on fewer than 5 wheels to the Testery, where language methods can be applied. |
Back to General Report on Tunny. Contents.