BABBAGE'S MONOGRAPH ON CILLIES
Babbage wrote to Welchman on 1 October 1982 and included a short monograph on Cillies, taking issue with Welchman’s description of them in The Hut Six Story and even his spelling of the word. (See page 115, The Bletchley Park Codebreakers: In Their Own Word).
Sillies
I think that these were originally christened Cillies (perhaps by Dilly Knox?). They were not as described in pp99-103. If we had a 3-part message like that at the bottom of page 99. We should of course have guessed the text settings, but this wasn’t a true cilli, and I don’t remember anything quite so obvious occurring.
The essential thing about a cilli operator was that when he had encoded a message he would leave the wheels unaltered when he came to encode the text setting of his second message (or second part of the same message). If he had done this then by ‘alphabetical subtraction’ of the number of letters in the first message from the 3-letter indicator setting of the second we would get back to the text setting of the first message and thus have a 3-letter decode. The nature of this ‘alphabetical subtraction’ depended on the wheel-order. When a wheel advanced from one letter to the next, the wheel (if any) on it left would normally stay put, but when the first wheel passed through a certain ‘critical position’ it would take its left hand neighbour with it. The critical positions for wheels 1, 2, 3, 4, 5 were
Q/R, E/F, V/W, J/K, Z/A
respectively.
Suppose we had a 5-part message whose indicator settings and indicators were as follows, where the figures in brackets give the number of letters in the text:
1) AVS MBU (161)
2) QGD LRT (166)
3) XZH FJK (167)
4) EKN YTL (171)
5) RMK BHC (150)
If we subtract 161 (or 6 x 26 + 5) ‘alphabetically’ from QGD we get QAY. 166 (or 6 x 26 + 10) from XZH gives XTX. 167 (or 6 x 26 + 11) from EKN gives EEC and 171 (or 6 x 26 + 15) from RMK gives RGV. QAY, XTX, EEC, RGV are very nearly the four consecutive ‘keyboard diagonals’ QAY, WSX, EDC, RFV, and we can in fact get back to these by ‘alphabetical subtraction’ if we make certain assumptions about the wheel order. Let us assume that we are dealing with a cilli operator and that the text settings of the first four parts are QAY, WSX, EDC, RFV. We get from QAY to QGD in 6 x 26 + 5 steps provided the right hand wheel does not pass through a critical position in going from Y to D. If it does pass through such a position we arrive at QHD (or at RHD if also the middle wheel passes through a critical position in going from A to H). If the middle wheel has a critical position between A and G and the right hand one does not have one between Y and D, we arrive at RGD. We deduce that the middle wheel is not 2 and that the right wheel is not 5. In order to get from WSX to XZH in 6 x 26 + 10 steps the middle wheel must have a critical position between S and Z, and the right hand wheel must have one between X and H. Hence the middle wheel must be 3 and the right hand one 5 or 2. The above assumptions lead us to the conclusion that the wheel order is 132 or 432 or 532, and it may be verified that with any of these wheel orders one does get from EDC to EKN in 167 steps, and from RFV to RMK in 171 steps.
Thus we have a 12-letter decode, or rather a 15-letter one, because we may assume with some confidence that the text setting of the fifth part is TGB, which is the fifth in the sequence of ‘keyboard diagonals’. In searching for the stecker pairings we have only 3 wheel orders to consider instead of 60, and a strong Herivel tip might reduce the number of ringstellungs to be considered from 26 x 26 x 26 to 1, at any rate as a first shot. I distinctly remember just such a situation occurring one day and my working out the machine setting by myself well before 7 a.m.
Apart from triads of letters forming recognisable patterns on the keyboard, some lazy cilli operators chose text settings three steps ahead of the indicator, not so much JAB JAB as JAB JAE, the point being that, after the text setting had been encoded, the wheels would be in the correct position for encoding the main text.
Other popular text settings were pronounceable 3-letter groups. One was SAK, which I suspect means something rather rude.
In searching for cillies of all these kinds one examined a sequence of messages from the same operator and carried out what I have described as alphabetical subtraction combined with intelligent guesswork about the wheel order. Perhaps Dilly Knox found a pronounceable sequence including CIL, short for CILLI, a girl’s name?
Image: Dennis Babbage in later life. Credit: Babbage family
