General Report on Tunny
22G page 54
| and therefore Δ2Kij → dot | (F 6) | ||
| Key has once been recognised by this method (see R3 p. 22) (R3 pp. 15, 76) | |||
| Further | |||
| Δ2Ψ' = | stroke at double dots in the TM | ||
 |
since  |
||
 |
at double dots in the TM (R0 p. 53) | (F 7) | |
(d) The sum of key streams.
There are a few words on this topic in 22W(b).
22G THE PLAIN LANGUAGE STREAM
(a) P and ΔP.
Machine methods of work on Tunny make it important that we should be able to recognise plain language not only by its linguistic, but also by its statistical properties.
The statistical properties of the P stream are obvious enough, the frequency of the various letters ranging from that of 9 (space) which normally occurs once in every 6 or 7 letters to that of stroke, 3, and 4 which should not occur at all.
In ΔP the frequency of each letter depends on the frequency of the 32 bigrams which add up to it. The letter count is not as bulgy as that of P, but is of greater basic importance in view of its contribution to the count of ΔD.
Fig. 2 (IV) shows bigram frequencies and their contribution to the various letters of ΔP in a sample of 25,600 letters of Jellyfish June 1944.
The first references to ΔP counts are on R0 pp. 21, 45 - 7 and to P counts on R2 pp. 83, 110 - 2)
(b) Heterogenous nature of P and ΔP.
Fish messages consist of a mixture of three component types of P: German language (in letter shift), numerals (in figure shift), and punctuation (involving frequent shift changes). The P and ΔP counts for these components are strikingly different and, even within each type the form of the count depended on the operators spacing and punctuation
Back to General Report on Tunny. Contents.