SUBJECT: Fish.Notes Report #F 22
TO : CO, SSA, War Dept. 17 April 1944
1. March Bream furnished the greatest volume of decodes. in the
history of the section, over one and one-half million letters of plain
text having been turned out.
2. The March Jellyfish wheels have finally been broken and a few
messages have been read. They have fulfilled all expectations with
respect to inteljiigence value. The circuit uses the X2 and P5 limitations
precisely as does Bream. The Mu37 patterns contained 22 dots. This is
quite favorable so that it is probable that a substantial amount of March
traffic can be read if time permits. Work on the.April Jellyfish wheels
is in progress.and..they look hopeful. The plain language characteristics
(judged from only a. handful of decodes) seem a little:better than the
Berlin end of Bream but not as good as the Rome end. (The Rome operators
have.bad habits which make that end of the circuit much easier to read
and Bream work always starts on the messages originating from Rome.. The
difference.is more important to Mr. Newman's section than to Maj. Tester's.)
3. The April. Bream wheels are broken but Mu37 has only 17 dots and
this makes it difficult to set the Psi patterns and read messages. At
present. only about 50% of the de-chies are being read, although it is.
anticipated that the percentage will increase later in .the.month. This
type or situation revives interest in the machine we contemplated for
setting Psi patterns by dragging probable words. Mr. Welchman has just
advised that it has been decided to proceed with this project but before
this was known the people here were urging me to suggest that we go
ahead as quickly as possible. The value of the machine method, as an
adjunct to the hand methods employed, is that the very fact that makes
the latter difficult adds to the sensitivity of the former.. This may be
slight but less contraction of the Psis will cause some reduction in the
number of stops and may make it feasible to work with probable words
shorter than 10 letters.,
4 In working on April Bream, Maj. Tester's section, for the first
time, derived the Psi patterns without a depth. In theory, no plain text
assumption applied.to a de-chi can be disproved (unless long enough to
repeat a portionof the cycle of'the shortest Psi wheel) because the
resultant Psi' key is always possible. But actually there is a high
probability.of the assumption being wrong unless it produces a substantial
number. of repeats and anti-repeats in the Psi' key. Although sure cribs
are not available, familiarity with the traffic proved a.help in this
process which, in any event, is most difficult and requires a high
degree of skill and a good deal of training and background. Once an
assumption. has been fitted into a position which yields a.likely stretch
of Psi' key, efforts are made to extend this forward and backward. After.