SUBJECT: Fish Notes Report #F 36
TO : CO, SSA, War Dept. 7 May 1944
1. Enclosed is microfilm (D-127) of Fish traffic described
in detail in the subsequent paragraphs. The delay in forwarding
this has been due solely to the fact that everyone connected
with the Fish problem has been under terrific pressure. The
messages sent represent a selection from a mass of traffic
which had to be gone through with great care to find those best
suited, from the viewpoint of length and freedom from corrup-
tion, for the solution of the particular problems involved.
2. The first 7 messages on the film are Gurnard traffic.
This is between Berlin and Belgrade or Zagreb. The messages
selected are all from Berlin. This is a new link and has
never been read. It surely embodies an auto-key element and
probably has X2 and P5 limitations. The message numbers and
page lengths are as follows:- #1751-87, #1101-66, #500-57,
#1076-41, #1741-34, #1079-27, #1077-21. Since each page con-
tains 500 (or slightly fewer) characters, some of these are
of very great length and should be ideal for recovering X
patterns. The texts of #s 1079 1077 seem best, The early
part of #1101 is pretty bad and it would be advisable to work
only with the portion following the the bad stretch on page 32.
#1751 fairly good throughout and can probably be used in
Its entirity. It is long enough, however, to permit several
partitions to be selected for independent study, Since these
are all April messages the X and Psi patterns are presumably
the same for all of them,. There is, of course, a possibility
that patterns changed some time during the month as has
happened on other circuits.
3. Since Jellyfish has been broken a number of re-
encodements from Bream have been found, Sometimes these are
character for character; sometimes just the substance is re-
trasmitted. As a result a watch is being kept for possible
re-transmissions on other links. The next item on the film
is Gurnard message #1569 followed by Bream decode #1372.
There is good reason to believe that this Gurnard message is
identical with or similar to the decode. The same is true of
the next pair of items, Gurnard #1114 and Bream decode #l334.
No cryptanalytic study has been made to verify these conjec-
tures and these pairs are sent for such use as you may be able
to find for them. Presumably they will be of value only after
the X patterns have been recovered. Message lengths are not
necessarily significant because ot the practice of sending
several messages at once.