of the auto-key element. One problem being considered is a statistical
means of differentiating between this traffic and ordinary Fish traffic.
Other special problems such as solution where wheel order is known but
settings unknown, or. where settings are known and wheel order unknown,
are receiving consideration. Each of these can be subdivided depending
on whether or not a depth is available. The possibilities of both
statistical and cribbing methods are being studied.
22. Without a machine to assist in enciphering and deciphering, all
the necessary operations are exceptionally laborious. Even the development
of £ and pi keys from a given wheel order and setting is very slow and
tedious. No plans have yet been made to construct a machine. The actual
machine is believed to have a zeroizing mechanism and is probably similar
to a captured machine (apparently of earlier design) which Lt. Col.
Rowlett and I saw on a recent trip to Knockliolt. He is planning to take
some photographs of this model.
23. Cryptanalysis of traffic enciphered with this machine seems to
present extraordinary difficulties and I shall try to keep abreast of
all developments here.
24. Paragraph 4 is not complete because it doesn't consider the
situation where only the X2 limitation is used. In such cases the first
MT, position is still always cross but the second is dependent on X2one back.
If the X2 pattern starts at 5, X2one back starts at 5 in the second column and
a conventional dot is put in the first column to compel a cross in MT. This
was the meaning of the red dot in Report #F 5. Looked at from the
viewpoint of the machine, rather than from the viewpoint of the conventions
used here, we can generalize as to both types of limitation and say that
MT must be cross in the first position and that from there on MT is MB as
affected by the limitation - but the limitation does not operate until
all elements comprising it have had an opportunity to become effective.