92 | 10 THE ENIGmA mACHINE
reduce the number of possible wheel orders from 336 to as few as ten or twenty, so making the
codebreakers’ work much easier (although these rules remained undiscovered until 1944).^10
However, the greatest weakness of the German Navy’s daily key system was the fixed ground
setting. Turing’s method of ‘Banburismus’ (described in Chapters 13 and 38) depended entirely
on the fact that the ground setting was not changed during the day. Had the Germans used a
variable ground setting, it would seriously have impeded the decryption of messages.
The navy had a number of different communication networks, each with its own daily key.
The principal network for U-boats and surface ships in home waters, including large parts of
the Atlantic, was Heimische Gewässer, known at Bletchley Park as ‘Dolphin’. Up until September
1942 Dolphin, and its companion Triton (used at that time only on the M3), were the only
general Naval Enigma keys being broken. The ins and outs of Dolphin will now be explained.
The message setting
After the Enigma machine had been set up using the daily key, and the operator had carried out
some further preliminaries, described below, he would then turn the three wheels to the start-
ing position for the first message (e.g. BDK). Each wheel was turned until the letter in question
appeared in its viewing window.
The German Air Force and Army allowed their operators to choose these three letters.
Unfortunately, operators often chose rather obvious triples of letters, such as the letters on keys
sitting diagonally on the Enigma keyboard (e.g. QAW and WSX) making life simpler for the
codebreaker. The German Navy, more security conscious than the army and air force, tightened
security by issuing Dolphin operators with a book containing lists of potential message settings.
This book was called the Kenngruppenbuch or, at Bletchley Park, the ‘K-book’. The K-book
contained all 17,576 possible trigrams (i.e. three-letter groups, such as PQR). The same K-book
remained continuously in use from 1941 until the end of the war. In addition, once an opera-
tor had selected a trigram from the K-book, he employed ‘bigram tables’ to disguise it before
transmitting it. A bigram is a two-letter pair (such as PQ).
Bigram tables were paper tables pairing bigrams with bigrams, for all 676 (i.e. 26^2 ) bigrams.
A table might read AA = PY, AB = ZR, AC = NV, etc. AA = PY told the operator to disguise
occurrences of AA by putting PY instead (and vice versa). Nine complete tables constituted a
set, and which table of the set was in use on any given day was shown in a special calendar issued
to operators. (New sets of bigram tables were introduced in July 1940, June 1941, November
1941, March 1943, and July 1944.)
The procedure for creating and using a message setting was as follows:^11
1 The operator chose a trigram at random from the K-book, say ARQ. He then set his
wheels to the ground setting (the three-letter group fixed for the day and part of the
daily key), say JNY; this was done by turning the wheels, letter by letter, until JNY
appeared in the viewing windows. He then encrypted ARQ by typing ARQ at the key-
board, giving (say) LVN at the lampboard. LVN was the message setting, the position
to which he now set his wheels in order to begin encrypting the message itself (by
typing it at the keyboard).
2 The operator needed to send the trigram ARQ to the message’s recipient, so that the
recipient could decrypt it at the ground setting and discover the message setting. So the