SImPSON | 139
hole with a light below. Sliding the paper along, you punched a confetti-sized hole on each
pencil mark.
You took two messages with the first two wheel-settings matching, for instance CCE and CCK
or ALP and ALA. One Banbury on top of the other, you would slide along 25 places one way,
25 the other. Depending on the number of times each letter appears in German, lots of holes
co-incide, particularly when German code passes over German code—one must be correct.
Noted the number of holes in each position, hoping for a group. A group of five, six or more
had you jumping up and down. The results were passed on to the Crib Room, to be matched
with clear German and make the menu for the Bombe.
The German Navy in the Mediterranean had much less traffic. Same machine, same system,
but it ran from mid-day to mid-day. One person could cope with it all. I rather fancied myself at
the job and was called ‘Med Queen’ on my shift. It was a lovely rush every second day on day
shift to get a ‘throw-on’ menu by 4 p.m.
Eileen Plowman’s and Christine Ogilvie-Forbes’ ‘slide’ must have been a synonym for the
cryptanalysts’ ‘shove’.
Banbury One from the Hut 6 roof (Fig. 13.2) illustrates these accounts. Along the top are
written, not at all carefully for posterity, ‘TGZ’ above columns 1–6 and ‘B/078’ above columns
9–13: these are the three-letter setting—or indicator—and number recalled by Christine
Ogilvie-Forbes. Over columns 41–52 we have ‘with TGC 11xx/159’. Alexander’s history enables
us to interpret this as ‘when this message and another with indicator TGC were compared, with
an overlap of 159 letters, 11 repeats were found of which four came as two bigrams’ (two-letter
sequences). Chapter 38 pursues the significance of these figures.
After careful preservation, a selection of the papers from the roof—including of course the
two Banburies—went on display at Bletchley Park in March 2015.
Tetras
In Banbury One the indicators (TGZ and TGC) have their first two letters in common. However
the second of the two pathways mentioned used pairs of messages that had only their first
indicator letters in common (and therefore only one wheel position in common). Such pairs
were so much more numerous that a stringent criterion for which pairs to test was needed.
Fortunately, the structure of language again came to the rescue.
In practice, the letters repeated between messages often came not as single spies but, if not in
battalions, in bigrams, trigrams, and up to enneagrams (sequences of nine letters) and beyond.
The tetragram, a sequence of four letters, proved particularly useful, and was colloquially a
‘tetra’.
The observation that some letters occur more frequently than others in language becomes
more marked when we look at such sequences of letters occurring together. A two-letter
sequence such as ‘en’ occurs more frequently in English than the combination of ‘e’ and ‘n’
counted separately, but for some bigrams (say, ‘mf ’ or ‘bv’) the opposite applies. As before,
this spiky distribution of bigrams carries through into the repeats between identically enci-
phered messages. Thus a repeating bigram is more powerful evidence of the alignment’s being
true than two single repeats would be. A three-letter repeat is better still—and so it goes on.