ROBERTS | 171
were plenty of what we called ‘depths’—messages that had been enciphered using the same
key (see Chapter 14 for an explanation of key). Our intercept stations were often able to detect
when a pair of messages were depths, and they passed this information on to us. I would select
two longish ciphertexts that the intercept people thought had probably been enciphered with
the same key, and I added the two ciphertexts together. Because the messages (if depths) had
had exactly the same encryption applied to them, this process of addition had the effect of
‘cancelling out’ the common key. Although the resulting string of letters looked random, in fact
it consisted of the two plaintexts added together. My problem now was to separate out these
two plaintexts.
For example, the process of adding the two ciphertexts together might produce this stream
of letters:
. . . F J M 5 X E K L R J J . . .
(the dots indicate simply that the letters carry on to the left and to the right). I would start by
choosing a German word—the trick was to choose a good word—and I would then add this
to the stream of letters, beginning at some position in the stream that my instinct told me was
promising. If this didn’t work out, I would try adding my chosen word at a number of other
positions. If I was lucky, sooner or later I found a position where adding the chosen word
produced plaintext from the other message of the pair. For example, I might try the common
plaintext expression 9ROEM9. This related to Roman numerals. Units in the German Army
were commonly referred to by Roman numerals, I, XXII, CXX, and so on. The Tunny operators
dealt with these numerals by typing the German equivalent of ‘Roman 1’, ‘Roman 22’, and so
on. The abbreviation for ‘Roman’—ROEM (from ‘Roemisch’)—therefore cropped up a lot in
messages, usually with a ‘9’ stuck on at either end, since Tunny operators used ‘9’ to indicate a
space between words.
figure 16.1 Jerry Roberts at
Bletchley Park.
Reproduced with permission of Mei Roberts.