Welcome to Anoraks Corner, the vulnerabilities of the Enigma machine
and how these were exploited.
1. Vulnerabilities due to the physical characteristics.
1.1 Because of the reflector, no letter can encipher to itself. So a guessed
decrypt cannot be correct if the same letter appears at the same point in
both cipher text and guessed decrypt.
1.2 The wheel rotational position at which carry occured was different
for each wheel. This vulnerability was exploited by sophisticated techniques
which could decide, from the turnover, which wheel was on the right hand side
and even sometimes, which wheel was in the centre.
This difference in carry position persisted through wheels 1 to 5, but
the later Naval wheels, 6,7 and 8 had carry notches at the same rotational positions.
2. Vulnerabilities arising from operational use.
2.1 The setting sheets. These could be captured by an enemy. They usually
covered all the days in a month and emergency procedures were evolved
to cover situations where they were known to have been captured.
2.2 Relying on the operator to chose message keys. Humans are notoriously
bad at chosing random sets of three letters. In early use keys such as
BBB occured regularly together with keyboard patterns such as ASD or
alphabetic sequences like XYZ. Later the same problem arose in chosing
indicator letters, compounded by there now being two sets of three letters
which inevitably became related like BER LIN.
2.3 Message lengths were restricted to less than 250 letters to avoid
repetions of the enciphering cycle. This meant that many messages were
multi-part and vulnerabilities arose in the designation of these
multi-parts. The message key and indicator were supposed to be
different for each part. Very often they were obviously related, the
indicator of the next part being taken as the wheels end position for
the previous part.
2.4 A self inflicted vulnerability. This may have been suggested by
Scherbius, and involved enciphering the message key twice in succession
starting at the indicator position. The reason for this was to ensure
that the message key was correctly received. If the same three letters
came out twice from the decrypted six cipher letters, then this confirmed
the message key. This was supposed to mitigate against corruption due
to bad Radio transmission of the Morse code.
2.5 Allowing message plain text to be guessed. In the UK an inferred
plain text was known as a "crib". (In modern cryptography, "known plain text".)
The German message structures were very poor,often starting with an originator,
such as VONVONBDUU and ending with a signature or vice versa. First three
letters ANX, "to" space occured regularly early on.
2.6 A general vulnerability of cipher systems is letter frequency.
This arises because letter frequencies in natural language are far from
random, in English and German "E" is predominant. When a message is
enciphered, the distribution of the resulting cipher letters is nearly
flat random. Thus the distribution of letters resulting from a decryption
can be used to determine if the correct decryption configuration has been
found, and this test can be mechanised thus allowing rapid testing of
configurations.
3. Exploitation of vulnerabilities.
3.1 Cipher text only attacks.
The only realy successful attacks were by Marian Rejewski and Henryk
Zygalski, the famous Polish mathematicians and code breakers.
Marian Rejewski attacked the repetition of the
message key and worked out a method involving "characteristics".
Click here to read details of his method.
It depended on all message key encipherments starting from the same
indicator position.
In September 1938 the Germans changed their procedure to different
indicators for each message. Rejewski's characteristics no longer workd.
Henryk Zygalski came to the rescue with his netz or grill method.
Click here for more details of Zygalski's Sheets.
Alan Turing, in 1939, proposed a letter frequency attack using what he
called the "E" rack. I don't think it was ever built and would only have
worked with known, or a very small number of unknown stecker plugs. However
you can now read about my investigations into
the "E" rack.
3.2 Known plain text attacks. (Cribs).
Cribs were the basis of nearly all the British attacks. First the
crib, the guessed German text, was written out letter by letter
underneath the intercepted cipher text. If there were any "clashes",
positions where there was the same letter in cipher and crib, the crib
was not in the correct position and could be slid under the cipher
text until no clashes occured.
Very early on no messages had been broken so the only source of cribs
was capture or stealing of cipher/plain text pairs. Once breaking had
started, message analysis could rapidly build up expected cribs, an
example of success building on success.
Hugh Foss
devised a "Geometric Method" in 1927, which required a long crib.
Dilly Knox devised his "rods" method in 1936 or 37, which also required
a crib. Click here for a more detailed
description of Dilly Knox's Rods.
However, it was Alan Turing who devised the most powerful tools for
exploiting cribs. Click here for more details on
how Turing worked out his ideas for the Bombe.