From: Felix Reuthner on
Hi,
a question about code-breaking in WWII:
Correct me If I'm wrong, but basically it worked like that: German radio
traffic was encrypted wit ENIGMA machines and settings that were changed
daily. In Bletchley Park, they had a nifty machine (or many) that could
go trough all possible settings (usually within a few hours), so it
could find the correct setting for the day. After that, decrypting all
German radio traffic for the given day was trivial.

Now the 1000$ question: How did the code-breakers identify the correctly
decrypted text among the zillions of garbled attempts? Did they have the
possibility to check for sequences that usually appeared in a
transmission? IIRC, there actually were attempts to get the Germans to
send specific texts, but my memory is a bit hazy there. Possibly that's
only from a novel. For example, if five enemy bombers are observed
dropping mines in a specific location, the British could guess what the
German radio operator would report. But I would be surprised if that
analysis could have been automated back in WW2.
So, how did they do it?

Felix

From: robertwessel2 on
On Nov 18, 5:35 pm, Felix Reuthner <s...(a)reuthner.net> wrote:
> Hi,
> a question about code-breaking in WWII:
> Correct me If I'm wrong, but basically it worked like that: German radio
> traffic was encrypted wit ENIGMA machines and settings that were changed
> daily. In Bletchley Park, they had a nifty machine (or many) that could
> go trough all possible settings (usually within a few hours), so it
> could find the correct setting for the day. After that, decrypting all
> German radio traffic for the given day was trivial.
>
> Now the 1000$ question: How did the code-breakers identify the correctly
> decrypted text among the zillions of garbled attempts? Did they have the
> possibility to check for sequences that usually appeared in a
> transmission? IIRC, there actually were attempts to get the Germans to
> send specific texts, but my memory is a bit hazy there. Possibly that's
> only from a novel. For example, if five enemy bombers are observed
> dropping mines in a specific location, the British could guess what the
> German radio operator would report. But I would be surprised if that
> analysis could have been automated back in WW2.
> So, how did they do it?


In fact, most of the Enigma decrypts were some form of known plaintext
attack. These texts were often called "cribs" and were collected from
a variety of sources, including regular broadcasts from weather
stations, assumptions about routine status messages ("nothing going
on"), as well as deliberate efforts to get a known message sent (for
example the deliberate mining of certain locations), and very
extensive traffic analysis. In some cases the plaintext came from
messages that were found to have been sent more than once - once in a
low grade code (for example weather reports).

They used machines ("Bombes") to step through vast numbers of
combinations, combining cryptanalysis and brute force.

The code breaking effort was a vast effort, and had considerable
resources.

The Wikipedia article is a decent summary:

http://en.wikipedia.org/wiki/Cryptanalysis_of_the_Enigma

http://en.wikipedia.org/wiki/Bombe

From: Gordon Burditt on
>a question about code-breaking in WWII:
>Correct me If I'm wrong, but basically it worked like that: German radio
>traffic was encrypted wit ENIGMA machines and settings that were changed
>daily. In Bletchley Park, they had a nifty machine (or many) that could
>go trough all possible settings (usually within a few hours), so it
>could find the correct setting for the day. After that, decrypting all
>German radio traffic for the given day was trivial.
>
>Now the 1000$ question: How did the code-breakers identify the correctly
>decrypted text among the zillions of garbled attempts? Did they have the
>possibility to check for sequences that usually appeared in a
>transmission? IIRC, there actually were attempts to get the Germans to
>send specific texts, but my memory is a bit hazy there. Possibly that's
>only from a novel. For example, if five enemy bombers are observed
>dropping mines in a specific location, the British could guess what the
>German radio operator would report. But I would be surprised if that
>analysis could have been automated back in WW2.
>So, how did they do it?

Boilerplate in messages is one way to verify correct decryption, at least
for a first pass. Anything that looked real after this could be looked at
by humans. For example, if I were trying to decrypt email messages,
I'd look for strings like:
"\nFrom: "
"\nTo: "
"\nSubject: "
"\nDate: "
in the first few hundred bytes of the message. Further, you can guess a
fair portion of the Date: header, and there will be common repetitions
of sources and destinations of messages.

You might also get used to the format of messages sent, for example,
a message sent at 5PM from a particular station on a particular frequency
might always have strings like:

Subject: Daily Weather Report

and another station might usually have:

From: Col. Wilhelm Klink, Commandant, Stalag 13
To: General Berkhalter
Subject: My Perfect No-Escape Record

in the headers for about half of the messages. There might be
standard headers indicating the origin of the message is the central
command. Long-winded but standard ways of saying "This message is
classified top secret" might also be useful bits of known plaintext.

It is also possible (I'm not familiar with World War II message
traffic) that every message had "Heil Hitler" after the header.
That could be a big weakness.


I have heard about a few attempts to get the Germans and/or Japanese
to send particular messages, often to break a code-within-a-code.
Things like map coordinates, bases, or unit designations might be
coded this way. For example, the plaintext might refer to Airbase
Delta Gamma, but you don't know which airbase that is, so you pick
one, overfly it, and see which airbase is reporting spotting enemy
planes.

I seem to recall identification of a code for a particular Japanese
base/island in the Pacific was made by doing something to the water
tower on the island suspected to be the one in question, then noting
that it reported a water shortage. I think I heard this on a PBS
documentary.

From: wjhopwood on
On Nov 18, 6:35 pm, Felix Reuthner wrote:
> Hi,
> a question about code-breaking in WWII:
>..... How did the code-breakers identify the correctly
> decrypted text among the zillions of garbled attempts?

Simply put, when the text of the message made sense
they knew they had broken it.

> Did they have the possibility to check for sequences
> that usually appeared in a transmission?

In essence, yes. Successful decoding depended on
finding the correct key to the Enigma wheel settings
for that message and probably all the other messages
transmitted that day..

Finding the key rested in large part on a certain amount of
tedious trial and error as well as guesswork coupled with
having the known identity of the sender and the addressee.

>From this information, an intelligent assumption
could follow as to what might constitute the basic theme
of the message and possible words and phrases which might
be used within that theme.

Frequently such sequences might be found at the
beginning and end of the message, the portions which would
contain the name(s) of the addressee(s) and of the sender.
These might be already known through signal intelligence which,
from call signs, direction finding, and traffic analysis produced
fairly accurate information regarding the identity of the sender
as well as to whom the message was being sent.

German operators sometimes got careless by using stock
words and phrases in the opening and closing sentences of the message,
those buffer phrases unrelated to the basic text
and inserted in an attempt to decoy enemy decoders. Such
opening phrases as "Heil Hitler" and the like were often found
in such locations and were of considerable aid to decoders.

Another decoding aid was an idiosyncracy of the Enigma machine which
kept it from ever substituting the same letter for itself.
(If the letter "H" was used in the real text it would never come out
an "H" in the coded text and the same with all the other letters)
This was of considerable assistance to the code breakers.

WJH

From: Rich Rostrom on
On Nov 18, 10:50 pm, gordonb.4t...(a)burditt.org (Gordon Burditt) wrote:

> I seem to recall identification of a code for a particular Japanese
> base/island in the Pacific was made by doing something to the water
> tower on the island suspected to be the one in question, then noting
> that it reported a water shortage. I think I heard this on a PBS
> documentary.

The incident you are thinking
of took place before the battle
of Midway. U.S. codebreakers were
at that time reading much of the
Japanese military cipher traffic,
and the intelligence analysts
believed they had uncovered plans
for an attack on Midway. However,
the messaged did not refer to Midway
by name, but to the code name "AO".

To make certain that "AO" stood for
Midway, the Navy had Midway report
that its water distillation plant
had broken down. This report was
sent by radio "in clear", and was
picked up by the Japanese. Shortly
afterwards, a Japanese dispatch
stated that "AO" was short of
fresh water. Bingo!

(The message directing Midway to
make this false report was sent
via the telegraph cable from
Hawaii to Midway, which was intact
and of couse secure. BTW, it was
part of the trans-Pacific telegraph
cable, and the next link - from
Midway to Wake Island - was also
intact, even though Wake had been
occupied by the Japanese. Of course
this cable went unused during the
war, except for an occasional
obscenity or insult.)