Key length in an old cypher

Jul 19 01:29 UTC 2007

In cryptanalyzing a text encoded with the Vigenere cypher, if the key is
as long as the text to be encyphered, and presuming that the key is
sufficiently tough, would the the encoded text be discoverable
noticeably faster than a search through the keyspace? If the key had a
different letter density distribution than the original text, would
there be any way of telling the encyphered text from a random list of
letters? As an example, if I used a page of a Polish novel as a key to
encypher a page of English (Polish favours consonants and consonant
groupings far more heavily than English, and uses pairs and triplets of
consonants not often seen in English), would the resulting letter-stream
be easily attacked statistically? If the adversary did not know that the
key was Polish and did not have a copy of the book from which the page
was pulled, would it be distinguishable from a letter-stream encyphered
with an OTP or similar? If the example of Polish does not sit well with
you, consider alternately a Welsh text (with its disproportionately
large streams of vowels). 

At the moment, this is simply a Gedanken experiment, but it's been
playing around my br@n3 for a couple of days.



Thanks

response 2 of 2:

Jul 26 00:26 UTC 2007

The complexity of which gives the non-speaker an impression of
randomness, even though the language adheres to well-defined (albeit
frightfully complex) rules, making it a strong candidate for a
crypto-key, while maintaining an encoding for the key that can be
carried and transmitted out-of-band in the form of a text (maybe a
trashy romance novel).