# Decrypting transposition ciphers

How do I see if the following ciphertext eaeairtntrnaeemtve is a transposition cipher, using letter frequency?

This article suggest how to detect by using letter frequency detection and cryptanalysis but I am unsure of how to do this. Any suggestions?

-
This was simultaneously cross-posted on Crypto.SE and this site (crypto.stackexchange.com/q/12689/351). Please don't cross-post. That is prohibited by StackExchange rules. In the future, if you discover that you posted on the wrong site, you can click the flag link to flag your post for moderator attention and ask them to migrate it, but you should never cross-post. –  D.W. Jan 3 at 8:21
@DavidRicherby, nope! Look again. The cross-post on Crypto.SE was posted 2014-01-02 15:15:03Z by user10877 -- i.e., yesterday, the same day as this was posted. It was subsequently marked as a duplicate of crypto.stackexchange.com/q/3826/351, which was posted on Sept 18 2012 -- but I'm referring to the post by user10877. Perhaps your browser is auto-redirecting you to the dup, so make sure to look at the post by user10877, not the one it was marked as a dup of. –  D.W. Jan 3 at 19:43
@DavidRicherby, regarding your comment about the questions being similar, make sure you look at the original revision. Incidentally, the wording of the two posts is identical. The Crypto.SE post was subsequently edited by Ilmari Karonen to improve the wording a bit, and this post was edited by Gilles after it was posted, so make sure to look at revision 1 of both posts: to see what I mean, compare crypto.stackexchange.com/revisions/12689/1 to cs.stackexchange.com/revisions/19453/1. –  D.W. Jan 3 at 19:44
@joker, if you didn't cross-post, how did these two posts end up with exactly identical wording? Compare the two "rev 1" versions I linked to, and notice how the wording is 100% word-for-word identical: not just similar, but identical. The timing of the two posts was also extremely similar: the Crypto one was posted on 2014-01-02 15:15:03Z, the CS one was posted on 2014-01-02 15:19:39Z (less than 5 minutes later). Got an explanation for that? –  D.W. Jan 3 at 19:47
@D.W. Ah, I see. Yes, I was being redirected to the duplicate; yes, the two original revisions you link to are identical (apart from capitalization of the title) and were posted within five minutes of each other. Good detective work! –  David Richerby Jan 3 at 22:24

It might be problematic to use letter frequency when you have short ciphertext.

Anyway, this is transposition cipher! so your plaintext is probably some permutation of the ciphertext.

You have $18 = 3 \cdot 6$ chars. Try to write them as a rectangle of $3 \times 6$ or $6 \times 3$ and read it in different ways (spiral, zigzag, etc.).

-

It is impossible to do cryptanalysis on such a short message. It could be anything, encrypted with anything. Even if you somehow knew that it's a substitution cypher, it could be any 18-character message where the right characters match (there must be seven different letters; characters 1, 3, 13, 14 and 18 have to be the same; characters 1 and 2 must be different, etc.).

If you had a longer cyphertext (say a few hundred characters), you'd just compute the frequency of the characters appearing in the cyphertext and compare that to the frequency of the letters in the language you think the message was written in. The frequencies in English are: e 12.7%, t 9.1%, a 8.2%, o 7.5%, i 7.0%, n 6.9%, etc. If a decently long cyphertext comes decently close to that distribution, you'd expect a transposition cypher. If you only have that the most common characters occur about 13%, 9%, 8%, 8%, 7%, 7%... of the time, you'd suspect a substitution cypher and you'd start guessing the letters based on their frequency. Note that the process is not automatic and you need to be flexible: for example, if the message is about the Zulus playing zithers in zoos in Zanzibar, the letter frequencies may be skewed quite some distance from the average over all English text.

-