I posted some info yesterday on reddit on my approach to solving the Z340 cipher.

Since you guys are pretty open-minded, I’ll tip my hand on a few additional thoughts I’m having. First, it really sticks out in my mind that there are 24 +’s and 12 B’s, the two most frequent symbols in a cipher by someone calling himself (herself?) "The Zodiac". How many signs in the Zodiac? 12. Now, if it were merely the case that there were 12 of the most frequent character in the cipher, that wouldn’t be a very remarkable fact in itself. But the +, in my hypothesis, is just a "masking" character that does more of what was done at the end of Z408 where you have multiple clear-text characters mapping to the same cipher-text character. Thus, the author of the ciphertexts had complete control of how many +’s and could easily control the number of any other cipher-text character – such as B – by altering the plain-text or altering the rate at which the plain-text character is enciphered into its homophones.

The cipher has proved quite frustrating in that it appears to have significant non-randomness, yet any "sensible" substitutions into these non-random bits yield gibberish everywhere else. For example, consider the ‘252’ – this could be any bit of English that goes "ERE", "EVE", "NIN", "DED" and so on. But I have not yet found a substitution of this form that does not lead to gibberish elsewhere in the cipher. To my mind, this makes it less likely that the Z340 is a homophonic cipher, applied just once. Perhaps the author constructed two keys and alternated. Perhaps the author applied one key to create an intermediate cipher, then applied a second key to create the final cipher. These seem to me to be the two most probable alternative hypotheses.

Also, I started looking at the idea that the cipher may be broken into parts. This would make sense given the author’s previous breaking up of the 408 into 3 parts. Taking the -…- line to be the imaginary break point, I noticed immediately some patterns in the most frequent cipher characters. + looks fairly evenly scattered, but B is bunched up at the bottom, while p is bunched up at the top, 5 is bunched at the bottom, c is bunched at the bottom, O is fairly evenly scattered, I is bunched at the bottom; F and the other less frequent characters all appear to be fairly evenly scattered. This is consistent with the hypothesis that the top and bottom halves are separately keyed, though maybe it’s only the very frequent letters (EATOIN) for which this was done.