Very impressive. Again, this was from the novellas, so you can look it up in there if anything does't quite fit. (Like "asbal last" should be "as balast", or where additional punctuation goes. But then, you probably can't get that quite right without the original.)
Letter frequencies was exactly the right way to start. There was a little more than just single letter frequencies however. The "e" thing may have been a bit confusing with this one. (For some reason I though I'd chosen the text so "e" was the most common letter). Anyways, it was only the second most common letter, so hopefully it didn't throw you off too much. I did however make sure the digraph and trigraph frequencies were helpful. Although, I doubt it occured to anyone to use them.
If you scan for the most common 2 and 3 letter sequences, you'll find that "TS" and "TSL" were the most common. These are expected to be "th" and "the". Even with the spacing and punctuation removed, you can still expect fairly good chances of this happening. Anyways, there were 34 occurances of "TSL" in the ciphertext, and the next most common trigraph only occured 14 times. (Which is about equivalent to a slap to the face
). Also, "TS" occured 60 times, and the next most common digraph occured 57 times (and it was "SL"). Given that and "L" was the second most common letter, hopefully it wouldn't have been too hard to guess that "e" was "L" and not "T".
Also, it would have been important to look for repeated letters. Some of the most common repeated letters were "AA" (14 occurances), "TT", and "KK" (both at 13 occurances).
Well, I'd hoped that anyone looking at trigraphs and digraphs would have been able to guess the letters used for "the" fairly easily. It got a bit harder after that, but once a few more letters are filled in, it should have gotten easier. But again, good job on this one. Did you really break it using only single letter frequencies?
Another one?