Looks like a trimmed jlap resets to a 0 iv aka the hash in the first line of the file. It would be nice to retain the nonzero iv, so that the overlapping hashes lined up.
Probably comes from json2jlap and not the newer trim code. (trim; check iv; update trimmed file; check iv again)