An interesting intermediate result after one night of number crunching:
The program is not finished yet, but there is definitely a structure in the results.
I don't really know what I'm seeing :-)
It might be he sections, of the Currier languages, who knows ...
I've used this input file:
It is a selected portion (almost all) of the Takahashi transcription
with the descriptors removed. (Knox, can I publish the link?)