I found a bug in the "fsg2eva" script, which I had used back in 
November to convert Landini's interlinear to EVA.  The "IK"
ending (but not "K", "IIIK", and "IIK") was erroneously
translated to "ik" instead of "im".  

Fortunately that group occurs only 79 times in the whole interlinear file;
discounting multiple transcriptions, that means about 50 occurrences in the text.
So, most of the statistics I computed since then are still valid.

In particular, for the QOKOKOKO note, the main consequences of fixing
that error are adding "im" to the list "K*" of significant elements, and 
lowering the exception rate in Friedman's text from ? to ?.