From Marc: I am quite surprised at the different count: as you say there was no decision to normalise, and I am supposed to have checked all of tcn at one point so I can't see how so many zs could have been thrown in later. Has the difference been checked against the manuscript, at least on a selection, just to make sure it's not a problem in tc? The lack of matching is not at all unusual, since in many cases the difference would not be audible. The ap's ability is very good, but with ordinary carelessness about such matters.
PHS: Action item: sample some of the cases of added zs in the tc/tcn against the facsimile and each other