Yes, 26/28 seemed unusual to me, too. All six numbers I gave you were
average values of a series of digitized files.  I can give you one example
from which you can see that the peaks are - unfortunately - no multiples and
therefore 'unnotchable', as you found out, too.  8.0/9.2, 16.5/17.8,
25.1/26.4. Also the peaks don't exhibit the same shapes and widths within
one file, which allows no uniform notching, too. Tricky thing.

When planning our series, we started out with the question ‘How does
treatment time affect the success of baking’?

In our experiments and with our tape, every treatment time (2-48h) stopped
the squeal, but the oxide losses seemed to occur randomly - this may be
related to the initial state of the tape at the very tape region. Now, this
is just our present guess and we wouldn't like to generalise on that nor
talk about it as proven evidence before we do more tests, and with other
tape types. But so far that left us very sceptical towards the baking
treatment. We are now trying to find financial support to further pursue the
study, and repeat the experiments with more tape types and also looking at
the other parameters.
As I said, our tape type is unclear, no marking on the back or leader and
the shedding has left sticky residues on the player heads, no powder.

We are also interested if your theory of a lowered glass transition
temperature is generally applicable. On how many sticky tapes have you
observed/measured that fallen Tg?

Best regards, Katrin