Eduard Kontar discovered an error in the spectral response matrix for
the front segment of detector G9. The line due to k-shell photon escape
from the germanium was put in at 9.88 keV instead of
(photopeak energy - 9.88 keV). This was a mistake I made and fixed long
before launch, but apparently I failed to propagate the fix to
G9. The false peak at 9.88 keV should only be relevant if
1) You are using the full (non-diagonal) response matrix, and
2) The spectrum is very hard from 10-30 keV (for instance because
one or both of the attenuators are in place)
The problem is of course diluted to the extent that other detectors
besides G9 are used at the same time.
As of this mailing the problem has been fixed.