Kathy Flanagan notes (8/23/96):

        VERY interesting. I am trying to find out
if gaps could account for it. It is interesting that there
is a bump 100 microns away, because 107 microns away
SHOULD have been the location of the first order if
the period of the MEG is correct. It would be located to the
less-dispersion side (closer to zeroth order). It is easily
possible that incorrect de-gapping could create a bump there.
(By overcompensating for the gaps, you overlap events from
one tap on top of events from the next tap.) In fact, incorrect
degapping can change the apparent centroid. 

        If possible, we should ask the HSI to me moved a bit
in the dipesion direction, and the MEG remeasured. That would
immediately resolve the gap issue. (Actually, the un-degapped data
would probably tell immmediately if gaps are involved.)

        The HSI is using the ROSAT electronics, which have
typical gap sizes of 8 to 10 pixels, at 6.4 microns per pixel.
Thus, a gap of 60 microns would be normal.  The plates are only 18mm
in diameter, so that they can not see both first orders at the same 
time. It is important that they independently check the plate
scale by moving the HSI a known amount, and remeasure the
dispersed spectrum.  This is a great problem! We need to make sure we
have a map, in advance, of where the "gap" zones are for the
big calibration tests in the spring. They are always at half-tap
locations, so it is easy to predict trouble spots in advance and
move the detector to compensate!