Light pulses in the phosphor readout of the HUT detector have a decay time of 0.20 ms. Since the Reticon diode array is scanned once every 1.024 ms, a small fraction of pulses have persistence bright enough to be flagged as genuine events on the subsequent scan of the diode array. These double-counted events can be easily identified in high time-resolution mode. Measurement of the pulse persistence during observations of high count-rate targets on Astro-1 shows that statistically 92.6% of the recorded counts correspond to true incident photons. In the standard reduction of HUT data, we have chosen to fold this simple multiplicative correction to the count rates into the inverse sensitivity curve.
Since the HUT detector is photon counting, errors in the detected count rate
follow a Poisson distribution. To assign errors to the individual pixels in a
spectrum we use the square root of the detected number of photons.
It is in this error bar calculation that the pulse persistence correction
must be explicitly incorporated since the actual number of photons is only
92.6% of the number of recorded counts. Thus, the assigned error for a
pixel containing N counts is not 
, but 
.