The 18 orbits scheduled for this program executed on 28 October, 29 November, and 16 December, 1995. During each orbit, a single long integration (1800 sec) was taken while the satellite was in the Earth's shadow and one additional short integration (300-400 sec) was taken to fill the time between target acquisition and entry into the Earth's shadow. Two orbits per visit were spent observing with each of three WFPC2 filters (F300W, F555W, and F814W). The short exposures were combined and used to improve cosmic ray rejection. In all, five 1700 sec darks and ten bias images were taken during the bright portion of the six orbits comprising each visit. The gain setting for these data was 7e-/DN (bay 4) and the read noise was roughly 5.2e- for each of the four CCDs. The field of view of each WFPC2 image is roughly 4.4 arcmin2 (725 × 725 well-exposed pixels in each of the 3 WF CCDs, with 0.0996 arcsec2 per pixel).
4.2. Basic Data Reduction and Calibration
The reduction of these data involved a mixture of the standard
pipeline and our own methods. The pipeline procedures used included
the corrections for analog-to-digital conversion errors, overscan
subtraction, bias image subtraction, flat-fielding, and point-source
calibration. The accuracy of these steps is largely provided in STScI
documentation. Whenever possible, we have independently verified the
errors we quote. All steps in the reduction are indicated in the flow
chart in Figure 2 with associated
errors explicitly
included in our error budget. The reduction is discussed briefly
below; greater detail can be found in
Bernstein (1998).
The following
section should be read by those who are interested in the reduction
procedures and detailed error discussions. As the goal of these
observations is an absolute measurement of the total flux of the
night sky, zero point calibration is as critical to the accuracy of the
results as flux calibration. Table 3
summarizes the errors discussed below and indicates the dominant sources of
error.