Period determination for short period stars is performed by observing target fields several times per night for several consecutive, or closely spaced, nights. Data are reduced to instrumental magnitudes in a similar manner to that described in Section 2.2. First the individual frames are debiased and flat--fielded, then fluxes within a four arcsecond aperture around each star (targets and randomly chosen background stars) are extracted. The nominal seeing was about 1.5 arcseconds at both the Wise observatory and at Lowell. An annulus with an inner radius of five arcseconds and an outer radius of seven arcseconds is used for background subtraction. The relative fluxes are put on a magnitude scale with the zero--point set to 25. Since the goal of the program is differential photometry, nightly airmass and color solutions are not needed. In any event, the sky conditions under which these data were taken were not photometric.
After instrumental magnitudes were calculated for the stars observed during the run, several comparison stars are chosen. These comparison stars were then compared among themselves, and the four most stable in each field were chosen as field standards. Differential magnitudes were then measured between the target stars each field standard star. For any single target star, there are usually three to four observations per night and four field standard stars per observation. The longest continuous run was at the Wise Observatory. It contained 11 nights of data. Counting each comparison star separately and removing unusable frames, each target had about 100 data points per filter. Observations made from MTSB had many more observations.
The first step in period searches involves trimming highly discrepant points, those that vary from the mean magnitude by more than six standard deviations. Normally, one or two highly deviant points were removed this way. Usually, such deviations occurred to many objects in the same field and were caused by either filter misalignment or terrestrial clouds. Then, two separate numerical period searching routines are used. One is a phase dispersion minimization (PDM) method (Stellingwerf 1978) and periodogram (Scargle 1982).