2. Period verification and refinement
The time baseline covered by the OGLE monitoring is around 120 days (i.e. a single observing season) for most of the stars, but in 9 cases the available data extend over two seasons. In this case, an improvement of the goodness of the fit could be obtained by calculating a solution for each season and then aligning the mean magnitudes (this procedure was applied to the measurements of OGLEGC 3, 4, 5). As a matter of fact, shifts up to 0.048 mag were observed in Field 5139BC, which are surely due to observational or instrumental problems. In two cases (OGLEGC 42, 45), we did not consider the data obtained in one season, as they were a small part of the total and probably affected by a misalignment which was difficult to quantify. In the remaining four cases (OGLEGC 9, 29, 38, 59) the procedure of the re-alignment did not introduce appreciable effects on the fit.
We made an independent period search. Since the baseline and the number of measurements were appropriate, all the values previously known were confirmed. Only the case of OGLEGC 34 deserves some comment. Kaluzny et al. (1996) suspected a double-mode nature on the basis of the period search carried out with the CLEAN algorithm. We performed the frequency search by using the least-squares iterative method (Vanicek 1971) and we obtained the power spectra shown in Fig. 1 (upper panel). The peak at f=26.1611 cd-1 is the highest, but the difference with respect to the alias at 25.1611 cd-1 is very small. When introducing f as a known constituent, the power spectrum did not reveal any significant feature in the range that would be expected for a second period (lower panel). The CLEAN algorithm is probably responsible for the result quoted by Kaluzny et al. (1996): because it cannot match the odd noise distribution, the signal is spread at different peaks. OGLEGC 34 is probably monoperiodic, but the period is uncertain and may be either one of the two values reported above; we have a slight preference for f=26.1611 cd-1 because it gives a better fit and a better residual power spectrum. Note also in the lower panel of Fig. 1 the increasing noise at very low frequencies, the fingerprint of night-to-night misalignments.
© European Southern Observatory (ESO) 1999
Online publication: May 21, 1999