## 3. Orbital period changes of AW UMa## 3.1. Times of minimum lightSeveral UBV photometric observations of AW UMa were taken during eclipses. These observations were analyzed using the procedure introduced by Kwee & van Woerden (1956) to determine the times of minima. We have determined the times of 13 primary and 11 secondary minima. Most of them (20) were already published (Pribulla et al., 1997). The heliocentric times of our new 4 minima of AW UMa together with all available photoelectric minima from the literature are listed in Table 3.
Our UBV photometry shows that while the primary minima of AW UMa occur almost simultaneously in all three passbands, the secondary minima in the U passband occur later than in the B and V passbands. The mean delay of the times of secondary minimum in the U passband in comparison with the average times of the secondary minima in the B and V passbands is (96 22) s. We have used the minima from Table 3 (except last 9) to find the systematic delay of the secondary minima with respect to their expected position in phase 0.5 given by primary minima. The resulting delay = (54 27) s is even larger than the = 36 s found by Demircan et al. (1992) and incompatible with an explanation by the internal light-time effect (Liu et al., 1990). Another phenomenon must be responsible for the shift. The large error in its determination suggests that the phenomenon is highly variable. ## 3.2. O-C diagram and its explanationWoodward et al. (1980) found that the orbital period of AW UMa decreases. According to Hrivnak (1982), the observed change can be either sudden or continuous and can be caused by mass transfer from the more massive to the less massive component. Altogether, we have used 108 minima (given in Table 3) to study the period change. The times of 9 minima deviated too much from the general trend (listed at the end of Table 3) and were therefore excluded from this study.
A linear ephemeris (2) was applied to construct the diagram (Fig. 2). The long-term period decrease can be explained as: -
for 11495 24621 we have: and finally for 24621 we have: The sum of the squares of residuals for the three linear fits is 0.00031816 d*Two sudden period changes*. The first jump was detected by Woodward et al. (1980). The second possible jump was detected by the authors (Pribulla et al., 1997). Fitting the data by a broken-line, we derived three linear ephemerides. The ephemeris valid for 10495 is:^{2}. The first sudden period change = (7.30.2) 10^{-6}occurred at JD 2 442 650105; the second one = (7.20.7) 10^{-6}at JD 2 448 850115. The time interval between two possible sudden period changes was 6200150 days (= 17 years). -
*Continuous period change*. A parabolic fit provides the following ephemeris:The sum of the squares of residuals is 0.00053518 d ^{2}. The total change of the period is = (1.420.06) 10^{-5}. While the largest residuals from the quadratic fit occurred in 1993-4, the activity of the system reached its maximum four years earlier (see Sect. 5.2). Therefore, the increase of the residuals was not caused only by the activity of the system. -
where sin*Continuous period change and the light-time effect*. The parabolic fit to the (O-C) data (continuous period change) shows its greatest residuals at the two times cited in case (1) above as possible sudden period changes. These residuals can be reduced by invoking a light-time effect due to an additional body in the system, superimposed on the continuous period change. The times of minima can be computed as follows:*i*is the projected semi-major axis,*e*is the eccentricity, is the longitude of the periastron, is the true anomaly. is the quadratic ephemeris of the minima in an eclipsing binary and*c*is the velocity of the light.A preliminary period of the light-time effect was found by Fourier period analysis of residuals from the quadratic fit as = 65001200 days. The parameters of the fit obtained by simultaneous fitting of the light-time effect and parabolic ephemeris obtained by the differential corrections method are given in Fig. 2. The sum of squares of residuals is 0.00033223 d ^{2}. The period of the body = 6250340 nearly coincides with the time interval between possible jumps (case 1). The body responsible for such a light-time effect would cause the systemic velocity to change with an amplitude of 0.9 km s^{-1}, which is much smaller than the observed one (see Sect. 7).
To summarize, the data can be interpreted as two sudden period changes or as one continuous period change (quadratic fit). In the latter case, the residuals show a systematic effect which can be interpreted as the light-time effect due to another body. Although the sum of the residuals is smaller in the first case, the resulting fits for both cases differ only slightly (see Fig. 2). The present data are not sufficient to decide which interpretation is correct. © European Southern Observatory (ESO) 1999 Online publication: April 12, 1999 |