During the observations of XY UMa in this work,
times of four primary and two secondary minima were
obtained and are given in Table 2 (click here). The
values were computed using the following light elements
given by Geyer et al. (1955):
H JD  Min  Filter  
2400000.+  
49433.5427  I  B  0.0088  0.0001 
49433.5436  I  V  0.0097  0.0008 
49436.4161  I  B  0.0082  0.0007 
49436.4175  I  V  0.0096  0.0007 
49443.393  II  V  0.0397  ...... 
49443.395  II  B  0.0417  ...... 

Figure 3: The versus epoch numbers diagram for XY UMa
The diagram of XY UMa is shown in Fig. 3 (click here). This diagram contains other times of minima which can be found from the literature: Apart from our minima we used all visiual, photographic, and photoelectric minima from the list compiled by Pojmanski & Geyer (1990), and BBSAG observers, and Hanzl (1991), and two photoelectric minima observed recently by Jeffries et al. (1995). deviations in Fig. 3 (click here) were calculated with the ephemeris given in formula (1). The arbitrary weights of the points were chosen from Pojmanski & Geyer (1990). So, small points on this diagram represent uncertain data of weight 0. Photographic and photoelectric data are represented by filled dots of sizes proportional to their weights. Also, asterisks on this diagram represented the residuals of secondary minima. As it is clearly seen from the figure, the orbital period of XY UMa changes with time. However, as it was stated by Pojmanski & Geyer (1990), the asymmetry seen in the vicinity of the secondary minimum affects the precise determination of the times of secondary minima, and accordingly, in the period analysis of XY UMa only the residuals pertaining to the times of primary minima were used as it was the case with earlier works.
In order to calculate the phases of the observations of XY UMa,
the light elements of the system have been derived by using
the photoelectric primary minima times with E > 17000
cycles (see Fig. 3 (click here)) as;
with the weighted least squares method. In the light curves
which were formed using these phases, primary minimum
coincides with the phase 0.0 (see Fig. 1 (click here)).
The diagram of XY UMa in Fig. 3 (click here) indicates that the orbital period of the system is changing in a sinusoidal character. There are two mechanisms which explain such periodic orbital period variations of eclipsing binary systems: (i) a third body, and (ii) apsidal motion. Apsidal motion is unlikely in the case of XY UMa, since the orbital eccentricity of XY UMa is too small, and the residuals of primary and secondary minima times do not give a sinusoidal curve caused by apsidal motion (i.e, times of secondary minimum of XY UMa in Fig. 3 (click here) do not trace out a sinusoidal curve of equal amplitude but exactly out of phase to a sinusoidal curve traced out by times of primary minimum). Therefore, the third body hypothesis can explain the sinusoidal variation of the orbital period of XY UMa. However, if this sinusoidal variation character is, in fact, shape of the orbital period modulation, the Applegate mechanism can explain orbital period modulation of XY UMa. We shall investigate these suggested two models in turn.
Such an curve in Fig. 3 (click here) could easily be represented
with a lighttime effect which results from orbiting around
a third body. According to third body hypothesis,
time delay which are resulting from the movement on the
eccentric three body orbit may be calculated as follows (Irwin 1959):
where a sini is the projected distance of the thirdbody
to the eclipsing pair (in km), c is the speed of light (in km/day).
e is the eccentricity; v is true anomaly;
is the longitude of periastron passage of
the orbit of threebody system, while is the epoch of periastron
passage and is the period of the orbit of threebody system.
The ephemeris for the minimum is now given by:
Unfortunately, the theoretical curve given this formula
do not fit to the old photographic data of Geyer (1977).
Therefore, a quadratic term (= ) was added to
the Eq. (4):
where Q is the coefficient of the quadratic term and
gives change (increase or decrease) of the orbital period
of the eclipsing binary (in day). Finally, using Eq. (5)
to observed minima times of XY UMa, we have
performed the weighted regression method with
differential correction for T_{0}, P, Q, i, e,
, ,
and parameters which are given in Table 3 (click here).
Parameter  Value  Standart deviation 
T_{0} HJD  35216.5039  0.0013 
P (day)  0.47899433  1.30 10^{7} 
Q (day)  1.85 10^{11}  3.38 10^{14} 
i (km)  
e  0.41  0.09 
12  
(cycles)  16205  3739 
(cycles)  24804  1168 
Figure 4: The versus epoch numbers diagram for XY UMa.
The residuals were calculated according to
linear part of the ephemeris given in Table 3 (click here).
The continuous line represents
an appropriate fit to the data
The Q parameter in Table 3 (click here) shows that the orbital period of XY
UMa is secularly increasing. The period increase is found
to be about s per century.
The mass transfer and mass loss phenomena can
be held responsible for the secular increase seen in
the orbital period of the system. Furthermore, existence of
MgII emission lines which is believed to be originated from
the circumbinary environment (Gurzadyan & Cholakyan 1995)
supports this proposal. The primary component of XY UMa is close to
fill out its Roche lobe (see Sect. 4). So, this primary component may be
responsible for the mass loss.
The orbital period increase of XY UMa and the
mass of the mass losing primary component greater than
the mass of other component indicates that the only case IV
from massloss cases discussed by Singh & Chaubey (1986)
is possible to XY UMa. The mass loss in the case IV
in which the ejected material escapes out from
the binary system may be calculated as follows:
Using Eq. (6) to m_{1}=0.90 and
m_{2}=0.75 which are given in Sect. 5,
we have calculated
for the mass loss rate for primary component to
observed value .
The residuals of all observed minima times of XY UMa which were calculated according to the linear part of the ephemeris given in Table 3 (click here) are plotted versus epoch numbers in Fig. 4 (click here). In this diagram the theoretical curve of the lighttime effect is also plotted.
The parameters given in Table 3 (click here) show that the eclipsing pair revolve around a center of mass of the three body system with a period of 32.5 yr. The projected distance of the thirdbody to the XY UMa is 1.487 AU. These values lead to a small mass function of f(m_{3}) = 0.0031 for the hypothetical thirdbody. So, the masses of the thirdbody were computed for different values of the inclination of threebody system and are given in Table 4 (click here). In this computation m_{1}=0.90 and m_{2}=0.75 which are given in Sect. 5 were applied.
Inclination  m_{3} () 
0.483  
0.361  
0.297  
0.259  
0.237  
0.225  
0.221  
According to Table 4 (click here) the minimum mass of the thirdbody was found to be 0.22 . This value is different from that found by Pojmanski & Geyer (1990) for thirdbody hypothesis on XY UMa. The cause of this difference is that Pojmanski and Geyer have been fixed the orbital period on the least square fitting process for the lighttime effect since the earlier photographic data could not be fit. Instead of this artificial method applied by Pojmanski and Geyer on analysis of XY UMa, in this work, a quadratic term was added to Eq. (4) given for lighttime effect and all the parameters were left free on fitting process. Such a method could be considered more reliable since it does not include an artificial effect in the process of fitting a theoretical curve to data.
If the relative orbital inclination of three body system is equal to orbital inclination of XY UMa, the semimajor axis a_{3} of the thirdbody orbit around the center of mass of the triple system is about 11.2 AU. This value shows that the thirdbody, if it exists, revolves around much beyond the outer Lagrangian points of XY UMa and its orbit should be stable.
If the thirdbody is a mainsequence star, the massluminosity function () for the mainsequence stars with given by Demircan and Kahraman (1991) gives the bolometric absolute magnitude of the thirdbody . Whereas, the bolometric absolute magnitude of XY UMa is obtained to be about in the photometric analysis in this work. So, the thirdbody is more than about fainter than XY UMa, and then the lines of the third body on the spectrum of XY UMa are unobserved. If one considers the distance to XY UMa as 100 pc given by Dempsey et al. (1993), the separation between the tertiary body and the eclipsing pair may be found as . As a result, either the faintness of the thirdbody in the system or its small separation with the eclipsing pair makes its direct observation with the current equipment almost impossible, leaving the thirdbody hypothesis in dark.
Zeilik (1997) has found the evidence for a third light from another star in his recent photometric analysis. He stated that if the maculation effect is large, and centered around zero degrees, the timing may be in error. So, he found that one have to use clean light curves to get the correct times for the primary minima in such cases. But, the main problem is, in his work, to correctly determine the effects distorted the light curves, and then to correctly clean the distributive effects from the light curves. It can be clearly seen that the probable errors in the cleaning processes will damage both the naturalness of the light curves and the normal light curves.
At this stage of the orbital period analysis, the cause of the orbital
period change in XY UMa is thought to be caused by
the period modulation of the sine variation.
Therefore, the sinusoid with amplitude , period and
a moment of minimum was added to the ephemeris (T_{0} and P)
to get the best fit to the data:
However, a quadratic term (= ) was added to
this equation since the earlier photographic data
could not be fit. So, using the sine equation
with a quadratic term to all observed times of minima of XY UMa,
we have performed the weighted regression method
with differential correction for
T_{0}, P, Q, , and parameters.
The parameters found are given in Table 5 (click here).
Parameter  Value  Standart deviation 
T_{0} HJD  35216.5021  0.0011 
P (days)  0.47899433  
Q (days)  
(days)  0.0079  
(cycles)  12953  190 
(cycles)  23713  615 

Figure 5: The versus epoch numbers diagram for XY UMa.
The residuals were calculated according to linear
part of the ephemeris given in Table 5 (click here).
The continuous line represents
an appropriate fit to the data
According to this Q parameter, the increase in the orbital period was found to be s per century. This increase in the orbital period is related to the case IV from mass loss cases discussed by Singh & Chaubey (1986). Accordingly, the mass loss rate for the primary component is estimated to be .
The residuals of all observed minima times of XY UMa which were calculated according to the linear part of the ephemeris given in Table 5 (click here) are plotted versus epoch numbers in Fig. 5 (click here) together with the theoretical curve of the sine equation with a quadratic term.
The primary component of XY UMa is a RS CVn type star with a surface magnetic activity. The starspot activity is evident from the light curve analysis in this work (see Sect. 4). So, for the cause of the orbital period changes seen in the system, the activity in the primary component might be taken responsible and the Applegate gravitational coupling mechanism (Applegate 1992) can explain the sinusoidal variation seen in Fig. 5 (click here).
The Applegate mechanism requires that the active star be variable, and that the period of the luminosity variation be the same as the period of the orbital period modulation. Applegate (1992) explains that maximum luminosity should coincide with an curve minimum if the star's outside spins faster than its inside but should coincide with an curve maximum if the outside spins slower than the inside. In the case of XY UMa, Geyer (1976) has presented that the average brightness of the binary system changed between 1955 and 1975 in a sinusoidal manner by 0.18 mag in V and 0.20 mag in B, indicating a periodic variation of about 28 to 30 years, and that the lowest system brightness was observed in 1961, and the highest in 1975. According to Geyer's paper, the mean brightness and the orbital period of XY UMa vary with the same cycle length, years. But, the maximum system brightness in 1975 or minimum system brightness in 1961 does not coincide with the minimum or maximum of curve in Fig. 5 (click here). However, due to the big asymmetries in the light curves of the system and insufficient data, this conclusion could not be taken as final. Only systematic and continuous observations may help to decide on the existence of any connection between the mean brightness and period variations.
Here, we use Applegate's formalization in an effort to examine the orbital period changes of XY UMa. We assume that the characteristics of the active component of the system are: , , and given in Sect. 5.
(i) The diagram of XY UMa in Fig. 5 (click here) shows a modulation with a semiamplitude of day and a modulation period of yrs. This gives the observed orbital period change of 0.181 s. (ii) The angular momentum transfer, which produces the 0.181 second period change, has been computed to be . (iii) The moment of inertia of the shell has been computed to be by assuming the mass of the shell is onetenth of the active component's mass. (iv) The variable part of the differential rotation is . (v) The energy required to transfer the angular momentum is ergs. (vi) If the energy requirement is supplied by the nuclear luminosity of the star with no energy storage in the convection zone, the star will be variable with the RMS luminosity variation of which give rise to only about mag light variation of the system. (vii) The mean subsurface magnetic field which is provide the RMS torque required to periodically exchange between the outer shell and inner part of the star is obtained to be about 10 k Gauss.
If the hypothesized magnetic cycle can be estimated from an diagram as the time between one period decrease and the next (Hall 1990), the diagram for XY UMa in Fig. 5 (click here) gives the magnetic cycle of the primary component as 31.1 yrs. This value agrees with the magnetic cycle for solartype dwarfs computed by Hall (1990). Finally, the Applegate model gives reasonable solution to the orbital perid modulation of XY UMa which is caused by the magnetic activity of the primary component.