Changes of the orbital periods are well documented for a large number of close eclipsing binaries of various types. There were developed several theories for their explanation which employ two categories of causes: the geometric and the intrinsic ones. The geometric ones comprise the apsidal motion and the presence of the third body (light-time effect, LITE). They require strict periodicity of the changes which course can be analytically expressed.
Nevertheless, courses of most period changes are non-periodic and cannot be explained by the geometric effects. Already Wood (1950) recognized that systems with large period changes contain at least one star which fills in its Roche lobe. As Huang (1963) showed period changes can be consequences of mass transfer in binary or mass loss from it. Also models of mass exchange in binary which were developed in sixties and early seventies seemed to offer a natural explanation for some period changes and showed the importance of the semi-detached configuration (e.g. Paczynski 1967; Plavec 1968).
Most following analyses were done for Algols, the typical representatives of the semi-detached binaries with evolved late-type star filling in its Roche lobe and transferring mass onto its companion, a main-sequence star of early spectral type. Analyses of the long series of observations revealed that period changes in a large part of Algols are much more complicated than the evolutionary schemes of mass transfer suggested and alternate course of variations is not exceptional (e.g. U Cep-Fig. 1 in Olson et al. 1981). Biermann & Hall (1973) presented a model which tried to explain this phenomenon. They assumed that the mass is transferred in bursts and the angular momentum is temporarily stored in the outer layer of the gainer, thus causing decrease of the period. In the subsequent epoch the increased turbulence in the gainer allows to return this angular momentum to the orbit and the orbital period increases again.
A significant change in opinions on the mechanisms of the period variations occurred in the eighties. An idea appeared that the mass transfer rate is too small to be important for the observed changes. Matese & Whitmire (1983) described a general mechanism which explains the observed period variations by changes of the stellar structure. Hall (1989) presented a correlation between spectral types of the binary components and character of the period changes. His statistics showed that binaries which contain at least one star of spectral type F5 or later show cyclic course of the O-C variations while only monotonic course or constant period is observed in systems with purely early-type components. Hall attributed the cyclic course to the magnetic activity of the late-type star possessing convective outer layer (COL). Applegate (1992) and Richman et al. (1994) presented a theory which explains the cyclic orbital period variations by changes of the internal structure, caused by the sub-surface magnetic field in COL of a component of the binary. This theory forecasts that changes of the orbital period should be accompanied by the luminosity variations of this star through the cycle and thus allow for an observational test. Hall (1991) found that the period of CG Cyg, an RS CVn-type system consisting of two main sequence stars (G9.5V + K3V), varies with the cycle-length of 50 years and these changes are accompanied by variations of brightness in agreement with the Applegate theory. Maximum brightness and period increase occurred simultaneously. On the contrary, Simon (1997c) revealed two consecutive episodes of shortening the orbital period in the Algol-type system RW Tau. Each event was accompanied by decrease of brightness of the loser. The course of the changes in case of RW Tau was inconsistent with the model proposed by Applegate (1992).
In summary, the evidences were accumulated which suggest that the late-type component is responsible for most observed period changes. An important question therefore arises: what will remain if we remove the suggested effects of the late-type component? If the supposed dominant influence of this star is avoided one could also have a better chance to evaluate the role of the mass transfer in the period changes. Further, if the latest sets of parameters of the respective systems are employed, it is possible to assess their role in the variations of the period. The following analysis which preliminary version was presented by Simon (1997ab) will try to bring some answers.
Copyright The European Southern Observatory (ESO)