Cyclical period changes in cataclysmic variables: a statistical study

We report the results of a statistical study of cyclical period changes in cataclysmic variables (CVs). Assuming the third-body hypothesis as the cause of period changes, we estimate the third-body mass, $m_3$, and its separation from the binary, $a_3$, for 21 CVs showing cyclical period changes from well-sampled observed-minus-calculated diagrams covering more than a decade of observations. The inferred $a_3$ values are independent of the binary orbital period, $P_\mathrm{orb}$, whereas the $m_3$ values increase with $P_\mathrm{orb}$ by an order of magnitude from the shortest period (oldest) to the longest period (youngest) systems, implying significant mass loss from the third body with time. A model for the time evolution of the triple system is not able to simultaneously explain the observed behavior of the $m_3(P_\mathrm{orb})$ and $a_3(P_\mathrm{orb})$ distributions because the combined mass loss from the binary and the third body demands an increase in orbital separation by factors $\sim 140$ as the binary evolves toward shorter $P_\mathrm{orb}$'s, in clear disagreement with the observed distribution. We conclude that the third-body hypothesis is statistically inconsistent and cannot be used to explain cyclical period changes observed in CVs. On the other hand, the diagram of the amplitude of the period change versus the CV donor-star mass is consistent both with the alternative hypothesis that the observed cyclical period changes are a consequence of magnetic activity in the solar-type donor star, and with the standard evolutionary scenario for CVs.