1999A&A...342..167S

We discuss the light variability of the Mira variable χ Cygni, present a homogeneous set of times of minimum and maximum light since its discovery date in 1686, and analyse the variations of the pulsation period. There is a substantial cycle-to-cycle variation in the pulsations (standard error ∼6.5d) with quite strong evidence for a linearly increasing period. Quasi-cyclic patterns in the O-C diagrams are visible, but there is no compelling evidence that these apparent cyclic period changes are real since the residuals left after fitting a linear period increase plus intrinsic period scatter have properties consistent with being white noise. The Lombard frequency domain test for a changing period applied to the residuals also gives unconvincing significance levels. The low gradient of the luminosity change with period is compatible with the star undergoing a rise in luminosity when H shell-burning takes over.