Cepheids are stars that pulsate with remarkable regularity: their brightness changes like a lighthouse beam, and the pulsation period is linked to the star's average density. In binary systems where a Cepheid is eclipsed by its companion, astronomers can precisely measure the star's radius and mass. Usually, these data are used to fit evolutionary models, but the pulsation period, the most accurately observed quantity, often gets overlooked.

In a new study, the authors decided to test what happens when the period is included in model fitting. They modeled four binary systems with Cepheids in the Large Magellanic Cloud and compared two approaches: fitting by radius and fitting by period. The results were unexpected: the two methods give different answers. When the model is fitted by period, the predicted radius is systematically smaller than observed. When fitted by radius, the period comes out too long. The discrepancy reaches several sigma.

Part of the problem appears to be related to a nonlinear increase in radius for Cepheids with large pulsation amplitudes. This effect has not yet been studied in detail. The authors recommend using the pulsation period rather than the radius in such studies and call for a systematic investigation of nonlinear effects in powerful pulsators.