Test of Periodicity in the Quasar OJ 287

OJ 287 displays an apparent periodicity in its violent optical light variations. By applying a progressively brighter faint-magnitude cutoff of the data when performing time series analyses of the light curve, the best-fit period is found to increase from 11.7-12.1 yr (depending on the method used) to 12.4 yr (based on an analysis of only the times of light maximum). In comparison, the average time interval between large outbursts is 11.7 yr. The derived period exhibits noticeable instability, which cannot be caused by the very slow modulation of the light curve or by the double-peak structure of the light maxima. Moreover, the probability of obtaining such a periodicity by chance in the 105 years of past observation is at least 0.02. Nevertheless, arguments based on the observed repetition of structure in the light curve, combined with a binary black hole model, suggest that a real underlying periodicity does exist.