The statistical significance of the superhump signal in U Gem

Context. Although its well determined mass ratio of q = Msec/Mwd = 0.357 ± 0.007 should avoid superoutbursts according to the thermal tidal instability model, the prototypical dwarf nova U Gem experienced in 1985 an extraordinary long outburst resembling very much superoutbursts observed in SU UMa systems. Recently, the situation for the model became even worse as superhump detections have been reported for the 1985 outburst of U Gem. Aims. The superhump signal is noisy and the evidence provided by simple periodograms seems to be weak. Therefore and because of the importance for our understanding of superoutbursts and superhumps, we determine the statistical significance of the recently published detection of superhumps in the AAVSO light curve of the famous long 1985 outburst of U Gem. Methods. Using Lomb-Scargle periodograms, analysis of variance (AoV), and Monte-Carlo methods we analyse the 160 visual magnitudes obtained by the AAVSO during the outburst and relate our analyse to previous superhump detections. Results. The 160 data points of the outburst alone do not contain a statistically significant period. However, using additionally the characteristics of superhumps detected previously in other SU UMa systems and searching only for signals that are consistent with these, we derive a 2σ significance for the superhump signal. The alleged appearance of an additional superhump at the end of the outbursts appears to be statistically insignificant. Conclusions. Although of weak statistical significance, the superhump signal of the long 1985 outburst of U Gem can be interpreted as further indication for the SU UMa nature of this outburst. This further contradicts the tidal instability model as the explanation for the superhump phenomenon.