We present new eclipse observations of the polar (i.e. semi-detached magnetic white dwarf + M-dwarf binary) HU Aqr, and mid-egress times for each eclipse, which continue to be observed increasingly early. Recent eclipses occurred more than 70 seconds
earlier than the prediction from the latest model that invoked a single circumbinary planet to explain the observed orbital period variations, thereby conclusively proving this model to be incorrect. Using ULTRACAM data, we show that mid-egress times determined for simultaneous data taken at different wavelengths agree with each other. The large variations in the observed eclipse times cannot be explained by planetary models containing up to three planets, because of poor fits to the data as well as orbital instability on short time scales. The peak-to-peak amplitude of the O-C diagram of almost 140 seconds is also too great to be caused by Applegates mechanism, movement of the accretion spot on the surface of the white dwarf, or by asynchronous rotation of the white dwarf. What does cause the observed eclipse time variations remains a mystery.
