Detection of rotation in a binary microlens: PLANET photometry of MACHO 97-BLG-41

We analyze PLANET collaboration data for MACHO 97-BLG-41, the only microlensing event observed to date in which the source transits two disjoint caustics. The PLANET data, consisting of 46 V-band and 325 I-band observations from five southern observatories, span a period from the initial alert until the end of the event. Our data are incompatible with a static binary lens, but are well fitted by a rotating binary lens of mass ratio q = 0.34 and angular separation d approximate to 0.5 (in units of the Einstein ring radius), in which the binary separation changes in size by delta d = -0.070 +/- 0.009 and in orientation by delta theta = 5 degrees.61 +/- 0 degrees.136 during the 35.17 days between the separate caustic transits. We use this measurement, combined with other observational constraints, to derive the first kinematic estimate of the mass, distance, and period of a binary microlens. The relative probability distributions for these parameters peak at a total lens mass M similar to 0.3 M(circle dot) (M-dwarf binary system), lens distance D(L) similar to 5.5 kpc, and binary period P similar to 1.5 yr. The robustness of our model is demonstrated by its striking agreement with MACHO/GMAN data that cover several sharp features in the light curve not probed by the PLANET observations, and which did not enter our modeling procedure in any way. Available data sets thus indicate that the light curve of MACHO 97-BLG-41 can be modeled as a source crossing two caustics of a physically realistic rotating binary. Thus, contrary to a recent suggestion, the additional effects of a postulated planetary companion to the binary lens are not required.