Fitting Photometry of Blended Microlensing Events

We reexamine the usefulness of fitting blended lightcurve models tomicrolensing photometric data. We find agreement with previous workers (e.g.Wozniak & Paczynski) that this is a difficult proposition because of thedegeneracy of blend fraction with other fit parameters. We show that follow-upobservations at specific point along the lightcurve (peak region and wings) ofhigh magnification events are the most helpful in removing degeneracies. Wealso show that very small errors in the baseline magnitude can result inproblems in measuring the blend fraction, and study the importance ofnon-Gaussian errors in the fit results. The biases and skewness in thedistribution of the recovered blend fraction is discussed. We also find a newapproximation formula relating the blend fraction and the unblended fitparameters to the underlying event duration needed to estimate microlensingoptical depth.Comment: 18 pages, 9 figures, submitted to Ap