Recent discussion of the effects of finite source size on high magnification microlensing events due to MACHOs motivates a study into the feasibility of observing such effects and extracting the source radius .
Simulated observations are generated by adding Gaussian error to points sampled on theoretical microlensing light curves for a limb darkened , extended source .
These simulated data sets are fitted in an attempt to see how well the fits extract the radius of the source .
The source size can be fitted with reasonable accuracy only if the impact parameter of the event , p , is less than the stellar radius , R _ { \star } .
It is possible to distinguish “ crossing ” events , ones where p < R _ { \star } , from “ non–crossing ” events if the light curve is well sampled around the peak and photometric error is small — i.e . \geq 3 observations while the lens transits the disk of the source , and \sigma _ { phot } < 0.08 { mag } .
These requirements are just within the reach of current observational programs ; the use of an early-warning system and multiple observing sites should increase the likelihood that R _ { \star } can be fitted .
The programs used to simulate and fit data can be obtained via anonymous ftp .