Probing the emitting region using anomalous lensed QSOs.

In gravitational lensing theory, it is predicted that two images which straddle a caustic will have the same magnification. However there are a significant number of well-observed cases where a factor of up to 10 is observed in the magnification ratio - ``anomalous lensed quasars''. These offer a unique window on the nature of both lens and lensed source: From recent modelling \citep{floyd_B+07,floyd_C+07} it appears that the size of the emission region convolved with the microlensing pattern is the main reason for the discrepancy. With appropriate modelling this can be turned around, and a measurement of the emission region of the quasar can be made. This is particularly interesting if a range of observations are made, either at different times (to more highly constrain the probability distribution) or in different wavebands (to constrain the size as a function of wavelength). We present recent multi-band observations of MG0414+0534 as a case study, demonstrating that the anomalous A2/A1 flux ratio decreases as we move blueward, and use the results to constrain the size of the r-band AGN emission region to <7 light days.