The June 1994 Protocam observations of M92 were not optimally sampled, having been stepped by integer pixels frame-to-frame. Nonetheless, the repeatibility of aperture photometry approaches the measurement limits predicted by simulations of the errors induced by the undersampling of 2MASS data ( e.g. Reference 1). For bright point sources, repeated measurements show a typical scatter of approximately 2.5%, a value that is very consistent with the Level 1 Requirements for photometric accuracy, and allows room for less than perfect photometric calibrations. As encouraging as these results are, we note that the scatter in the psf-fit photometry is still relatively large for bright sources. Furthermore, we know that psf-fitting performs better for faint objects, is more robust to sampling errors due to bad pixels, and that it is the only viable option to carry out photometry in crowded fields. Therefore, there is considerable motivation for identifying means by which the photometric quality of the 2MASS data can be improved during acquisition and processing. Below, we describe several possible improvements, each of which is being tested using the existing Protocam data, and hopefully new test data from the planned April run.