The analysis of star catalogs. 1: An intercomparison among the FK3, the FK4, and the FK5

Progress in the construction of the fundamental catalogs allows one to ask the following questions: 'Now that we have the FK5 with which to measure by, how good was the FK4? The FK3? How realistic were the error estimates they gave for their equatorial coordinates and their proper motions? Have their residual systematic differences continued to decrease in amplitude?' and so forth. Similar questions could be asked of the GC and the N30, and a companion paper will similarly analyze them. Finally, the identical series of questions could be asked of the AGK2, AGK3, AGK3U sequence. In this paper we develop a general method to objectively investigate, post facto, the true quality of star catalog error estimates. We present a complete mathematical formulation of the questions above and delineate the calculations needed to address them. Our initial numerical applications are restricted to a straightforward examination of the mean errors of the coordinates and proper motions in the FK3 (using the FK4 and the FK5 for comparison) and in the FK4 (using the FK5 for comparison). However, we can simply and consistently interpret the results of these computations -- for the FK catalogs fail to follow the predictions of mathematical statistics -- only if the FK3 and the FK4 contain higher angular frequency systematic errors. These errors, at the mean epochs of place of the older catalogs, are larger than the random errors in the FK5 system. We expect that our new ability to discern these features means that we can also prevent them during compilation. Finally, note that the concepts behind our procedures are completely general and can be applied to any set of compiled data -- astrometric, photometric, geodetic, and so forth.