This study tests several explanations for discrepant results in an earlier study (Cook et al., 1985) which presented a partial pre-calibration method for equating new editions of the Scholastic Aptitude Test (SAT) to the same scale as older editions. In contrast to full pre-calibration, which seeks to equate all items from two or more editions, the partial method designates a subset of items to serve as an equating section; both methods rely on item response theory (IRT) and employ LOGIST to achieve calibration. Several verbal and mathematics equatings selected from the larger study were subjected to further tests. Estimation error of item parameters was small; instead, the source of difficulty rested with the IRT characteristic curve transformations, which failed to adequately equate the series of calibrations runs to the same scale. Possible differences in ability levels at different test administrations could not account for the discrepant equatings. Finally, the study identified items that functioned differently (DIF items) in the two groups used for equating and calibration. Eliminating these items from the partial pre-calibrations runs, however, did not improve the equatings significantly. Figures depict the calibration designs, item response functions, and other results. (LPG)