Abstract
Data of Rohrschneider (30 solutes on 23 stationary phases) and of McReynolds (10 solutes on 226 stationary phases) were treated using the Benzecri's “analyse factorielle des correspondences.” Solutes and phases are represented in the same multidimensional space, proximity between the representative points indicating a strong interaction. Separate classifications are also obtained for stationary phases and for solutes as in other statistical methods. Only two factors have to be introduced to account for 92% of the variance. The main factor (85%) was identified as polarity of the phase. The maps obtained for solutes and liquid phases are readily usable to select a potentially effective stationary phase for a given problem of separation.