ABSTRACT
Canonical correlation has been used in geography to distinguish the interrelationships existing between two data sets. In this paper, a nonparametric method that may be performed by hand is presented as an alternative to the conventional parametric method. Besides its inherent pedagogic advantages, the hand analysis is quick and easy to perform, and it produces essentially the same answer as the computer dependent, parametric approach. It is stressed that the method affords geographers a simple way of formulating hypotheses at the exploratory stage of their data analyses.