Henry Kaiser suggested a rule for selecting a number of components m less than the number needed for perfect reconstruction: set m equal to the number of eigenvalues greater than I. This rule is often used in common factor analysis as well as in PCA. Several lines of thought lead to Kaiser's rule, but the simplest is that since an eigenvalue is the amount of variance explained by one more component, itÂ doesn't make sense to add a component that explains less variance than is contained in one variable. Since a component analysis is supposed to summarize a set of data, to use a component that explains less than a variance of I is something like writing a summary'of a book in which one section of the summary is longer than the book sectio~it summarizes--which makes no sense. However, Kaiser's ma-jor justification for th5 rule was that it matched pretty well the ultimate rule of doing several component analyses with diff-nt- numbers of komponents, and seeing which analysis made sense. That ultimate rule is much easier today than it was a generation ago, so Kaiser's rule seems obsolete.