R help - Sep 2004 - Discriminant function analysis eigenvalue standard error calculations

I am trying to use the machinery of linear discriminant function analysis to
identify group membership for a dataset with many groups, and few members
(2) per group. The MASS package lda() function conveniently returns
eigenvectors specifying the linear combination of my group member covariates
which best discriminant members into groups. I would like to obtain a
standard error for the corresponding maximum eigenvalue (although the
eigenvalue itself does not seem to be returned by lda() ?). Does anyone have
any suggestions on an appropriate method to consider in calculating a
standard error?
Thanks in advance - 

Peter Holck