The transformation T,
described in a paper of Baum and Eagon, is frequently a growth transformation
which affords an iterative technique for maximizing certain functions. In this paper,
the Jacobian matrix J of T is studied. It is shown, for example, that the eigenvalues
of J are real and nonnegative in a large number of cases. In addition, these
eigenvalues are considered at critical points of T. One necessary assumption used
throughout is that the function P to be maximized is homogeneous in the variables
involved.