We consider various possibilities for the uncorrelated reference for t
he calculation of properties. According to the philosophy of Lowdin, t
o whom this volume is dedicated, the uncorrelated reference state for
response properties ought to be taken as the unperturbed Hartree-Fock
state for properties of all orders in the applied field. Frequently, h
owever, it is operationally convenient to use the coupled Hartree-Fock
like results as a standard of comparison for second- and higher-order
properties, even though this is not consistent with Lowdin's choice f
or the uncorrelated reference state. In this method the reference stat
e is the perturbed Hartree-Fock state. Numerical examples demonstrate
a rather large difference between the two uncorrelated references. We
consider the pros and cons of each choice.