Quiggin and Chambers have introduced the notion of invariant preferences, and shown that the only invariant expected-utility functionals are those associated with a quadratic utility function. This note identifies the class of preferences which simultaneously satisfy invariance, two-fund portfolio separation, and linear risk tolerance to determine if there exist meaningful classes of preferences, which inherit much of the quadratic family's theoretical and empirical tractability, but do not necessarily inherit its more unattractive properties when regarded as preferences over wealth.