A PRIORI REVISABILITY IN CONDITIONS OF EMPIRICAL PRESSURE?

In Dynamics of Reason, when he speaks about the second layer of principles in his model of scientific knowledge, Friedman admits that they could change in response to "empirical pressure". Since there is no explicit defense of the a priori character of these principles when viewed as revisable in conditions of empirical pressure I wonder what would actually stop the empiricist of arguing that in this case the principles are empirically revisable, i.e. their a priori status, no matter how "relative", would lose most of its power.