The fundamental lemma was formulated above with one omission: we never made precise how to fix an isomorphism between Cartan subgroups in and . Such isomorphisms exist, because the two Cartan subgroups have the same root data. But the statement of the fundamental lemma is sensitive to how an isomorphism is selected between and a Cartan subgroup of . If we change the isomorphism, we change the -orbital integral by a root of unity . The correctly chosen isomorphism will depend on the element .
The ambiguity of isomorphism was removed by Langlands and Shelstad in [ 19]. They define a transfer factor , which is a complex valued function on . The transfer factor can be defined to have the property that it is zero unless is strongly regular semisimple, is strongly regular semisimple, and there exists an isomorphism (preserving character groups) from the centralizer of to the centralizer of . There exists such that
The correct formulation of the fundamental lemma is to pick the base point so that Condition 8.0.1 holds.
For classical groups, Waldspurger gives a simplified formula for the transfer factor in [ 31]. Furthermore, because of the reduction of the fundamental lemma to the Lie algebra (Section 7.2), the transfer factor may be expressed as a function on the Lie algebras of and , rather than as a function on the group.
More recently, Laumon (while working on the fundamental lemma for unitary groups) observed a similarity between Waldspurger's simplified formula for the transfer factor and the explicit formula for differents that is found in [ 27]. In this way, Laumon found a simple description of the matching condition implicit in the statement of the fundamental lemma.