Lateral viscosity variations are making their way into postglacial rebound
studies [ Gasperini and Sabadini, 1990; Gasperini et al. , 1990;
1992; Wu, 1993]. As shown by Mitrovica and Peltier [1993a; 1993b],
the formal resolution of the post glacial rebound data is an integral over a
rather large depth interval. This raises some concern as to how well postglacial
rebound studies can resolve lateral viscosity structure. Clearly a study using
the formalism developed in Mitrovica and Peltier [1991b] would help clarify
this issue. An interesting observation by Spada et al. [1991] suggests
that seismicity around deglaciated regions may help constrain lower mantle
viscosity. They argue that for a lower mantle with a viscosity of
Pa s differential stresses of order 100 bars may remain in the
lithosphere, while for a lower mantle with a viscosity of 
Pa s
differential stresses of order a few bars may remain. The differential stress
of order 100 bars may be sufficient to induce seismicity on pre-existing
faults. They further speculate that earthquake activity along the passive
margins of eastern Canada and Fennoscandia indicate that differential
stresses exist and are consistent with a high viscosity lower mantle.
In the postglacial rebound studies with non-Newtonian rheology, Gasperini et al. [1992] find that an average effective viscosity for a large region controls the relaxation spectra. This, coupled with the large vertical averaging [ Mitrovica and Peltier, 1991a; Wu, 1993], probably explains why Newtonian models have had such great success in modeling postglacial rebound.