There have been several attempts to form and solve an inverse problem for the
viscosity of the mantle using post-glacial uplift data [ Parsons, 1972;
Peltier, 1976]. Analyses of the relative sea level, or uplift history,
over Scandinavia and Hudson Bay were performed by Mitrovica and Peltier
[1993a; 1993b]. The horizontal extent of the Laurentide ice sheet suggests that
this subset of the relative sea level (RSL) data should be sensitive to the
viscosity at greater depths than other data subsets [ Mitrovica and
Peltier, 1991a]. They conclude that the preference of a uniform viscosity in
the lower mantle of 
Pa s from previous studies [e.g., Haskell,
1936; Cathles, 1975] is more appropriately interpreted as a constraint on
the uppermost part of the lower mantle (i.e., 670-1800 km), with very weak
sensitivity to changes in viscosity of up to an order of magnitude below this
depth or in the upper mantle. Therefore, models with increases in viscosity
with depth cannot be ruled out by the RSL data as long as the average viscosity
in the 670-1800 km depth range is
Pa s.