Ice draft data derived from upward looking sonar observations collected during a Scientific Ice Expeditions (SCICEX) submarine cruise in the Arctic Ocean have been used to test the ice thickness distribution theory of Thorndike et al. <1975>. Two separate ice draft surveys, 40 days apart, were made during the fall of 1996 in a circular Lagrangian region ~180 km across. Air temperature and deformation data from buoys in the region were used to force an ice thickness distribution model in an effort to reproduce the changes observed over the 40 day interval. Initial tests with an elementary ridging treatment were unsuccessful in predicting the observed change in the ice thickness distribution. The shape of the distribution suggested that both ridging and rafting of ice were involved in the redistribution process. Modifying the theory to include rafting along with ridging resulted in much improved agreement between the modeled and observed ice thickness distributions. This result, taken together with many other field observations, leads us to believe that rafting is an important component of the mechanical redistribution of ice thickness during the fall.