In 1990 Germany began the reunification of two separate research systems. Yet, the institutional unification of these system does not necessarily imply their actual unification. Here we study the evolution of the network of co-authorships between East and West German scholars between 1974 and 2014 to identify the fields that integrated more successfully, and also, the factors predicting re-unification success. We find that the unification of the German research network was fast during the 1990s, but then stagnated at an intermediate level of integration. Next, we study the integration of the twenty largest academic fields (by number of publications prior to reunification) and find an inverted U-shaped between a field's East or West ``dominance'' (a measure of the concentration of the scholarly output of a field in East or West Germany prior to 1990) and the field's subsequent level of integration. We check for the robustness of these results by running Monte Carlo simulations, and a differences-in-difference analysis. Both methods confirm that fields that were dominated by either West or East Germany prior to the reunification integrated less than those whose output was balanced among East and West. Finally, we explore the origins of this inverted U-shape relationship by comparing the mixing patterns, and show that this inverted U-shaped relationship can be explained as a consequence of a tendency of scholars from the most productive regions to collaborate preferentially with scholars from other top regions. These results shed light on the mechanisms governing the reintegration of networks in the content of scholarly communities that were separated by institutions.