Linguists want simple, elegant, and convincing grammatical principles, but the degree to which actual linguistic constraints meet these criteria is often inversely correlated with the amount of their empirical coverage. The problem seems to be particularly strong in the domain of typological generalizations: there are only very few examples of crosslinguistic generalizations that hold without exceptions. Here, I would like to raise the question of whether some of the difficulties can be circumvented if we assume that linguistic communities do not always realize all the possibilities which their grammars allow. What might constitute an unrealised possibility is perhaps best exemplified by prenominal genitives in German. Often, they are consi...