The historically constructed nature of ethnicity has become a widely accepted paradigm in the social sciences. Scholars have especially have focused on the ways modern states have been able to create and change ethnic identities, with perhaps the strongest case studies coming from colonial Africa, where the gap between strong states and weak societies has been most apparent. I suggest, however, that in order to better understand how and when ethnic change occurs it is important to examine case studies where state-directed ethnic change has failed. To rectify this oversight I examine the case of the “lost counties” of Uganda, which were transferred from the Bunyoro kingdom to the Buganda kingdom at the onset of colonial rule. I show that British attempts to assimilate the Banyoro residents in two of the lost counties were an unmitigated failure, while attempts in the other five counties were successful. I claim that the reason for these differing outcomes lies in the status of the two lost counties as part of the historic Bunyoro homeland, whereas the other five counties were both geographically and symbolically peripheral to Bunyoro. The evidence here thus suggests that varying ethnic attachments to territory can lead to differing outcomes in situations of state-directed assimilation and ethnic change.