This paper is concerned with the geographical setting of Roman Algeria and the relation between human settlement and geographical factors, and will contain some comments on the writing of Roman provincial history. The present may be a suitable moment for those who are no longer occupied directly with scholarship to review the first principles and method of provincial history writing, which were perhaps not altogether sound in the published works, and particularly in the general histories, of the pre-war decade. There were two main types of study. The first were monuments of learning and minute scholarship, but tended to convey no general impression at all. The second type were so broad in outline as to be mainly false, a fault particularly marked in the various economic histories and general surveys of the Roman Empire. It is extraordinary that for Roman Africa, with its tens of thousands of inscriptions and superficial ruins and its dozens of town sites, there still exists no coherent or detailed study of part or whole except Toutain's old and excellent account of Tunisia published in 1896. Instead, scholars have concentrated either on particular town sites or on the military frontiers and military stations. The military studies are all linked to Cagnat's monumental work about the Roman Army in Africa. The result is that there exists a coherent body of knowledge about this great topic.