of prehistory, historical perspective almost invariably means speculative reconstruction. This circumstance, together with the fact that ethnology has often been called upon by other sciences, such as history, sociology, law, ethics, to give a categorical answer to the problems of ultimate origins, is responsible for the over-crowding of anthropological books and journals with fantastic speculations which are at best of interest as material for the ethnologist rather than as contributions to ethnological science. The critical student is therefore doubly concerned about a careful methodology of historic reconstruction in ethnology. For him Dr. Sapir's Study in Method will prove a rare treat. The author brings to his task good general knowledge of anthropological fact and theory as well as distinctly unusual qualifications as a linguist. This explains the unequal value of the two parts into which the work informally falls. The first, considerably the longer, deals with time perspective in connection with general cultural data, and presents no more than a clear and concise summary of work done by others, often with greater wealth of argumentation and more fortunate in formulation. The briefer second part examines linguistic evidence from the same standpoint; it brings original data and opens new vistas.
To turn to some of the generalizations arrived at in the first part. Culture elements which are presupposed by other elements in order to make the existence of the latter possible, must be regarded as earlier in time (p. 15). A well-defined style in any domain of culture always stands for relative age (p. 18). The larger the territory covered by a cultural trait, the older, ceteris paribus, the trait (p. 28). The interrupted distribution of a feature may serve to establish its minimum age, for it must clearly be ascribed to a period preceding that in which were active the factors responsible for the discontinuous distribution (p. 41). Incidentally the author takes pains to emphasize, with great justice, that the various aspects of culture, such as social organization, religion, art, mythology, technical features, display vastly different modes of behavior in connection with cultural diffusion (p. 32). Therefore, adds the author, it is of the greatest importance to ascertain the paths of diffusion of culture in North America, a task hardly begun (pp. 35–36).
Passing to the linguistic discussion. Dr. Sapir cautions that linguistics can be drawn upon for historical reconstruction in culture only to the extent to which language reflects culture (pp. 51–52). This is eminently the case with vocabulary, which can often be utilized for purposes of relative chronology. Noting, e. g., that the Tsimshian term for phratry defies analysis while that for crest is readily analyzable, one is led to conclude that some form of phratric division antedated among the Tsimshian the appearance of phratric and clan badges (pp. 55–56). Another test is the "criterion of morphologic irregularity": a culture concept associated with an archaic linguistic process is itself an old one; although the reverse conclusion cannot be drawn with safety (p. 64). The analysis of grammatical categories may also throw light on cultural conditions: thus the existence of
