PEROMYSCUS MANICULATUS.--Six subspecies of _Peromyscus maniculatus_ occur in the state of Washington. The geographic range of one of these (_rubidus_) lies mainly in the states of California and Oregon and includes, so far as is known, a single small island in the Columbia River that is politically within the state of Washington. Another (_hollisteri_) is restricted to certain islands in northern Puget Sound and obviously has become subspecifically differentiated in postglacial time. The remaining four subspecies, namely _oreas_, _austerus_, _artemisiae_ and _gambelii_, have extensive geographic ranges. These subspecies are not confined to their geographic ranges by geographic barriers. Deer mice occur in the deep forests and the open desert, on high mountains and in low valleys, and are almost everywhere the commonest species of mammal present. The study of several populations of deer mice from any general area usually shows small but constant differences between them. Dice (1939: 21) studied stocks of deer mice from nine localities in southeastern Washington and found significant differences between several of them. A statistical study of mice from the San Juan islands shows that the population of almost every island is different in one or more respects from the mice of any other island. Geographically separated populations of "wild caught" mice of the subspecies _austerus_, of the mainland, were statistically compared and significant differences were found between these populations, too. Small, differentiated populations are to be found in many parts of the state, and each subspecies appears to be an assemblage of such tiny genetic units. These genetic units probably are the microgeographic races of Debzhansky (1937). They have been intensively studied by Sumner (1917 A, B) and Dice. An especially important paper by the latter author (1940) summarizes much of the available information on speciation in _Peromyscus_ and clearly discusses the microgeographic races of _Peromyscus_. The numerous microgeographic races of _Peromyscus maniculatus_ in Washington present diverse combinations of characters which could result from the random fixation and elimination of genes (Wright. 1932: 360-362). Such a hypothesis, however, requires at least partial isolation of the populations involved. The mechanism of such isolation, in such populations of deer mice as we have studied, is not readily apparent. Some microgeographic races are not separated by noticeable geographic or ecologic barriers and the distance between their ranges is not too great to be traversed by a deer mouse. The tendency to remain on a home range may have the same effect as isolation would be supposed to have.