The evolutionary trend toward an ontogenetically earlier U-pattern in the first two molars in the primitive lineage of the Geomyinae suggests that the U-pattern characteristic of the Entoptychinae was simply an earlier tendency toward the same specialization that occurred later in the subfamily Geomyinae. If so, early entoptychines would have been characterized by an H-pattern in the first stages of attrition, like _Dikkomys_, and later developed union at the edge of the protomeres. However, in the entoptychines, all the molariform dentition, and not merely the first and second molar, became specialized; consequently the U-pattern was produced on the occlusal surfaces of each of the cheek teeth. As in _Pliosaccomys_, the transitional phase, in which the two columns were united at their mid-points, was eventually eliminated from the pattern of wear and only the U-pattern, that now appeared in the initial stages of wear, was retained. In the entoptychines of the early Miocene there is no suggestion of the H-pattern that characterizes the Geomyinae, except in the position of the cusps before wear in the lower molars of _Pleurolicus sulcifrons_, which, according to Wood (1936:6), suggests the H-pattern. In earlier unknown Oligocene stages of evolution, the prisms possibly united first at their mid-points, and the columns may have joined at the side of the tooth only in the terminal stages of wear. The U-pattern of pre-Miocene entoptychines, therefore, may have become the dominant occlusal pattern only in the later stages of phyletic development. According to the recently expressed views of several paleontologists, the Entoptychinae constitute the primitive lineage of the family and the early Geomyinae constitute a specialized offshoot of the entoptychine ancestral assemblage. The structure of the Entoptychinae, especially of the less advanced genera, closely approximates that of the hypothetical primitive morphotype. But, according to my view, the subfamily Geomyinae constitutes the ancestral assemblage and its structure is essentially that of the primitive morphotype of the family. At any rate the structure of the early geomyines more closely approximates the structure of the ancestral stock than the more divergent entoptychines. Therefore, the genus _Dikkomys_ of the early Miocene, the first known geomyine, is considered to be a generalized geomyid, and, although it is a contemporary of the more specialized entoptychid assemblage, is considered to be more closely allied to the ancestral stock.