areas, as for example Myocastor, which has shifted to the lowlands to the west and south.

Possibly climatic changes were responsible for the faunal shift from the region that is now a plateau in Central Brazil. This climatic change may have resulted from the gradual uplift of the eastern part of the continent. This uplift prevents part of the trade winds which come from the east from carrying the same amount of moisture inland as they did previously. In fact, the Andean revolution, even if it occurred as late as Late Tertiary, would have had no perceptible influence on the amount of water precipitated on the more eastern parts of the continent. Oliveira and Leonardos (1943:617) point out that after a Cretaceous submersion of the central part of Brazil, there was a general uplift. The authors (op. cit.:689) mention the presence of continental Cretaceous deposits in the Central Plateau of Brazil, in support of these changes, and state that "pelo menos em certas zonas do litoral a elevação do continente prolongou-se até o Pleistoceno."

Berry (1942:373) concluded, among other things, that there was a southward extension "in South America of equatorial floras in the lower Miocene," and (op. cit.:372) that ... "east of the Andean Axis in the south temperate zone there was a normal mesophytic flora ... instead ... of present day large steppes."

My idea is that a tropical forest still covered the Central Plateau of Brazil in (early?) Pleistocene times and that populations of Proechimys of a primitive type, similar to P. g. steerei, for example, lived in that extensive forest-climax. The gradual uplift of the plateau, however, gradually brought about drier conditions in this region. As a result a large cliseral change was initiated, which shifted the forest-climax to the more humid eastern escarpments and lowlands that were gradually being developed, while the savanna climax was being established on the plateau. Eventually the effect of the decreasing moisture was locally accentuated by the erosion of the sandstones (Oliveira and Leonardos, 1943:690) in northeastern Brazil, thus depriving it of a natural reservoir of rain water. An arid belt was developed which now constitutes an efficient geographic barrier to the distribution of many kinds of animals.

One marginal species may have shifted eastward with the forest-climax to effect the Recent distribution. The eastern species became completely isolated from the main group, accumulated mutations, and evolved into the subgeneric type Trinomys. The generic trend that gave rise to Trinomys probably remained more stable as far as supraspecific changes are concerned. The lack of barriers in the distributional area of the original group favored the dispersal and submergence of mutations and, therefore, there was but little further supraspecific evolution. The speciation in both subgenera finally resulted from gradual differentiation of varying populations since they show combinations of the generic biotypes and possess few truly qualitative characters.

The cliseral changes in the Central Plateau, which developed the dry belt, a barrier, might explain the evolution of a few more supraspecific groups of mammals, as indicated by the presence of similar forms in the Amazonian region and in Southeastern Brazil. Among these Echimys and Phyllomys, in the same family with Proechimys, show differences that are parallel to those observed in Proechimys. One of these parallel changes is the increased lamination of the cheekteeth. Although Echimys, from the Amazonian region, has upper molariform teeth with the four laminae fused, Phyllomys has the four laminae completely separated.

None of the genera known from the Upper Oligocene and Miocene of Argentine deposits seems to be directly ancestral to Proechimys.

SPECIATION

The detection of differences of systematic worth between populations of animals, represented by skins and skulls, is a step preliminary to deducing the factors responsible for the differences. Ordinarily the factors which cause heritable differences have to do with geographic isolation and adaptation to ecological conditions. When differences in the structure of the animal are known, a person is led to speculate on the factors which could cause them. For one thing, does the observed degree of difference tend to isolate animals possessing the "new" character from the other animals? It would seem to me that the isolation once started by one of these differences tends to be accentuated with time and the difference itself thus then becomes a factor responsible for further differentiation.

Whether or not transition from one character to another occurs gradually, in its geographic expression, and thus whether or not intergradation occurs between two subspecies, can be ascertained by the analysis of a series of population-samples appropriately distributed geographically. If two characters of systematic worth are known to blend in one part of the geographic range of a subgenus, and if the same two characters are seen in two other populations, far removed geographically from each other and without any samples of annectent populations to provide actual evidence of intergradation, then such intergradation is to be inferred.

The available collections of Proechimys mostly were made haphazardly with the result that there are extensive areas from which no specimens as yet are available. Thus, actual proof of intergradation is often lacking in areas where it almost certainly occurs. In some extensive areas, however, many samples, from relatively regular intervals, have been available and they provide genuine proof of intergradation. These instances have served as a guide for estimating whether other samples should be considered to be full species or instead merely subspecies of the same species.

Lack of intergradation in any of the characters may be accepted as the criterion of full species. Where two populations occupying the same range (sympatric populations) show different qualitative characters, they almost certainly do not crossbreed. Furthermore the characters that distinguish such kinds of nonintergrading animals are likely to be considered as of full specific value when detected in far distant parts of the range of the subgenus.

In a genus that is widespread and continuously distributed, it is useful to know which characters always distinguish full species and which ones, sometimes or always, distinguish only subspecies, since in a population from a small island, there is, ordinarily, less individual variation than in a corresponding population from the mainland or a larger island; under certain circumstances a person might be tempted to give specific rank to the population when its characters actually are analogous to those separating subspecies elsewhere.

Sometimes it is convenient to recognize species-groups, a systematic category without nomenclatural status, intermediate between the species and the