Morphologically _P. dimidiatus_ and _P. iheringi_ are sometimes difficult to distinguish, especially on the basis of cranial features. Nevertheless close attention to the small, nearly round, incisive foramen of _P. dimidiatus_ versus the larger, more elongate foramen in _P. iheringi_ will permit separation of the two. However, the two species live in the same place and one is led to infer that there may be greater differences in their physiology than in their morphology. In fact Dr. H. W. Laemmert, from the Serviço de Estudos e Pesquisas Sobre a Febre Amarela in Brazil, informs me that while _P. dimidiatus_ was highly susceptible to the virus of yellow fever (18 out of 24 with virus in circulation), _P. iheringi_ showed a lower rate of susceptibility (3 out of 25 with virus in circulation). _P. longicaudatus roberti_, belonging in the other subgenus, showed no susceptibility at all. At Teresópolis, Estado do Rio de Janeiro, the two species were found in two different forests, only a few kilometers apart, but _dimidiatus_ lived at a higher elevation, where the humidity was remarkably higher. Naturally the plant associations were different in the two forests. This seeming ecological adaptation of the two kinds of _Proechimys_ may explain why _P. iheringi_ ranges farther north; the forests to the northward are less humid. One of the four species, _P. setosus_, subspecies _elegans_, was used by Winge (1941:80, 82) as representative of the genus _Proechimys_ when he was estimating the relationships of that genus. Because _Cercomys_, with four crests in each of its cheekteeth, was, on other grounds, regarded by him (_op. cit._: 80) as "... the most primitive genus within the group.", and because he noted in _P. s. elegans_ 4 crests in P4 and in some first molars, he concluded that _Proechimys_ was "very closely related to _Cercomys_." His conclusion seems to be correct, but actually other species of _Proechimys_ (subgenus _Trinomys_), for example, _P. dimidiatus_, have four or more crests in each cheektooth, and, therefore, may be considered as more closely related to _Cercomys_ than is _P. setosus_. If a large number of crests indicates primitiveness, _P. dimidiatus_, always with four, is more primitive than any other species in the subgenus _Trinomys_. Also, the large skull, long hind foot, short tail and thin aristiforms of _P. dimidiatus_, in my opinion, are primitive characters. SUBSPECIFIC VARIATION IN THE SUBGENUS TRINOMYS