160 MONOGRAPHS OF NORTH AMERICAN RODENTIA, 
When this exterior triangle is most perfect—most like the antecedent one— 
then also it bears the most perfect supplementary internal spur; but oftener 
the two together have an indeterminate contour and a common dentine islet. 
The last upper molar is the diagnostic toath of this section of the genus. 
Certain European species show it exactly as in our forms; but in North 
America, as far as is known, no Arvicole but xanthognathus and the varieties 
of riparius show the peculiarity. This tooth consists essentially of an ante- 
rior transverse elliptical loop, one interior lateral closed triangle, two exterior 
lateral closed triangles, and a long oblique posterior crescent. The ellipse 
is succeeded first by the first exterior triangle, then by the single interior 
triangle, then by the other exterior triangle; the long anterior horn of the 
crescent bends inward to form a second interior saliency; the long outward 
convexity of the crescent bears the second exterior triangle upon its back, as 
it were; the posterior horn of the crescent curls inwardly to form a loop that 
finishes the tooth behind. With endless minor modifications, as matters of 
individual variability, this crescent is a/ways recognizable and rarely obscure. 
Generally, it is seen at first glance, as something different from the U-, V-, or 
Y-shaped trefoils that end this tooth in our other subgenera. Really, of 
course, it is not a continuous enamel-wall thus stretching crescentie across 
the tooth; simply, the second (counting from backward) internal reéntrance 
is so deep that it pushes before it a fold of enamel till this touches and gen- 
erally fuses with the external wall of enamel just behind the second external 
triangle. It is, in fact, this fusion that produces the last-named triangle itself: 
(In the other subgenera, the corresponding prism of the tooth is simply the 
exterior leaflet of the posterior trefoil, opening directly into the midleaf, 
through lack of the fusion that takes place in riparius.) Now let this second 
internal reéntrance be not quite deep enough to effect this closure, and we 
have the first modification of the crescent to be remembered, a slight break 
in its convexity, just at the posterior angle of the second external triangle. 
When, as occasionally happens, this break is considerable, the integrity of 
the crescent is destroyed, and we have a trefoil-like loop simulating that of 
the other subgenera. But, even in these most obscure cases, we have always 
found something in the configuration, perhaps not susceptible of definition, 
* This little subsidiary triangle is never, to our knowledge, developed at all in our other sections 
of the genus, and therefore, when evident, is a good character; but it is very liable to be overlooked— 
in fact, it was only after repeated examinations that we verified the nice distinction Baird drew (p. 514) 
in the matter of this tooth. 
