64 FLORA OF THE MONTANA FORMATION. 
are, as he lias pointed out, in agreement with the Point of Hocks speci- 
mens, but in nervation there are quite important differences. In the 
Point of Eocks specimens the finer nervation consists of line quadran- 
gular areolations, whereas the Montana specimens have a more open 
nervation with no evidence of the quadrangular arrangement. But the 
most important point of difference is in the leaf character of the Mon- 
tana examples, which precludes absolutely their reference to the genus 
Trapa — that is, the leaves are compound, being trifolioiate, while the 
original specimens from Point of Rocks are simple so far as now known. 
It is of course possible that these latter examples represent detached 
leaflets, but against this supposition we have the fact that among 
numerous specimens, both from Point of Eocks and from the Canadian 
localities, not one has been found that showed any indication of being 
compound. On the other hand, few detached leaflets are found in the 
Montana material, most of them indicating in one way or another the 
fact that they are compound. 
In the living Trapa the leaves are arranged in the form of a rosette, 
similar in appearance to what must have been the habit of the Montana 
specimens, but in no case are they compound. The Montana specimens 
are not only trifolioiate, but are compound in a very peculiar manner. 
The normal or typical form appears to consist simply of three similar 
leaflets, but in some exceptional instances both lateral leaflets have 
been again divided into two perfect leaflets, thus possessing five leaflets 
but still retaining the trifolioiate character. This is incompatible 
with Trapa, and it would seem that Professor Ward's specimens from 
Montana must receive a new name. Just what their affinities are is 
difficult to say. 
Since the above was written a number of specimens referable to the 
Montana form have been obtained in Wolverine Creek, in the Yellow- 
stone National Park. They formed the characteristic rosette, and 
thus differ from the true Trapa f microphylla. 
Habitat. — Point of Eocks, Wyoming. Bad Lands, Eed Deer and 
Eosebud rivers, and Piucher Creek, Canada. Lance Creek, Converse 
County, Wyoming, base of "Ceratops beds." 
Trapa! cuneata n. sp. 
PL V, fig. 6. 
Leaf obovate-cuneate, regularly rounded above, wedge-shaped below ; 
margin entire for lower two-thirds of leaf, coarsely crenate- toothed at 
apex; midrib slender, somewhat flexuose, with several very slender, 
flexuose, often forking, branches; finer nervation, consisting of very 
numerous small quadranglar areolae. 
This form is represented only by the example figured. It is obovate- 
cuneate in outline, being 2 cm. in length and a little more than 1 cm. 
iu width at the broadest point, which is far above the middle of the 
