200 Rhodora [NOVEMBER 
(2) upper leaves sometimes separate instead of sometimes perfoliate; 
(3) internodes shorter; (4) plant stouter; (5) plant more leafy; 
(6) leaves more rugose, (7) thicker, (8) paler beneath, (9) more densely 
soft-pubescent; (10) calyx-segments much shorter, (11) less foliaceous, 
(12) narrower, (13) more acute; (14) corolla often duller and greenish 
on lower half instead of dull purplish red, firmer; (15) the tube scarcely 
dilated or two-lipped instead of decidedly two-lipped and upwardly 
dilated; (16) the lobes shorter and more erect, (17) and scarcely 
surpassing the anthers; (18) style more exserted; (19) fruits more 
numerous (6-8 instead of 2-6) and more crowded, (20) more globose, 
(21) smaller, (22) dull greenish-orange instead of orange or bright 
orange-red; (23) an inhabitant of more sandy soil, (24) and of lower 
more level woods and thickets; (25) flowering about two weeks earlier. 
To these differences the writer may add as no. (26) a more densely 
crisp-pubescent stem in T. perfoliatum. A comparative study of 
about 175 herbarium specimens in addition to many in the field has 
failed to show that these differences are all valid. With the character 
of perfoliate leaves only nos. 9, 12, 13, 14, 15, 16, 18, 23, and 26 
could be even generally correlated. Under most of the other numbers 
no differences could be found while some numbers were indiscrimin- 
ately variable. Where the above differences seemed to correlate, 
the correlation was, however, chiefly in averages and the extremes 
overlapped very badly, showing tendencies rather than distinct groups. 
To a certain extent it was found that the same characters were not 
always combined, thus forming various combinations of characters. 
Sortings on different characters showed no possibility of making dis- 
tinct groups, the only groups approaching distinctness being those 
given in the key below. Perfoliate leaves were found occasionally in 
the non-perfoliate group; dense crisp pubescence frequently in the 
group with normally loose villous pubescence; narrow acute sepals 
where these should be broad and blunt; non-flaring corolla where 
this should be flaring. A constant difference in length of stamens 
could not be made out. In the var. glaucescens the sepals were 
narrow or broad. Though originally not doubting that T. perfoliatum 
and T. aurantiacum were distinct species, the writer has now come to 
the belief that there is far too much intergrading to make possible 
the retention of both as species. It is therefore proposed to treat 
the North American Triosteums as in the following key. Because of 
the exceptions, the varieties under T. perfoliatum are best recognized 
