48 Contributions from the Gray Herbarium 
criterion by which to distinguish species in this genus is found in 
the character of the seeds. This character was made use of by 
Watson, Proc. Am. Acad. xiv. 283 (1879), in his revision of To- 
fieldia. His description of the seeds of T. glutinosa, ‘‘ Minute, 
with close brownish testa, and a contorted tail at each end ”’ and 
of T. occidentalis ‘ angular-ovate, with loose white spongy testa, 
and a slender tail at the outer end nearly as long as the body ”’ 
in each case, accurate, and it is surprising that the variable and 
indistinct character of the degree in which the bractlets are connate 
should have been used in place of the definite seed-characters as a 
means of distinction. Yet this character of the bractlets is used by 
Rydberg, 1. c., as the principal difference between 7’. occidentalis 
and T. iarmaia | i.e., T. glutinosa as to western specimens. In 
reality 7’. intermedia bis nearly the seeds of 7’. glutinosa (but as I 
shall show they are not exactly the same) but not at all the seeds of 
T. occidentalis, which species is confined in the Northwest to west- 
ern Washington, Vancouver Island and southern British Columbia 
but extends south to northern California, the type locality. T. 
intermedia, on the other hand, ranges from northwestern Wyoming 
to Montana, eastern Washington and Alaska, and may be dis- 
tinguished constantly from 7’. glutinosa, which occurs as far west 
as Banff, Alberta, only by the different seeds. These are similar in 
shape to those of 7’. glutinosa but are inclosed in a close-fitting 
thin net-like white testa suggesting that of T. occidentalis but not 
spongy and loose about the seed. In general the sepals of 7’. inter- 
media are, as Rydberg indicated, broader than those of T. glutinosa 
but the strength of the species really lies in the difference in seed- 
coat. 7’. occidentalis is even more distinctive since its seed-coat 
is so unusual and since it is normally a much more robust, taller 
plant than either of the more eastern species. Watson based this 
latter species on a nearly mature specimen from northern California 
which shows most of the bractlets connate only at base. Material 
from the same region, however, which is in blossom, often has the 
bractlets quite as connate as in 7’. intermedia, and considerable 
variation in this respect is not infrequently observable in a single 
specimen. 
’¥ THYSANOCARPUS ai le Hook., forma madocarpus (Piper ), 
comb. nov. T’. curvipes Hook., subsp. madocar Piper, Contrib. 
U.S. Nat. ob xi. ci. 306 (1 906). : Ae 
