48 Rhodora [Maat 
2-4 cm. long; and the 3-9 flowers are as small as in var. linearifolius 
or even smaller (1-1.5 cm. long). Although var. myrtifolius is specifi- 
cally separated by some authors from L. palustris by its wingless 
stems, shorter leaflets and smaller flowers, these are all characters 
which are so subject to variation.that no one of them alone is con- 
stant enough for diagnostic use. The nearly wingless stem of var. 
myrtifolius, though a reasonably good character when taken in con- 
nection with the other points, occurs also in numerous specimens with 
the low stature and elongate linear-lanceolate leaflets of var. lineari- 
folius and even in the coarser larger-flowered L. palustris itself. In 
fact, the pubescent extreme of the large-flowered L. palustris, which 
commonly has the stem somewhat winged, was treated by the late 
Theodore G. White in his Revision of the Genus Lathyrus ! as a variety 
of the “wingless”-stemmed L. myrtifolius, his L. myrtifolius macran- 
thus, having the “flowers large (2-2.5 cm. long)" and thus clearly 
opposed to the statement in his key that L. myrtifolius has the 
“flowers less than 1-5 [1.5] cm. long.” 
Not only do the flowers vary in size and the stems in stoutness and 
the degree to which the wings are developed, but the number, outline, 
and length of the leaflets are perplexingly variable, so that, as already 
stated, no one of these characters can be relied upon to distinguish 
L. myrtifolius as a species; but an examination of nearly 200 American 
specimens of the group has shown that, though none of these char- 
acters can be taken as final, they do occur in combinations which 
taken together mark off some very well defined varieties. Three 
of these varieties are characterized above. A fourth, var. pilosus 
(Cham.) Ledeb. was taken up in the 7th edition of Gray’s Manual 
as a plant resembling typical L. palustris in stature, foliage, and large 
flowers, but differing in being pubescent. This large-flowered plant 
with large usually broad leaflets, the L. myrtifolius, var. macranthus 
of White, was identified during the revision of the Manual by com- 
parison with Asiatic material labeled L. palustris, var. pilosus and with 
the figure so called in Reichenbach’s Icones Florae Germanicae (xxii. 
- t. mmcclvii, fig. v). But subsequently a difficulty has arisen through 
the discovery that much of the plant referred in the Manual to var. 
linearifolius is quite as pilose as the coarser plant taken to be var. 
pilosus; and examination of the original description of L. pilosus 
1T. G. White, Bull. Torr. Bot. Cl. xxi. 444-458 (1894). 
