Parnassia and its bearing on the Affinities of the Genus . 495 
tracheides is proved by radial longitudinal sections, in which they are easily 
distinguished by their spiral thickenings, which have undergone great 
stretching (PL XXXVI, Fig. 2, px). In this longitudinal section also, as 
in the transverse section, the protoxylem is seen to be attached to the 
normal centrifugal xylem and separated from the centripetal elements by 
parenchyma. 
Returning to the transverse section (PL XXXVI, Fig. 1), we notice 
that the xylem is surrounded at a little distance by a number (about ten) 
of small groups of elements which appear to represent phloem. 
We do not find the ring-structure completely developed in the xylem 
of every filament of Parnassia palustris , nor is it continuous throughout 
every individual filament in which it occurs. Typical deviations are shown 
in PL XXXVI, Figs. 3 and 4, which represent the variations in structure in 
different regions of a filament which, for a short distance near the base, 
possessed a continuous ring of xylem resembling that shown in PL XXXVI, 
Fig. 1. In PI. XXXVI, Fig. 3, we have an incomplete ring broken on the 
ventral side, while in PL XXXVI, Fig. 4, it is broken laterally, so that the 
centripetal and centrifugal xylem are entirely separated. In other cases 
the ring is broken both dorsally and ventrally. 
The mesarch structure, which we have been discussing, is retained to 
the top of the filament. When the bundle enters the connective, however, 
its organization undergoes a change. The xylem increases in amount and 
begins to branch. The branches remain attached to one another for a short 
time, but ultimately separate into a number of distinct strands, so that, in 
the middle region of the connective, there is a considerable xylem complex 
(PL XXXVI, Fig. 5). It is difficult to count the exact number of strands, 
since there is often no definite limit between them, but there are certainly 
sometimes as many as ten. The xylem is embedded in and surrounded by 
cells rich in contents, among which a certain number of elements which 
suggest phloem can be distinguished. Higher up in the connective the 
branches gradually die out, and the dwindling vascular complex is eventually 
reduced until it consists only of a single small xylem group, which dis- 
appears before the extreme tip of the anther is reached. 
The description so far given applies to young flowers in which the 
anthers are still retained. When we examine the filaments of old stamens, 
which have shed their pollen and lost their anthers, we find that the xylem- 
ring in the filament has become more massive by the outward spread of 
lignification, and that the internal tissue, parenchyma, and protoxylem have 
been torn, presumably in the course of the elongation which the filament 
undergoes when the pollen is about to be shed. 
I have found the mesarch structure, which I have described, in the 
filaments of Parnassia palustris from more than one locality. I have also 
examined the flowers of three other species — P. fimbriata. Banks ; P. wont a- 
