995 
Medullary Rays of Quercus . 
numerous near the ends of the ray as to dissect it into numerous small 
fusiform rays : in addition, vertical marginal fibres extend into the ray. 
From this point the next two types seem to diverge along different 
lines. 
Q. virginiana (8) (Fig. 25) has broad rays more divided than in 
any of the species so far described; they are transversely crossed by 
thick frayed bands of fibres, which differentiate the ray into superposed 
segments. The segments in turn are thoroughly permeated by a complex 
system of thin oblique or vertical lines of fibres. 
In Q. tomentella (11) (Fig. 26) the rays are very much narrower than 
any of those so far described, and are repeatedly divided by uniseriate to 
multiseriate bands of fibres into shallow, slender, oblique fusiform segments 
(or rays)." 
In the series of rays so far given Q. virginiana stands out as the most 
dissected type as regards broad rays, and this position in the Series IV 
agrees with that as indicated by the statistics provided in the second Table, 
though not so closely with the position as judged by mere inspection 
of the distinctness of the pore-zone. 
But when Q . densiflora (6) (Fig. 29) is considered, the correspondence 
between indistinctness of pore-zone and subdivision of the ray vanishes, 
for the broad ray in this species is repeatedly divided by numerous 
vertical fibres and shows an excellent series of stages of vertical subdivision 
of the ray. The meaning of this exception becomes clear when it is noted 
that the species belongs to an entirely different section ( eu-Pasania ) or 
genus ( Pasania ), and the anomaly vanishes when it is compared with the 
other species belonging to the same section or genus, namely, Q. fene strata, 
which will be discussed later in this paper. Thus genetic affinity as well as 
habit influence the structure of the broad rays. 
Yet when the remaining species are considered, namely, Q. semecarpi - 
folia , Q. glauca and Q. lamellosa , and Q. incana , the correspondence between 
division of the ray and indistinctness of annual ring ceases. 
From these, Q . lamellosa (12) and Q. glauca (11) might be excluded, 
because they do not, like the first series of forms, belong to the section 
Lepidobalanus or Erythrobalanus but to Cyclobalanus. The rays of Q. 
lamellosa (Fig. 29) are broad, show transverse frayed partitions formed by 
several series of fibres, and are penetrated at their margins by occasional 
thin vertical strands of fibres: they are thus at about the same stage of 
subdivision as those of Q. hypoleuca or the less divided ones of Q. agrifolia . 
Q. glauca (Fig. 29) possesses rays shallower than those of Q. lamellosa , and 
these are divided by thicker closed bands of fibres into segments that are 
here and there crossed by isolated strands of fibres, so that the rays are 
comparable with the more divided kinds of Q. agrifolia. 
But it is Q. seme carpifolia (Fig. 30) that destroys parallel sequence of 
