58 
Hoar.—A Comparison of the Stem 
though they occur nowhere in the main stem. It is generally conceded that 
the leaf is more conservative than the stem, and hence we might expect that 
medullary bundles are an ancient character of the Araliaceae. However, 
they are not found in the Cornaceae, which are considered to be lower in the 
evolutionary scale. 
In regard to the wood of the Araliaceae, we find often very broad rays 
which, in the herbaceous species, entirely separate the bundles. The vessels 
have generally simple, elliptical perforations. In two instances (Gilbertia 
and Fatsia) a few bars occur showing an indication of a transition to 
the scalariform type. The end walls of the vessels are more or less inclined 
to the lateral walls. Their lateral walls when in connexion with parenchyma 
of the rays commonly have simple pits. As regards the wood prosenchyma, 
the pits are always simple and we find delicate septations. Such cells often 
contain starch, and are called ‘ septate tracheides ’. 
Finally, in the stem of the Umbelliferae the cortex often contains 
strands of collenchyma which lie directly under the ribs appearing upon the 
outside. Sometimes these are lacking or are replaced by a ring of scleren- 
chyma. The fibro-vascular system is normally in the shape of a ring 
of bundles, with often medullary bundles in addition. The bundles are 
usually isolated, though sometimes they are fused together and are con¬ 
tinuous. Their rays are usually either uni- or biseriate. The vessels have 
usually simple perforations, though sometimes the scalariform type is found 
accompanying them. The pitting, where the vessels are in contact with the 
parenchyma of the rays, may be simple or bordered. The wood prosenchyma 
shows both simple and bordered pits. Anomalous structures sometimes 
occur, such as cortical bundles and an extra-fascicular vascular ring. The 
pith is usually lost in the internodes, thus leaving the stem hollow. 
Having briefly summed up the situation, and having stated the general 
anatomical characters of the order, I will now describe in some detail the 
results of my investigation. Past evidence from material worked over in 
this laboratory seems to indicate that the distribution of parenchyma is one 
of the most important and reliable anatomical characters found in plant 
anatomy. In the Gymnosperms we find it either at the end of the year’s 
growth (terminal) or scattered throughout the annual ring (diffuse). In 
general, among the Dicotyledons the lower families have their parenchyma 
scattered throughout the annual ring (diffuse), while higher families have 
it clustered about the vessels (vasicentric). This character has been found 
to be so constant and of so great value that it has aided much in determin¬ 
ing the relationships between families. Thus the Rosaceae, which from 
a systematic standpoint seem to be close to the Leguminoseae, show a 
different type of parenchyma distribution and hence, apparently, a much 
more remote relationship than is ordinarily assigned to them. 
Thus far in the anatomical study of the Umbelliferae the value of the 
