THE ANATOMY OF THE NODE IN ANGIOSPERMS 307 
of this and other genera of the family, however, only a single bundle 
was observed (^fig. 20) . This shows no signs of being divided into three 
parts, and without doubt represents the median one of the original 
three strands. In various species of Myrica the two lateral bundles 
are usually small and have often disappeared. Spiraea, Exochorda 
and their immediate allies alone among the Rosaceae, as far as the 
writer has observed, possess but a single unlobed and undivided leaf- 
trace {fig. 16). In such closely related forms as Physocarpus, however, 
there are the three widely separated strands {fig. 15) and we are forced 
to the conclusion that the trace of Spiraea represents only the median 
one of these. This method of reduction seems to have been a common 
one. From the fact that the single bundle given off to the leaf in the 
Ebenales, Ericales, Contortae and certain other groups is continuous 
and undivided we may infer that the unilacunar condition in these 
forms has arisen by the loss of the two lateral traces rather than by 
the fusion of all three, as in the Centrospermae and Tubifiorae. 
In these two ways, then, the presumably primitive trilacunar 
condition of the node has been reduced. In several groups, however, 
the opposite has happened and amplification has taken place. This 
tendency is shown in many forms but reaches its highest development 
in the Polygonales and Umbelliflorae. In these orders the leaves are 
typically sheathing and are supplied by a large number of bundles, 
each of which causes a gap of its own in the cylinder ( iig. 6) . In 
certain of the Polygonales, however, such as the dioecious species of 
Rumex {fig. 5), and the climbing forms of Polygonum, the node is 
trilacunar. In the Araliaceae, also, which are presumably more 
primitive than the Umbelliferae, the number of bundles and gaps is 
smaller than in the latter family, being as low as five in certain species. 
In young plants of the Polygonales and Umbelliflorae the first formed 
leaves are almost always supplied with three strands which are in- 
serted separately on the vascular ring. In the Ranunculaceae, too, 
the more primitive genera have three bundles and gaps, and the more 
highly specialized ones, many. It therefore seems very probable that 
this multilacunar condition has been derived by the amplification of 
our hypothetically primitive trilacunar type. 
If this supposition is correct, it furnishes evidence of much value 
in determining the relative antiquity of monocotyledons and dicoty- 
ledons. The typically sheathing leaf of the former group is supplied 
by a large number of bundles which come off around the entire peri- 
