Anatomy of Teratological Seedlings . /. 527 
appearance intermediate between ordinary fusion and reduction by the dying 
away of a xylem mass, and this intermediate condition has been noted in 
a more pronounced degree in some tricotylous seedlings of an undeter¬ 
mined species of Matthiola . Two of these show a loss of protoxylem in one 
of the two xylem plates which ultimately merge, whilst a third shows a loss 
of metaxylem, so that before fusion one xylem plate is represented simply 
by a separate strand of crushed protoxylem elements. 
It was thought possible at first that the reduction by fusion and disap¬ 
pearance might be successive stages in the acquirement of persistent triarchy. 
If this were true, however, one would expect the dying away of a pole to 
occur always low down in the root, since it would constitute the stage 
immediately preceding complete and persistent triarchy. This is, however, 
not the case, for in some instances among the tetracotylous seedlings the 
tetrarch stage is reduced to a triarch one in the hypocotyl by the dying 
away of a xylem plate, whilst reduction by fusion of poles is found in the 
root. 
All things considered it seems preferable to regard this type of reduc¬ 
tion in the number of poles as either a subsidiary line of evolution or as 
due to a sudden dominance of the ancestral hypocotyledonary diarchy over 
the newly acquired triarch symmetry. 
It will be obvious, however, that though there is a considerable body 
of evidence which lends support to our hypothesis concerning the origin of 
polycotyly, there are at the same time a number of difficulties which 
present themselves. In the first place, it may be suggested that there are 
structures which are obviously half bundles of a cotyledon which each show 
double structure, as for example in the Calycanthaceae, Fagaceae, Euphor- 
biaceae, Sapotaceae, Ebenaceae, and certain of the Rosaceae, in which the 
so-called diagonal arrangement occurs. A careful consideration of these 
will show an essential difference however, since in the majority of cases 
a very evident median protoxylem exists between the paired double bundles, 
and even in the extreme case of Calycanthus Chauveaud has demonstrated 
the existence of such a strand in the very young state. No trace of 
a median protoxylem has ever been found by us between the twin double 
bundles in the Cheiranthus seedlings, and such might reasonably have been 
expected had they originated from modified half bundles. It must be 
admitted^ however, that no median protoxylem appears in the separate 
parts of the split cotyledons between the two collateral bundles of the 
type a seedling, which are recognized as half bundles.. There is, however, 
a further important distinction, for, in the groups named above, the consti¬ 
tuent halves invariably constitute a divergent system, and in no case do 
they unite at a lower level, while in the type {3 Cheiranthus seedlings the 
bundles form a convergent system, and in many cases unite either in the 
hypocotyl or in the root. 
