526 Holden and Bexon.—Observations on the 
portion of the hypocotyl, each ultimately fusing with its fellow to form the 
two poles of a diarch root. It will be perceived that these seedlings afford 
an admirable illustration of an advanced stage in the evolution of polycotyly 
along the lines we have indicated as probable. 
In addition to these two methods by which the plant may become 
polycotyledonous, another possibility presents itself. Several investigators 
of seedling anatomy have reported the ‘ doubleness ? of the vascular bundles 
of the early epicotyledonary leaves in some species. 
Thus Davey ( 5 ) reports this in Juglans spp., Fagus sylvatica , and 
Castanea sativa , while it is described by Thomas (20) in Cheiranthus 
maritimus and Dr aba Aizoon. It has also been observed by us in some of 
the Cheiranthus seedlings examined during this investigation. In Juglans 
nigra the double bundles from the first two epicotyledonary leaves form, quite 
independently, two of the poles of the tetrarch root, and the same pheno¬ 
menon is reported by Compton ( 3 ) in Caesalpinia sepiaria and in Pitheco - 
lobium Unguis-cati. Bearing in mind these facts and also that Hill and de 
Fraine ( 9 ) interpret some of the phenomena in the polycotylous gymno- 
spermous seedlings as indicating that a plumular leaf has been displaced to 
the cotyledonary whorl, it seems quite possible that the cotyledon number 
may be increased by the precocious development and displacement of one 
of the epicotyledonary leaves, the double bundle of which shares directly in 
root formation. It is evident from these that the ‘ promotion ’ of a displaced 
epicotyledonary leaf to the rank of a whole cotyledon would not necessarily 
involve an intermediate phase as half cotyledon such as Hill and de Fraine 
suggest in their system of evolution. There is, however, no direct evidence 
of the displacement of an epicotyledonary leaf in our material. It has been 
pointed out previously that there are two methods of reduction of the root 
pole number, namely by fusion of poles, and by the loss of one xylem pole. 
This second method is described also by Compton ( 4 ) in Cannabis sativa , 
tricotyl D, and in the majority of the abnormal seedlings of Phacelia tana- 
cetifolia . The bundle which shows this behaviour is regarded by Compton as 
belonging to a c subsidiary * cotyledon, but with this interpretation we cannot 
agree, if the term ‘ subsidiary ’ is to be applied in the sense used by Hill and de 
Fraine. It is conceivable that the two methods of reduction are connected 
with different mqjthods of cotylar increase, reduction by fusion occurring 
when polycotyly is produced by the division of the apex of the cotyledon, 
and reduction by disappearance appearing in seedlings possessing a displaced 
epicotyledonary leaf. The occurrence, however, of intermediate stages 
between fusion and suppression militates strongly against this interpretation 
of the method of reduction under consideration. 
Thus in one or two Cheiranthus seedlings which showed reduction from 
triarchy to diarchy by fusion of xylem poles, one of the two poles became 
somewhat smaller than the other, just prior to fusion, thus presenting an 
