AFFINITIES OF DAVIDIA INVOLUCRATA. 321 
upon these lines, especially in the case of uniovular ovaries, have caused considerable 
confusion. For example, Baillon states that in Lustigma oblongifolium “the raphe 
is at first dorsal, but owing to a partial torsion the micropyle is brought to one side," 
and that in Fothergilla alnifolia the ovule is “twisted as in Hamamelis with its 
superior micropyle on one side—at first ventral.” Again, if we examine the nucellus 
of Aucuba we find that it is situated obliquely at early stages, attaining ultimately 
the dorsal position ; while in the case of a biovular ovary the ovules rest tangentially 
in the chamber. The ovules of Symphoricarpus are similarly situated except the 
lowermost ones of the pluriovular chambers, which may take up the dorsal position. 
Ducamp has shown that the lateral preceeds the ventral condition in the Araliacem, 
and that the nucellus of Hedera, at least, is ontogenetically lateral. The mode 
of development of the incomplete integument of Davidia and the position of the 
micropyle in young ovules brings this genus into line with the former, The use 
of the terms ventral and dorsal is obviously two-fold. In some natural orders the 
position of the micropyle is a useful criterion in distinguishing two forms of ovule, as, 
for instance, in the Rosacez. It does not follow, however, that it is serviceable when 
more generally employed. Ina similar way, the terms ascending and descending may 
be employed within a limited range. 
A great deal of confusion has been caused in connection with the set of ovules under 
consideration, by supposing the “dorsal” ovule to possess an inwardly directed 
mieropyle, and the ** ventral" ovule an outwardly directed micropyle. As a matter of 
fact, the micropyle in both types of ovule may be lateral. The lateral micropyle in no 
wise indicates that we are dealing with ovules of an intermediate character, as suggested 
by Harms*, ‘The micropyle is in these cases of subordinate value as a criterion. There 
seem to be two distinct forms of ovule, at least, which have evolved along independent 
lines. The fundamental points of difference are difficult to set down, and can only be 
grasped by a comparative study of ovular development. The differences may, to a 
certain extent, be expressed by the position taken up by the ovule in the loculus. 
There exists, on the one hand, a tendency towards a radial position at maturity, so that 
the vascular bundle of the raphe lies between the nucellus and placenta (Aralia); and, 
on the other hand, towards a tangential position at maturity. Under certain circum- 
stances of sterilization the tangential position changes to the radial, only in this case 
the respective positions of vascular bundle and nucellus are interchanged (Aucuba). 
The above considerations lessen the logical value of the remarks made by Harms (7. c.) in 
his discussion of the affinities of Cornaceæ in 1897. 
The ovule of Davidia is similar to that found in Nyssa and Alangium and in the 
Araliaceze—a natural order characterized by the possession of the “ventral” ovule— 
while it differs very considerably from that of the Cornoidezm. Davidia is strongly 
marked off from the Hamamelidacee in that the latter possess two integuments and a 
bulky nucellus. The nucellus of Davidia resembles that of Cornus in bulk, a resem- 
blance which may be regarded as indicating that the two forms in question are in an 
equivalent evolutionary phase rather than as affording evidence of affinity. 
* Harms in Ber. Deuts. Bot. Gesell. xv. (1897) pp. 21-29. 
9B2 
