AFFINITIES OF DAVIDIA IN VOLUCRATA. 319 
they are firmly wedged in theloculi. Dehiscence takes place by the removal of the upper 
part of the back of each loeulus. The seeds before dehiscence are about three-quarters 
of an inch long and are enveloped in a brownish, scaly testa, which is all that remains 
of the extra-nucellar tissue of the ovule, and consists of crushed and broken cell-walls 
with remains of vascular bundles. The endosperm is absorbed by the embryo uniformly 
by means of its limiting layer of cells, which at the time functions as an epithelium : 
the endosperm cells adjacent to it are crushed and disorganized. After dehiscence, the 
radicle bursts through the endosperm at the micropylar end and the seeds germinate 
while still fixed in the fruit as described by Hemsley *. 
7. Discussion oF AFFINITIES. 
There are three groups of plants near Davidia, possessing flowers with multiseriate 
stamens :—1, Nyssa and Alangium, outlying forms of admittedly Araliacean affinity 
among the Cornacesx; 2, the Schefflereze; 3, many genera of the Combretaces, to which 
order Davidia was provisionally referred by Baillon. The supposed resemblance of 
Lindley's Alangiaceze to the Combretacex is due to Baillon’s ingenious but fallacious 
argument for the derivation of bi-chambered from unilocular Alangiums by the centripetal 
advance of parietal placente. Baillon f states that at one time the Combretacee 
were thought to comprise plants with ovules inserted near the summit of the ovary, 
and later it was seen that their placentz were parietal and centripetal. Imagine 
the latter to advance further, and we have an ovary with two cells, incomplete or 
complete, as sometimes observed in Nyssa and in the bicarpellary Alangiem. In Nyssa, 
however, the process is precisely the reverse of that described by Baillon. The 
occasional presence of a rudimentary style and of an incomplete additional chamber, 
together with the variable position of the single style, are facts pointing to progressive 
reduction. If this be so in Nyssa, the case could hardly differ in the Alangieæ, where 
we have two-chambered (Marlea) or incompletely two-chambered ovaries (Alangium). 
It was on this point that Baillon referred Nyssa and Alangium to the Combretaceze 
rather than to the Araliacez. He considered their affinity to either of these orders 
greater than their affinity to the Cornacee. 
The ovary of Davidia presents certain general resemblances to that of Aralia, notably 
in possessing a distinct ovarian vascular system and one-seeded chambers—the seeds in 
both cases resting radially in the loculus with the vascular bundle between the nucellus 
and axis. The ovaries of both genera are inferior, but the degree of epigyny in Davidia 
is greater than it is in Aralia, and associated with this the carpellary system of the 
former is more concentrated. The two forms are in nearly equivalent evolutionary 
phases. There are, however, important differences both in the mode of suspension and 
vascular connection of the ovule. To make this point clear let us turn for a moment 
to the Caprifoliacese. The ovaries of Leycesteria and Lonicera are pluriovular, 
the ovules being arranged in biseriate rows in the chambers. The lower ovules of the 
* Hemsley in Journ. Linn. Soc., Bot. xxxv. (1903) p. 556. 
t Baillon, Hist. Pl. vi. p. 271. 
SECOND SERIES.—BOTANY, VOL. VII. 35 
