322 MR. A. S. HORNE ON THE STRUCTURE AND 
The occurrence of a multicellular archesporium in Davidia is probably not to be 
regarded as primitive. Lloyd * states that in the Rubiaceze the archesporium contains 
7-15 megaspore-mother-cells, the largest number occurring in species of the Galiem, 
Each mother-cell, with the exception of the arrested ones at the side of the archesporium, 
divides to form four megaspores, which, except very infrequently, are not separated by 
walls. He considers this the result of specialization. The condition obtaining in the 
Galieze is simulated by the sterile megasporangia of certain genera of the Caprifoliaces, 
In the sterile ovules of Symphoricarpus there is generally a single archesporial cell 
giving rise to four megaspores which are not arranged in a linear row. One of these 
germinates and produces an incomplete embryo-sac. The incomplete ovules of 
Viburnum Lantana are little more than stalked nucelli. The sporogenous tissue may 
arise from one or more subepidermal cells which are megaspore-mother-cells, and may 
give rise to megaspore groups of four (in the fertile ovules the arrangement of the 
megaspores is linear). Each group of megaspores gives rise to a number of nuclei 
(generally an even number), which are arranged in hollow, sometimes segmented proto- 
plasmic sacs. The condition of these sterile megasporangia recalls the anomalous 
structures which have been observed in Peperomia and Gunnera. In Sambucus nigra 
ovular abortion has culminated in the production of the merest rudiments. These are 
found in the superior portion of the ovary towards the base of the stylar canal. Each 
one consists of a subepidermal layer of cells apparently of an archesporial nature. 
These cells divide to form groups of four without the formation of cell-walls. The 
ovular rudiments in Sambucus afford the last traces of ovules phylogenetically dead. 
The multicellular archesporium of Davidia might therefore be regarded as a development 
correlated with the abortion of ovules upon somewhat different lines. 
The Araliacez, Cornaceæ, and Davidia possess the following characters in common :— 
(1) epigynous flowers; (2) ovaries with one-seeded chambers; (3) ovules with a single 
incomplete integument; (4) small nucellus; (5) endosperm. The ovaries of many 
genera of Cornacesz are uniovular. It has been pointed out that this condition has been 
attained in the Caprifoliaceze by progressive abortion—the different genera proceeding 
towards the uniovular condition along different lines. Hence the uniovular condition, 
whether it refers to the ovary or loculus, should not receive too much weight in 
establishing mutual relationships; at the same time, characters of apparently minor 
importance may serve to indicate a particular line of development, and on that account 
prove of value. Thus the terminal ovules of Davidia and the Araliacez have had a 
different origin. The form of ovule regarded in a similar manner becomes of funda- 
mental importance, and serves to emphasize the gap between the Cornoidex and the 
genera of Cornacez possessing the Aralian form of ovule. Davidia, therefore, stands 
apart from the Cornoides, and is allied rather with the Araliaceze, the Nyssee, and the 
Alangiee. The manner in which the terminal ovule is suspended separates Davidia 
from the Araliaceæ. Taking all the factors into account, the nearest resemblance is t0 
the Nysseæ and Alangieæ. The ovaries of these forms show evidence of having been 
* F. E. Lloyd in Mem, Torr. Bot, Club, viii (1902) pp. 27-111. 
