562 OX SEEDLINGS 



laterally between the ribs, or irregularly at the apex, or rarely 

 by valves. Seeds very numerous, and very small, with a thin 

 transparent testa, sometimes adnate to the nucleus, sometimes 

 loose, hyaline and produced at both ends a long way beyond 

 it. Recent researches l have shown that the seed, at any rate 

 in many cases, contains something more than the rudimentary 

 undifferentiated embryo which was formerly supposed to 

 constitute the whole. Endosperm is present, and occupies 

 the greater part of the seed, one of the cells being often very 

 curiously developed. In Burmannia capitata according to 

 Johow^a portion of the nucellus remains in the ripe seed 

 (perisperm). In this species the number of the endosperm 

 cells is very small (six in a longitudinal section), but they are 

 each very large. At the apex lies, embedded in it, the undiffer- 

 entiated embryo of ten cells. At the base is a small conical 

 group of cells distinguished from the endosperm by their small 

 size and dark colour ; they are a remnant of the original 

 nucellus and therefore perisperm. The endosperm cell at the 

 extreme base of the embryo-sac is most curiously developed, 

 the wall bordering on the perisperm having peculiar tap-like 

 or coralline protuberances which penetrating the cell-contents 

 reach almost to the opposite wall. This as well as the conical 

 nucellar process (perisperm) may play some part in germination. 



Apteria setacea is similar, but the differentiated endo- 

 sperm-cell is hexagonal and appears symmetrically striated 

 throughout, the protoplasm having apparently completely 

 passed over into cellulose or a closely related substance ; the 

 perisperm is very small. The embrj-o, which agrees hi all de- 

 tails with that of Burmannia javanica and Gonyanthes capitata 

 as observed by Treub, is simpler than in B. capitata, containing 

 only from three to six cells. 



Burmanniacese is a small tropical Order of small herbace- 

 ous saprophytes containing about a dozen genera. Their 

 germination does not seem to have been studied. 



1 Treub, Extrait des Annales de Jard. Buitenzarg, iii. pp. 120-2. 



2 Johow, Pringsh. Jahrb. xvi. (1885), p. 415. 



