86 Oliver . — On the Structure , Development , 
leaves nearer to the growing-point than fruits which have 
been normally developed. The cleistogamic flowers are 
simply slightly arrested normal ones, and their presence on 
the floating parts is perhaps due to the fact that, for some 
reason, it was difficult for them to reach the surface. It 
would be interesting to investigate whether cleistogamic 
flowers cannot be produced at will in Trapella , an experiment 
which might easily be performed — should the plant come 
into cultivation — by artificially keeping the flower-buds below 
the surface of the water. It seems quite probable that in 
this way cleistogamic flowers would be produced, and these 
in considerable numbers, showing that they are interchange- 
able, and that the production, in any case, of one form or 
the other depends on external causes rather than on any 
internal tendency. 
Mature fruits, in whichever way produced, are similarly 
appendaged. They are naturally held at the level of the 
water or at a depth of some 3 or 4 — or if cleistogamic of, 
at the most, 8-10 cm. Fish occur to me as the most pro- 
bable agents in dispersal here, and the incurved fruit- 
appendages are admirably adapted to clinging to their spines. 
Jaggi 1 has suggested that fish are instrumental in distributing 
the somewhat similarly appendaged fruits of Trapa natans ; 
but both Ascherson 2 and Nathorst 3 give it as their opinion, 
that more probably it is by ducks or other aquatic birds. 
In Trapella we have the same conditions to deal with, and 
a much greater proneness of the fruits to become entangled ; 
it is on this account a great puzzle to me that this plant 
should have so circumscribed a distribution 4 . Still it may 
1 J- Jaggi, Die Wassernuss, Trapa natans , L., 1883. 
2 Ascherson, in Bot. Centralbl. Bd. xvii. p. 248. 
3 Nathorst, in Bot. Centralbl. Bd. xviii. p. 278. 
4 Dr. Henry states that he only found Trapella in one pond out of some twenty 
he had seen. The pond in question differed from the others, he says, in being on the 
top of a hill, so that the water was little affected by floods of rain ; nor was the 
pond used for irrigation purposes, so that its undisturbed condition would be 
especially favourable to our plant, and might account for this being its sole 
habitat in the district. In the event of fish being the agents in question, distribution 
would depend on the facilities for their visiting other ponds. 
