476 
Saunders.—A Reversionary Character 
75, p. 73, Pflanzenfamilien, loc. cit.). A comparison of the above figures 
reveals the interesting fact that here, as in the Rhoeadales, a certain inter¬ 
play takes place between the two kinds of carpels, that is to say, the 
association between form and function is not absolutely rigid. Thus in 
SelenipediLum three of the carpels are of typical valve form, the edges 
incurve in the normal manner, and re-entering, bear the ovules on the now 
free terminal surfaces ; the other three carpels are much reduced, quite solid 
and sterile. In Phragmopedilum there are similarly three carpels of the 
latter type, but here the remaining three have become decidedly massive— 
the process of consolidation has already converted them into the pseudo or 
valve or semi-solid form. They are prolonged to the centre, where they 
meet, but although still valve-like in oritline they exhibit the feature asso¬ 
ciated with the solid conformation, viz. a shifting of the placenta position. It 
seems possible that one whorl in this type having become sterile as well as 
solid, the form of the other has been evolved as a compromise : fertility is 
retained, but consolidation is arrested at a point which leaves enough of the 
primitive form to satisfy mechanical needs. In Trichopilia the trend of 
modification is in the same direction ; the massiveness of the fertile carpels 
is even greater, but their later development is not carried so far : they do 
not meet centrally, hence the ovary remains unilocular. In Cypripedium , on 
the other hand, this shifting takes place without any marked departure from 
the valve form, so that the large ovarian cavity is retained with a deceptive 
appearance of typical parietal placentation. If, as follows from the state¬ 
ment accompanying one of the above figures (see legend to Fig. is, £ Pflan- 
zenreich loc. cit.), the three valves in Selenipedilum are formed from the 
three which in the other genera cited become small, solid, and sterile, we 
have paralleled here among the carpels a reversal of their ordinary 
behaviour such as is seen in the Rhoeadales in the case of Randonia (see 
above, p. 473.) Finally, in such forms as Angraecum and Pleurothallia (see 
4 Pflanzenfamilien ’, loc. cit., p. 73) we find the counterpart of the ‘ compound 
valve * met with in the Papaveraceae and Cruciferae (see above, p. 474). 
If further support for this interpretation of the Orchid gynoecium were 
needed, we find it in the appearance exhibited by certain of the Alismaceae. 
No more striking corroborative evidence could be adduced than that 
presented by the structure of the ovary of Triglochin maritimum , L., with 
its six fertile carpels, and T. painstre , L., with three normal-sized fertile 
valves alternating with three which, though still enclosing a cavity, are 
obviously much reduced and on the way to becoming solid (Figs. 56, 57). 1 
Reduction carried a stage farther would produce the small sterile carpels 
of the Orchidaceae. Furthermore the parallelism between these reduced 
carpels and the commissures of the Cruciferous siliqua is shown by the mode 
of dehiscence. In T. maritimum the ripe fruit splits septicidally into the 
1 See illustrations in Flora Londinensis, v, Pis. XCVIII and XCIX. 
