METAMORPHOSIS. 



167 



out, is excluded, for a second leaf would not occur in 

 front of the first except in a case like that of our 

 Buddleia, where the two opposite leaves of a node be- 

 come terminal to the shoot ; in the Jatropha-Rower 

 the stamens are lateral structures, so that each stamen 

 can only represent a single leaf. On the other hand, 

 supposing the dichotomy-hypothesis to be true, the 

 forking, in whichever of the two planes it takes place, 

 has never yet resulted in the formation of two distinct 

 leaves. Hence, in whatever way the matter be en- 

 visaged, one leaf only is involved in the phenomenon. 



Comparison of the Sporophyll of Ophioglossacese. 



The sporophyll of the Ophioglossacea? has been 

 probably derived from an ancestral type in which both 

 portions, and not merely the ventral one, were fertile, 

 as seems indicated by the abnormal specimens of 

 Botrychium in which sporangia occur on the sterile 

 blade (probably a reversionary phenomenon). 



The view is here maintained, as indeed Chrysler has 

 recently clearly shown, that the fertile ventral spike 

 of the adder's-tongue fern (Ophioglossurn) is the result 

 of the fusion of two lateral fertile leaflets across the upper 

 surface of the sporophyll by their external margins. 

 If now we imagine, as a morphological (even though it 

 may seem far-fetched) possibility, this fertile spike 

 becoming fused along the midrib of its parent-blade, 

 its two parallel rows of sporangia becoming changed, 

 by the solution of the septa, into two elongated 

 sporangial sacs, while similar sacs occurred on the 

 margins of the main blade, the typical anther-structure 

 would be attained. But normally (and note the im- 

 portance of the fact itself) it is only the stalk of the 

 "spike" which has become thus fused. In fact, the 

 two median anther-loculi maybe regarded, on one view 

 of the matter, as homologous in their origin with the 

 fertile " spike " of Ophioglossurn. This view of the 

 matter has been set forth by Celakovsky. 



