27 



defined, and themfelves furrounded by a pellucid iimbus fo entirely refembling 

 that of many of the Ctramia, as to give them every appearance of true capfules. 

 I kept the fpecimens feveral days, but could not oblerve any feeds efcape from 

 them, nor have I fince been able to dilcover any tiling at all fimilar in either 

 this or any other fpecies of OJci/latoria, and fubfequent difcoyeries have encreafed 

 my fufpicions that they were not capfules, and have induced me to believe that 

 C. decarticans, as well as the other fpecies of this family, are propagated by 

 feed in a different manner. 



In examining fome fpecimens of C. dijlorta, I obferved a number of detached 

 globules of the fame color, and of about equal diameter with that of the fila- 

 ments, and I alfo obferved that in fome filaments which were partly empty, the 

 remaining joints had affumed a fimilar globular fhape. Some of the detached 

 globules had become of an oblong form, and a diffepiment was then obfervable 

 in the middle, while others were more elongated with four joints, and others 

 were ftill longer, fo as to form a regular feries, beginning with the globule, and 

 ending in a perfect filament. I have, therefore, no doubt, fo far as it is poffible 

 to ftate any opinion on objects fo minute and obfeure without any doubt, that 

 each joint at length becomes a feed, which efcapes at the apex of the filament, 

 and that by its evolution the fpecies is propagated. I have obferved a precifely 

 fimilar appearance, in C. mirabilis, and have alfo feen detached granules, ap- 

 parently of the fame nature among the filaments of C.fontinalis and C. muralis, 

 and in both thefe fpecies, thofe filaments which are partly empty have their 

 remaining joints of a more globular form, than they are in thofe which continue 

 perfe£t. 



I have afcertained that the filaments of C. diflorta conjugate in a lingular 

 manner, (which, together with the fructification, is reprefented in my fupple- 

 mentary plate A) and that the fuppofed ramifications of this fpecies are thus con- 

 flituted. C. d.Jlorta is therefore moft clofely allied with C. mirabilis and C. 

 mnjuscula, and I incline to the opinion that here, as well as in the Qpnjttgata, 

 an union of their filaments is in fome manner effential to their fructification. 



d 2 



