the Cystocctrp in Rhodomelctceae . 
301 
Conclusions. 
It is clear from what has been said that the four species 
examined present a remarkable uniformity in the structure of 
the procarp at the moment of fertilization ; the fifth and last- 
formed pericentral cell facing the axis giving rise to three 
branches 
(a) A four-celled lateral carpogonial branch ; 
(р) A two-celled lateral sterile branch ; 
(с) A one-celled inferior sterile branch. 
After fertilization there is also complete correspondence up 
to the formation of carpospores : — 
(a) The carpogonial branch is shunted off and withers. 
(1 b ) The two-celled lateral sterile branch branches again 
and becomes four-celled. 
(c) The one-celled inferior branch adds a cell. 
( d ) The fifth pericentral cell, now the auxiliary cell, shuts 
off a superior sporogenous cell. 
(c) The central cell gives off laterally two cells from which 
numerous (about twelve) paranematal filaments are derived, 
converging to the pore, and forming an imperfect lining. 
The divergences come in when spore-formation has pro- 
ceeded to some length 
(a) In Rhodomela subfusca no absorption of neighbouring 
cells by the sporogenous cell takes place at a late period in 
the formation of spores, if at all. 
(b) In Polysiphonia nigrescens and P. fastigiata such 
absorption does take place, extending to the auxiliary cell 
and the sterile branches. 
(c) In P. violacea the absorption extends to the central 
cell. 
The paranematal layer is probably constant in Rhodo- 
melaceae. I have seen it in Polysiphonia sertularioides , 
P. byssoides, P. urceolata , and Chondria tenuissima , in addition 
to the species mentioned above. 
To revert now to the comparison of these species with 
