of the British Museum I offer my grateful thanks for his most valuable criticism and sueeestions 

 during the writing of this paper ; and to all the officials of the Botanical Department I 

 express once more my warmest acknowledgements for the continuation of that kindness and 

 courtesy they have always shewn me and for a share of the ready help and interest which 

 they extend to all visitors to the Department. 



Systematic. 



Up to the present time the species of Halimcda have all been founded on external 

 characters, with the exception of one species, H. matrophysa Ask. and one variety H. Opuntia 

 Lam. var. macropus Ask., which are both based on the size of the peripheral cells as seen in 

 surface view. Consequently many species have been made and as many more might still be 

 made, for the form of the joints varies very largely and is rarely constant even in individual plants. 



Owing to the opportunity of examining large collections, I was able to link together 

 the existing species by so many intermediate forms as to 



prove the instability of external characters as a basis of pfU r/t-y J J ^] C ) 

 classification. Indeed the connecting links were so complete 



. . [oints of H. Opuntia : all from one plant. 



that ït would have been necessary, using these characters 



alone, to reduce the number of species to two ; one having flat, unribbed, more or less discoid 



joints, and the other, joints ribbed and trilobed. 



It was however possible that some anatomical character might be found, sufficiently 

 stable to enable one to divide sharply into species this mass of apparently ^ 



confused forms. To this end the size of peripheral cells, depth of calci- \J A-/ V? 



fication and chromatophores were compared. None of these characters \J \j \y 



proved any more stable however than the external form, but at last a / B c 



comparison of the filaments of the central strand at the apex of each A - H - s racilis - 



1 B. //. incrassata. 



joint shewed a marked difference even in plants which were externally c - H - °P"> : '^'- 



very similar. As has been said above, communication between the central filaments as a whole 

 was shewn by Professor Askenasv to take place in H. incrassata and H. macroloba by means of 

 openings or pits; (figs. 34, 46.) but an examination of apparently typical specimens of H. Opuntia 

 Lam., the commonest and most variable species of the genus, shewed an entire absence both 

 of the pits and of the consequent welding of the strand into one connected mass. On the 

 contrary, at the apex of each joint where they naturally become pressed together owino- to 

 constriction, it was seen that communication was the result of a breakino- down of the side 

 walls between each pair of adjacent filaments. The fusion is however not complete, for in the 

 node the two partially fused filaments are again quite free and continue so in the next joint, 

 branching at intervals, till they again come into contact at its apex (fig. 25). 



This mode of communication between the central filaments was found to be constant 

 in most of the plants ascribed to H. Opuntia, but it was also found in the types themselves 

 of H. cordata, H. Renschii and H. triloba. These species were, in the various collections, all 

 connected by intermediate forms with II. Opuntia, so that for want of any distinguishing feature 

 I have been compelled to place them all under H. Opuntia. 



SIB0GA-EXPED1TIE LX. 



