ON THE BRITISH SPEOIES OF CEISIA. 161 



below the ovicell on the same side becomes free from the 

 internode (if any tubular aperture is developed), with the corre- 

 sponding point on the zooecium next above that zooecium on the 

 other side of the internode. The " length " of the ovicell is 

 the distance of the middle point of the line drawn as above to 

 the uppermost point of the ovicell, exclusive of its tubular 

 aperture, if any. 



This measurement, on the whole, gives the most constant 

 results. The relation of the zocEcia below the ovicell to the 

 ovicell itself is very variable, and it is impossible to take one of 

 these zooecia as a fixed point, a more definite result being 

 obtained by taking two as described above. In C. eburnea, 

 if the ovicell is the second member of an internode the 

 imaginary line is drawn from the base of the tubular aperture 

 of the first member of the internode in a direction parallel to 

 that connecting the apertures of the upper pairs of zooecia in 

 the internode (a method which usually gives more satisfactory 

 results thau might be supposed from fig. 6). 



The numbers given in the table do not profess to give the 

 total limits within which a given part may vary, but the limits 

 given will probably be found to include nearly all the variations 

 which are observed in the ordinary forms of colonies. 



In cases where it seemed to me possible to define the normal 

 size of any part, I have put this down as the " average " size. 



An examination of this table shows that C. eburnea, 

 C. aculeata, and C. ramosa form an almost continuous 

 series, as S mitt indeed has stated. C. denticulata can hardly 

 be confused with any of the other forms, amongst which 

 C. eburnea is the one that resembles it the most. 



It will be seen that in every single measurement given 

 C. ramosa is the largest of the four species, although in the 

 size of the entire colony C. denticulata surpasses it. In 

 many of the measurements, however, the upper limit of 

 C. aculeata overlaps the lower limit of C. ramosa. As 

 these are the two species which most closely resemble one 

 another, and which are probably very closely allied to one 

 another, it seems to me that it is not possible, in all cases, to 



