CELLULAR DEGENERATION AND FORMATION OF PIGMENT. 117 



of the ca3nosarc-tube. These breaks seemed to occur more 

 frequently towards the extremities of the branches than 

 in the main stem. The constrictions are formed in the 

 ccenosarc above and below a pair of hydrotheca? (PL XII, 

 fig. 8; text-fig. 6, a). A rough broken surface exposing the 

 endoderm is not present, but a complete covering layer of 

 ectoderm is formed, and the hydranths can be seen to be in 

 a living condition. Some hypertrophy of the ectoderm and 

 cf the so-called muscular sheet of the hydranth can be 

 observed (text-fig. 6, a). Ultimately the coenosarc and 

 hydranths succumb and the tissues are reduced to isolated 

 masses and strands of a black substance. 



A considerable portion of the ectoderm of the coenosarc 

 is greatly thickened owing to the multitude of interstitial 

 cells. The myo-epithelial elements form an outer epithelium 

 of large cells which rapidly taper inwards (fig. b). The inter- 

 stitial cells are particularly liable to pigment-degeneration 

 (fig. b), although it is also found in the thin ectoderm layer 

 of the hydranth (figs, c and d). 



The pigment granules formed in the ectoderm are of con- 

 siderable size (I'O jii), and they are dark brown or black. 

 They are of a peculiar character, consisting of little, flat, 

 round plates and of bacterium-like rods (figs, i and k). Some 

 zoologists may insist upon the view that these structures are 

 really parasitic organisms ; but they arise in the cells in a 

 manner quite similar to that in which pigment granules, 

 which do not mimic organisms, arise in other hydroids, and 

 accordingly I doubt whether they are independent organisms. 

 It is true that sometimes the rodlets may be seen in lines (fig. d, 

 middle of figure) as if the individuals had arisen by transverse 

 fission ; but the structures are black, the microchemical tests 

 for fungus-cellulose were inconclusive, and no iron was 

 detected. The size of the largest of the coccus-like granules 

 (1*0 /i) is about that of ordinary cocci, but the bacterium-like 

 bodies (l"3ju) are smaller than most bacteria. There was no 

 conclusive evidence of the division of the granules into chains 

 or " diplococci." Also, the granules vary greatly in size. 



