520 GILBERT 0. BOURNE. 



ovoid body has become somewhat flattened, and a process of 

 mesogloea, apparently formed by the cells contiguous to it, 

 stretches downward towards the mesogloeal lamina, to which, 

 however, it is uot yet united. A somewhat later stage is shown 

 in fig. 18, in which there are several such mesogloeal processes, 

 apparently in course of formation from the cells overlying 

 them, and the final stage is shown in fig. 19. In the last- 

 named figure a shallow cup with irregular margins and with 

 strise deeply stained with picro-carmine is seen to be attached 

 to the mesogloeal lamina by a band-like process of the latter. 

 To the edges of the cup is attached an external cuticular 

 membrane, stained blue by picro-nigrosin. The desmocyte 

 shows different staining reactions according as the section has 

 been stained for a short or long time in picro-nigrosin. It 

 always stains in picro-carmine : if it is left for a short time in 

 picro-nigrosin, the carmine is not displaced ; if for a long time 

 the picro-nigrosin displaces the picro-carmine and the desmocyte 

 and its striations are stained blue. The same is the case with 

 the similar structures in Madreporaria. 



I was long inclined to attribute to these structures the pre- 

 dominant share in the formation of the corallum, the more so 

 because their earlier ovoid shape and their final cup shape sug- 

 gested that they were similar to goblet-cells, the cup shape 

 being the expression of a goblet whose contents had been voided. 

 But further study convinced me that this was not the case. 

 The desmocytes are most abundant and best developed in places 

 where coral growth is least active ; only their earlier stages are 

 to be observed among the groups of calicoblasts in regions of 

 active coral growth. They are scarce relatively to the granular 

 calicoblasts, and in my flat preparations of hard and soft parts 

 at the growing points of the corallum I could not detect any 

 of them amongst the numerous calicoblasts clothing the ob- 

 viously newly formed points of coral. Moreover I am of the 

 opinion that the corallum is formed as a result of the disinte- 

 o-ration of calicoblasts. This is suggested by fig. 20, and fig. 21 

 is a drawing made from a partially decalcified preparation 

 stained with eosin and methylene blue. The calicoblasts are 



