THE ANATOMY OF THE MADREPOBARIA. 11 



from the rest by running far deeper into the corallum or 

 rachis. This may be a specialisation for circulatory purposes, 

 as has been shown by Dr. Wilson to be true for certain Alcyo- 

 naria, or connected with production of the generative elements, 

 as is the case in Antipatharia ; in M. Durvillei certainly the 

 latter, perhaps also the former, holds good. 



Madrepora aspera (Dana). 



For a fragment of this coral, fortunately the upper part of a 

 branch, I am again indebted to Professor Moseley. 



The species was founded by Dana (7), who gives a good 

 figure of the colony. 



A. Corallum. — A trans vers esection of the corallum (fig. 10) 

 shows that the polyp cavities {a a') are arranged in a definite 

 ring, and not merely confined to three sides as in M. Durvillei, 

 round a central cavity into which project six septa, more or 

 less fused together at their free edges. This central cavity 

 (c. c.) is continuous with that of the apical polyp of the branch. 

 The arrangement of the internal longitudinal canals is not so 

 definitely concentric as in M. Durvillei, but the method of 

 circumferential growth of the corallum appears to be similar 

 in both species, since the costse appear to fuse over the external 

 longitudinal canals {v. fig. 10, z, and p. 3). 



In the apical polyps are found six distinct entocoelic septa, 

 and six smaller exoccelic, of which all are not always present ; 

 in the others generally only an axial or abaxial septum. A 

 similar difi'erence between them was observed by v. Koch (8) 

 in M. variabilis, where both exosepta and entosepta were 

 present in the apical polyps, but entosepta only in the rest. 



In this form, as in the former species, there appears to be 

 no relation in number and position between costse and septa, 

 the former being by far the most numerous. 



The costse are apparently formed as in M. Durvillei, 

 that is, at the points where the endoderm and mesoderm 

 apposed to the exterior surface of the corallum touch the 

 external body wall {v. p. 5 and fig. 4), but in both species, 

 owing to alcoholic contraction, the latter has so shrunk on to 



