130 Mr. H. M. Bernard on the 



away from the branch. It is, in reality, a kind of forking, 

 only the stem remains the more important and less diverging 

 prong. The result of . repeated branchings with free fusions 

 iDetween parts that touch is to form a rather closely matted 

 tangle low down near the ground, the meshes in the tangle 

 being more or less angular. This angular character of the 

 meshes is, however, frequently obscured by curvings of the 

 branches. Broken fragments falling down into the tangle 

 freely fuse on again, and help to make the net thicker. In 

 claiming this very peculiar method of growth as characteristic 

 of the genus I am aware that it is not immediately evident 

 in all the types. It is very marked in Ridle} 's original type 

 {A . Forbesi) , in Quelch's types (A. gracilis and A. soUda), 

 and in one of the new types {A. echinulata '^) , whereas it is 

 not so marked tliough traceable in A. erecta *, and apparently 

 least visible in Eehberg's type {A. spinosa). In these last 

 two forms the branching does not come off at such a wide 

 angle, and hence the whole corallura is more symmetrically 

 arborescent. But in A. erecta, so far as I remember the 

 photographs shown me by Dr. Marenzeller, the larger clumps 

 were very close tangles of thin knotted stems, and Rehberg's 

 figure of A. spinosa (I. c.) appears to show distinct traces of 

 a tendency to sudden angular bendiugs of the stems and 

 branches. 



These points, then, the protuberant calicles, showing 

 distinct lamination of their radial structures, and the peculiar 

 character of the branching, serve, I think, to separate Ajiacro- 

 pora from Montipora, with which genus it is, however, funda- 

 mentally associated in the structure of the coenenchyma and 

 in the presence of calicles with degenerate septal apparatus 

 exactly like those of Montipora. 



Interrelationships of the Madreporidffi. 



As we have above seen, the only argument for allying 

 Montipora with Forites, as was done by Milne-Edwards and 

 Haime, and later by Duncan, falls to the ground as soon as 

 the secondary character of the trabecule is established f. 

 Hence we have no hesitation in claiming the genus with its 

 ally Anacropora as together forming a subfamily of the 

 Madreporidffi. I shall now endeavour to show that the 

 remaining three accepted genera — Madrepora, Turbinaria, 



* Full descriptions of these are given in the Museum Catalogue. 



t In 1889 Dr. Ortmann suggested, without going into details, that 

 Montipora might be deduced from Porites through Alveopora (Zool, 

 Jahrb. (syst.) iv. p. 584). 



