GENEEAL EEMAEK8. 17 



B. Axial corallites 3 to 4 mm. diameter. 



a. Axial corallites hemispherical or short cylindrical, 1 to 2 mm. exsert. 



b. Axial corallites long cylindrical, 3 to 5 mm. exsert. 



C. Axial corallites 2 to 3 mm. diameter, short cylindrical or hemispherical. 



D. Axial corallites 2 to 3 mm. diameter, long cylindrical, 3 to 6 mm. exsert. 



It will be seen that divisions C and D correspond to the two sections of division B 

 excepting that the axial corallites have a smaller diameter. 



The only other form of classification which need be noticed is that adopted by Ortmann 

 which to some extent is a compromise between those of previous authors. Ortmann in his 

 paper on the Corals of the Strassburg Museum, discusses shortly the classifications of previous 

 authors, and considers that of Dana preferable to that of Milne-Edwards and Haime ■ in this 

 I agree with him. The arrangements of species which he adopts, both in the paper referred to 

 and in the later one on the Ceylon Corals, were, like Klunzinger's, evidently only intended as 

 an aid to the identification of the species recorded in the respective papers. Ortmann follows 

 Dana in his main divisions, excepting in one or two cases, in which I think his system appears 

 to lose rather than gain by the change. In his subdivision of the main groups, however, he 

 appears to have selected characters which in many cases bring together groups of species 

 which are undoubtedly closely related. For instance, in considering those species with axial 

 corallites of a given diameter, he distinguishes between those which are cylindrical and those 

 which have a thick wall and rounded margin. Again, the corymbose species are divided into 

 two sections : in the one the branchlets are " catkin-like" (" katzchenformig "), with spreading 

 scale-like corallites ; in the other the branchlets are what Dana termed "spiciform." 



It may be well to consider separately the various characters which have from time to 

 time been employed in a subdivision of the genus Madrepora, and endeavour to ascertain 

 their present value for the purpose. In doing so it must, however, be distinctly understood 

 that it is premature to attempt a classification based on the morphology of the polyps ; and 

 the only characters at present available depend on the variation of the corallum and the 

 corallites which compose it, together with such guesses at the conditions by means of which 

 these structures are produced as may appear to receive support from the evidence already 

 obtained. 



1. Habit. — In certain cases allied species have undoubtedly a similar habit : compare, for 

 instance, the various species referred to Lepidocyathus and the majority of those referred to 

 lyiopora. Again, the majority of the truly arborescent species seem closely related to one 

 another, though others which are not arborescent probably belong to the same group. From 

 a consideration of habit alone Dana would not have arrived at his section G, which was 

 founded to include M. labrosa, M. securis, and M. cuneata, otherwise he would have in- 

 cluded M. palmata, M. cyclopea, and possibly M. conigera. This was done by Milne- Ed wards 

 and Haime, and the artificial result attained by the consideration of habit alone is at once 

 realized. Dana's group G remains good to-day and received the name Isopora from Studer ; 

 the essential feature of the group is not made clear in the definition, but that is a point 



D 



