20 GENERAL REMARKS. 



section and the walls of the corallites are provided with prominent costulae, between which, 

 there is every reason to suppose, the longitudinal canals described by Fowler are situated. 

 In M. hemprichi, on the other hand, the corallum is much denser in section, the wall is not 

 costulate, but very finely echinulate instead ; and one may presume the absence of costulae 

 to necessitate, or rather to be the result of, a different arrangement of the extra-calicular 

 tissues. Evidently, then, if M. cervicornis a,nd M. hemprichi d'lSer in other skeletal characters, 

 one appears justified in placing them in different sections. I have assumed that a similarity 

 of structure involves a close relationship and have therefore, in the classification now pro- 

 posed, placed little reliance on habit. To a certain extent the species now described fall into 

 well-marked groups, if one does not regard habit as of prime importance. For instance, 

 Studer's subgenus Isopora, although connected with the arborescent types through such species 

 as M. bruggemanni and M. ortmanni, appears sufficiently isolated to justify its separation. 

 The species M. echinata, Dana, agrees with M. echidnma, M. subglabra, M. longicyathus, 

 M. speciosa, and M. horrida and others in the possession of a type of slender tubular 

 corallite not found outside the group. The nearest approach to this condition is found in 

 another group of species, M, tubulosa, M. rosaria, M. hydra, M. confraga, &c., characterized 

 by the presence of a thick-walled rod-like corallite with more or less rounded apex. Again, a 

 considerable number of species, agreeing more or less in habit, group themselves around 

 M. seriata, Ehrb., in the possession of a very thick and porous-walled axial corallite, which 

 is hemispherical or very slightly prominent at the margin. Whether such divisions are well 

 founded or not remains to be determined, but in any case they appear to be more convenient 

 than those previously proposed. I have thus to a considerable extent made use of the form &c. 

 of the axial corallites in many of the divisions now proposed, but in other cases, where the 

 resulting groups appeared too extensive, I have made use of other characters. A considerable 

 number of corymbose species have a short cylindrical type of axial corallite, which varies very 

 little in different species, not even to any considerable extent in diameter. Some have delicate, 

 curved, scale-like radial corallites, others labellate or nariform ones. The one type gives the 

 catkin-like appearance to the branchlets referred to by Ortmann, the other the spiciform type 

 of branchlet. M. millepora, Ehrb., will serve as an illustration of the former group, and 

 M. gpicifera, Dana, of the latter. If, however, one endeavours to realize the type of radial 

 corallite present in M. millepora, irrespective of its length, and then looks around to see if 

 this type occurs in any other species, it is found at once that in M. hebes, Dana, M. monti- 

 culosa, Briiggemann, and a number of other species, which differ entirely from M. millepora 

 in habit, the same type of corallite occurs. In such cases I have regarded unity of type in 

 the radial corallites as of more importance than habit for systematic purposes. If such an 

 arrangement serves to facilitate the identification of species its purpose will have been 

 fulfilled. It appears more convenient to use names to distinguish the chief divisions, but 

 I have not at present felt justified in considering them of generic value, although some 

 appear to merit that distinction. 



