GENEEAL EEMAEKS. 19 



subject to variations dependent on the rate of growth &c. of the particular specimen, and in 

 many cases the normal surface is hidden by a secondary deposit ; and partly because the 

 actual density of the corallum is usually increased -with age by the secondary deposition 

 of lime. 



4. Degree of Coalescence of Branches. — This is a feature which appears to me to be of 

 little value unless associated with other characters. Even in species which at a late stage 

 of colony-formation exhibit a network of branches, or where the main branches become fused 

 into a solid mass, there is no doubt that at an early stage such confluence did not occur. 

 For example, there are several specimens of M. efflorescens in the collection of the British 

 Museum. The oldest forms a solid vasiform plate in which confluence of the main branches 

 takes place quite to the margin of the colony ; the youngest has obliquely radiating, non- 

 confluent branches with the habit characteristic of M. prolifera, Lamk. Again, the types of 

 M.fragilis, B.-Sm., and M. rambleri, B.-Sm., differ only in the latter being an older colony 

 than the former ; and owing to the manner in which the branches extend, they are bound to 

 become confluent if growth goes on long enough. Evidently, then, the coalescence or non- 

 coalescence of branches is an unreliable character, when taken alone, even for specific 

 purposes. 



5. Incrusting or Pedicellate Conditions. — Little need be said under this head : apparently 

 all species of Madrepora are incrusting in the first instance, and the extent to which incrustation 

 takes place is, in some cases at any rate, dependent on environment rather than on the 

 habit of the species or group of species. Large incrusting masses occur of species which 

 usually form cespitose or even arborescent colonies. The pedicellate condition has also been 

 already recognized as not constant in certain species. 



6. Size and Form of the Axial Corallites. — These characters were first used by Klunzinger 

 and aff'ord valuable assistance in the grouping of species. I am, however, of opinion that 

 size alone is of little value. In the primary sections of Klunzinger's system based on size 

 one often has difficulty in fixing the position of a species owing to the fact that the dimensions 

 vary beyond those of the section. It would not be difficult in the case of certain species to 

 find in one specimen some axial corallites which have the dimensions of one section and 

 others of another. Evidently the divisions 4-6 mm., 3-4 mm., &c. are quite artificial and, 

 though convenient for Klunzinger's purpose, cannot be made use of in a general classi- 

 fication. 



With regard to the characters which in the present treatise are regarded as the most 

 reliable for the subdivision of the genus, I have already stated that for the present 

 any attempt at a classification of the Madreporse must be based primarily on the 

 structure of the corallum. This, however, does not preclude an attempt to take into 

 consideration the structure and arrangement of the soft tissues in their relation to the 

 corallum so far as these are known, and from the basis thus obtained to endeavour to 

 distinguish skeletal characters which may be the outcome of a diff"erence in structure or 

 arrangement. For instance, in M. cervicornis, Lamk., the corallum is openly reticulate in 



