32 MADREPOEARIA. 



succeeded, for the primitive form both of the calicle and of the colony can be fairly easily 

 reconstructed, as has been already described (see Sec. III.). Certainty, however, is out of the 

 question. Hence any system which is buUt upon this primitive form would partake of that 

 element of uncertainty, and should be only hypothetically constructed. It should not be 

 given to the world as a definitive classificatiou. This is not all. The discovery of the primitive 

 form and even the recognition of the various kinds of modification which have given rise to 

 all the known descendants of that primitive form, are easy tasks compared with that of 

 arranging these descendants into genetic groups. Granted that the primitive form may stand 

 more or less accurately revealed by a simple process of comparison — that is, when enough 

 specimens are brought into line — stiU the laws under which that form varied, and hence the 

 interrelationships of the resulting variations, remain hidden from us. The usual plan of 

 assuming that degrees of genetic relationship run parallel with closeness of resemblance will 

 not work among the corals. With highly complicated organisms it is more justifiable because, 

 the larger the number of characters, the less is the probability of close resemblance being 

 due to anything but close blood relationship. But even in these cases, we begin to tread 

 uncertain ground as soon as we come to the finer subdivisions based on one or two minor 

 variations. In the more primitive corals the organisation is simple, and all the variations 

 on the ground or generic type seem to be of minor importance and extremely unstable. 



As stated above, we are merely in a position of feeling our way ; the data for grouping 

 the forms into " species " are not yet forthcoming. Nor can we even group them morpho- 

 logically with any feeling of certainty. In order to make a satisfactory morphological 

 classification we have to decide which are the primary and which are the secondary characters. 

 In former volumes, although the word "species" was retained, the classification was 

 admittedly purely morphological, and it was assumed that the growth-form was the chief 

 character. But Mr. Pace tells me that the growth-form of TurUnaria depends largely 

 upon the nature of the substratum.* The primitive cup-form can only survive in clear 

 water free from sand and mud. And in the genus which we are now describing, the 

 analysis of the growth-forms given above (Sec. III.) shows that they have some intimate 

 dependence upon the form of the central calicles. But we do not yet know whether the 

 growth -form fashions the calicles or the calicles fashion the growth-form. It is certain 

 that they must profoundly influence one another. But until we know, we are unable to give 

 the first rank to either of them. And even if we were to make a plunge and fix upon one of 

 these two as the leading taxonomic character, its subdivisions would run into one another 

 and must inevitably be artificial. The attempt, then, to divide the specimens into an ideal 

 classification either genetic or morphological has to be postponed until the data are available. 

 And here let me say that it has been maintained t that we should have a better chance of 

 success in our attempts to form " species " in corals if we studied the soft parts. In the 

 first place, we can only study what we have got, viz. the skeletons, and in the second place 



* See Journ. Linn. Soc, xxviii. (1901) p. 358. 



t Mr. J. Stanley Gardiner, in his reply to my ' Unit of Classification,' Proc. Camb. Phil. Soc, 

 xi. (1902) p. 423. 



