DISTRIBUTION OF GROWTH-FORMS. 133 



to the eminences on their surfaces. These seem to be controlled by some definite law which 

 we express by saying that there is a tendency for the eminences to repeat the form of the 

 stock. This shows at least that the differences are not purely accidental and therefore 

 negligible, and that our investigations have to be greatly extended before we are in a 

 position to make any definite statements as to their physiological relationships. 



D. Columnar. 



I am not certain whether there is any true columnar growth-form among these Atlantic 

 and West Indian Porites. The following may, however, be provisionally placed under this 

 heading : — 



P. St. Thomas 1. The thick stems of this slowly branching form suggest its being 



placed here. 

 P. West Indies x. IS. This again, the " P. macrocephala," D. and M., may be pro- 

 visionally placed here. 

 P. Alessandria 2 (?) 



E. Branching. 



In this Volume we have figured some forty-four branching Atlantic and West Indian forms, 

 and these, with figures published elsewhere, give us a list of close upon sixty growth-forms to 

 be compared and grouped. We have already noted that the usual group names, clavaria, 

 furcata, and divaricata, refer only to the character of the branching, and are far too crude and 

 insufficient even for that purpose alone, * to say nothing of the many other characters which 

 such a method utterly ignores. But while it is easy to say what method of classifying will 

 not do, it is by no means easy to sketch a method which is much more satisfactory than those 

 we reject After many attempts we fall back upon the metameric growth principle especially 

 prominent in these many branching Porites. We can classify by the characters of the colonial 

 units, so far as we can recognise them as sections forming the stock almost as links form a 

 chain. 



It will be noted that this does not depart far from our previous method of referring all 

 growth-forms back to an ideal initial plano-convex colony which grew up centrally to form 

 columnar and branching stocks. The difference between the methods is, however, obvious. 

 Under this earlier method the subsequent growth was, as it were, uncontrolled by any known 

 principle ; now we know at least that some principle of repetition guides the growth. This is so 

 far helpful that it enables us to picture to ourselves somewhat more sharply and precisely the 

 initial colony of each growth, and to layjdown a provisional classification according to what we 

 judge to have been its most striking characters. Every one who has tried to classify corals 

 will find this a considerable relief, because many specimens seem to start from broken frag- 

 ments or overturned specimens. Hitherto there has been no possibility of knowing what 

 was the normal form of the stock. We of course know of individual cases of overturned 

 Madrepores and Turbinarias, which at once repeated their normal forms from the bases or 

 sides of the overturned stocks, indeed from any part so long as it was uppermost. But these 

 few cases hardly seemed sufficient to justify our assuming that any stock growing in this way 

 returned instantly to its normal method. It is a gain therefore, to have, for the future, a good 

 reason for believing this always to be the case. 



* This insufficiency has already been felt by Dr. WaylaiiJ Vaughan as related, p. 11. 



