58 PAPERS FROM THE DEPARTMENT OF MARINE BIOLOGY. 



The customary remarks about the unsatisfactory condition of the classi- 

 fication of the Madreporaria will be repeated. Certain of the groups, it 

 seems, are natural, for instance, the Fungiidae, the Eupsammiidse, the Acro- 

 poridae, and the Poritidae; but the classification of other groups, especially 

 the so-called Astraeidae, is in a highly unsatisfactory condition. Mr. Mat- 

 thai's is the last attempt at classifying them. Although he has rendered 

 valuable service in describing the anatomy of a number of corals and in 

 publishing many good illustrations, especially of the types of species, he has, 

 in my opinion, not only not established that the corals considered by him may 

 be classified according to their genetic relations on the basis of the character 

 of the directive mesenteries, but the placing of such species as Orbicella ver- 

 sipora, Favia stelligera, and Acanthastrea echinata in the same genus {Favia) 

 appears to me to invalidate such a method. I am therefore splitting his genus 

 Favia into three families, distributing the species among four genera, and 

 am suggesting that a fifth genus may be represented. The basis of the clas- 

 sification I am using should be made evident. Besides some characters 

 obvious to all students of the groups, I am utilizing the method of asexual 

 reproduction and the character of the septal margins. It seems to me that 

 each subgroup of corals may reproduce by gemmation and by fission. For 

 instance, Orbicella annularis reproduces by gemmation, while Favia fragum 

 forms new corallites by fission, but they are more closely related to each 

 other than is either to Mussa angnlosa, which in its reproduction resembles 

 Favia. According to my treatment, Orbicella and Favia belong to different 

 families within the same superfamily; whereas Mussa belongs to another 

 family, but, as at present only one family with its peculiar septal charac- 

 ters is known, there is no known group of genera to form a superfamily. 



The system of classification I am using is only tentative and needs 

 critical testing, as do all systems. A searching study of the septal structure 

 of all the mussoid genera and of many of the species is one of the essentials 

 for such a test; and the results of such an investigation should be compared 

 with the results from similar studies of the septal structure of Orbicella, 

 Favia, and Favites. Perhaps the criteria on which I am basing families will 

 not stand such a test. 



A classification to be valid must be phylogenetic, and for that we must 

 strive. M. Dollfus in 1906' closed a review of my paper on the simple genera 

 of the Madreporaria Fungida with the statement: 



"Nous ne ferons qu'une critique, unique mais importante, a cette classification, 

 c'est quelle n'a rien de phylogenetique; les especes et les genres de tous les temps 

 et de tous les pays y sont melanges, aucun caractere evolutif n'apparait. C'est 

 peut-etre que la perforation des cloisons^ qui lui sert de base principale, est au fond 

 un caractere secondaire, tandis que certaines modifications laissees dans I'ombre 

 devraient etre, au contraire, considerees comme des symtomes preponderants et 

 nous donneraient le tableau de la marche de la vie dans tout le groupe, ce qui nous 

 manque jusqu'ici absolument." 



'Rev. crit. de Paleozool., lo An., No. I, pp. 65-67, 1906. 



*See my paper on Hawaiian corals, pp. 127-128, 139-140. M. Dollfuss is correct in his surmise. 



