June ii, 1891J 



NATURE 



'31 



on the Challenger Tunicata, but further work since — 

 on some very extensive collections from Australian seas 

 and on the Ascidians of our own coasts — has convinced 

 me that the only rational explanation of the protean 

 forms and labyrinthine inter-relations of the Ascidians 

 is to be found in regarding the group as one in process of 

 evolution, where many of the species, genera, &c., have 

 not yet become markedly differentiated by the elimination 

 of intermediate forms, and where the animals are so much 

 at the mercy of their environment that a special pre- 

 mium is set upon useful characters (if, indeed, there are 

 any "specific" characters which are not useful), and where, 

 consequently, the relations between modification of struc- 

 ture and conditions of existence brought about by the 

 action of natural selection are exceptionally evident. 

 Adopting, then, this view, the following difficult subjects 

 of dispute, and probably others with which I am not 

 concerned at present, can be, I think, satisfactorily ex- 

 plained : (i) the connection of the Simple with the 

 Compound Ascidians, and the classification of the latter; 



(2) the value of some modifications of the branchial sac ; 



(3) the position of the Polystyelidae ; (4) the relations 

 between the sub-families and genera of the Cynthiidas ; 

 and (5) the numerous " species " of the genus Botrylkcs. 



(i) If the attempt is made (as in most classifications) 

 to regard the Compound Ascidians as a group distinct 

 from the Ascidias Simplices, and forming either a parallel 

 or a divergent line in regard to the latter, one meets at once 

 with the serious difficulty that the Compound Ascidians 

 show affinities with the Simple at several distinct points. 

 Three investigators approaching the Compound Ascidians 

 after the previous study of certain Simple Ascidians — say, 

 the first fresh from Ciona, Ecteinascidia^ and Clavelina, 

 the second from Perophora, and the third from Styela 

 and Polycarpa — could each make out a good case for the 

 view that his new subjects were most closely connected 

 with the genera he had just been working at. The first 

 could demonstrate the undoubted relations, in external 

 form and in structure of branchial sac, between Clavelina 

 and Chondrnstachys, Colella and the other Distoniida; ; 

 the second might point to the similarity (on which I per- 

 sonally lay no stress) of Perophora and the Botryllidas, in 

 the relations of ahmentary canal to branchial sac ; and 

 the third could show the close similarity between the 

 Styelin^e and the Compound forms Synstycla, Goodsiria, 

 and Chorisocormiis in nearly every detail of internal 

 structure : and all three would be partly right, and 

 therefore unlikely to agree upon any one system of 

 classification. 



But when the attempt is made seriously to form 

 a conception of the past history or evolution of the 

 forms in question, it becomes obvious that the Com- 

 pound Ascidians are not a natural, but an artificial 

 group. That is, they are not the whole surviving de- 

 scendants of a single group of ancestors, but are poly- 

 phyletic in origin, being derived from several distinct 

 lines of ancestry which have arisen independently from 

 different kinds of Simple Ascidians, and have since ac- 

 quired the common characteristic of being able to re- 

 produce by gemmation so as to form compact colonies in 

 which the members (ascidiozooids) are embedded in a 

 common test or investing mass. We know with as much 

 certainty as we know anything in such phylogenetic in- 

 quiries that the ancestral Tunicates were not colonies, 

 and that reproduction by gemmation was not a primitive 

 character. This property has, then, been acquired 

 secondarily by some ancestral Simple Ascidians, and may 

 very possibly have been acquired more than once (though 

 this is not at all necessary for my theory of the poly- 

 phyletic origin). It follows from this view (which I have 

 expressed before, but now feel more certain of from recent 

 work), that if we are to retain the group Ascidiae Com- 

 positae, or S>nascidi3e,in our system, we must represent it 

 as linked on to the Ascidias Simplices, at three points at 



NO. 1 1 28, VOL. 44] 



least, and we must not attempt to arrange the families 

 and genera in a series diverging from any one of these 

 points alone ; or if we do, we need not be surprised when 

 we arrive at obviously unnatural arrangements which are 

 in conflict with the classifications of our fellow-workers. 



On the other hand, we might abolish the group Ascidiae 

 Composita; altogether as a sub-order of Ascidiacea;, on the 

 ground that it is not a natural group {i.e. a compact set of 

 descendants from a common ancestor — a single branch of 

 the genealogical tree). 



But if we adopt this course with the Compound Ascidians, 

 the same argument might be used in connection with 

 other polyphyletic groups throughout the animal kingdom. 

 They should all be broken up, it might be urged, as 

 being artificial assemblages. And thac would be a per- 

 fectly logical and definite position to take up, and one 

 for which a good deal could be said, but before adopting 

 it zoologists should remember that it involves a loss as 

 well as a gain. If it gives " the system " a certain preci- 

 sion, and an advance of a step or two towards the goal of 

 a completely natural classification, it at the same time 

 destroys the recognition of characteristics which certain 

 forms possess in common. In whatever manner they 

 have been obtained, there is no doubt that Compound 

 Ascidians of the present day possess certain features by 

 which they can be identified as Compound Ascidians, and 

 this fact is surely worthy of recognition in our " system." 

 My own opinion, then, is that the group Ascidia; Com- 

 positas should still be retained, but that its polyphyletic 

 origin and multiple connection with the Ascidiic Sim- 

 plices should be carefully borne in mind when drawing 

 up any scheme of classification, or discussing affinities. 



(2) Some of the ideas noted above, and others to be 

 discussed below, took definite form lately in reading a re- 

 cently published memoir by M. Fernand Lahille,i in which, 

 while giving a number of important original observa- 

 tions on the anatomy and bionomics of the Ascidians (and 

 especially of the Compound forms) of the French coasts, 

 the author introduces what I cannot help thinking in some 

 respects an unfortunate attempt to remodel the classi- 

 fication of the Tunicata on lines which he communicated 

 a few years ago to the French Association (Congrcs de 

 Toulouse, 1887), and now elaborates in detail. He 

 regards the branchial sac as the most important organ 

 in the Tunicata, and so it is in some respects ; but that 

 is not sufficient reason for regarding its modifications in 

 structure as the sole characteristics of the primary groups. 

 For example, the Appendicularians, instead of being called 

 Larvacea or Copelata, and characterized by the presence 

 of a tail containing the urochord, are placed in a group 

 " Atremata," defined by the absence of stigmata in the 

 branchial sac. The openings in question (stigmata) are 

 not even such important structures as the primary bran- 

 chial clefts (gill-slits), but are merely the secondary slits 

 placing the cavity of the branchial sac in communication 

 with the peribranchial or atrial cavity, and are of nothing 

 like such high morphological value as the presence or 

 absence of a urochord, and of the two primitive atrio- 

 pores, and the other well-known characteristics employed 

 in former classifications as distinguishing the Appendicu- 

 lariidae. Some of the Thaliacea are placed by Lahille in 

 a group (Hemitremata) of primary importance, by them- 

 selves, because they have the stigmata rudimentary or 

 imperfectly formed, while the other Thaliacea are united 

 with all the remaining Tunicata, because they are sup- 

 posed to be alike in having complete stigmata.- 



Then, again, an altogether fictitious value is given by 

 Lahille to the presence of internal longitudinal bars in the 

 branchial sac, especially since he shows (as had been done 

 by former writers) that these bars develop as outgrowths 



' " Recherches sur les Tuniciers des Cfltes de France " (Toulouse, 1890). 



* Which, however, is not really the case. The apertures in the walls uf 

 the branchial sac in Lahille's "Eutremata" are not always homologous 

 ■.tnicfiire'. In the genus Culeoltts, for example, there are no true stigmata. 



