350 



PERCY SLADEN TRUST EXPEDITION 



maxilliped are primitive features. On the other hand, the uncleft condition of the 

 outer flagellum of its antennule is probably an exaggeration of one of the secondary- 

 features of the Pontoniinse. Its vestigial antennal scales and small rostrum are no 

 doubt adaptations to its habitat, and in any case are without parallel in the subfamily. 

 Its reduced gill formula might be expected from its habits, and need not indicate any 

 affinity with Conchodytes. Its heavy, compressed body with slender walking legs ending 

 in biunguiculate dactylopodites, its large chela, and, very distantly, its rostrum, rather 

 suggest Feridimenceus, but in view of other diflPerences this is probably a convergence. 

 It is perhaps best regarded as independently evolved from the ancestor of Periclimenes. 



Pontonia Conchodytes 



Pontonides 



Coutierea Stegopontonia 



\ / 



Coralliocaris 



Harpilius 

 Harpiliopsis 



Periclimenseus 

 Typton 



Ancyclocaris ? 



Pontohiopsis 

 Palsemonella 

 Urocaris 



Urocaridella 



On some difficult questions in Phylogeny. 



The making of phylogenetic trees, once a favourite pastime of zoologists, has of late 

 years come under the censure of superior persons, on the ground that we know too little 

 about the mode in which evolution takes place to speculate successfully as to its course. 

 Yet to represent in a graphic manner the lines of I'esemblance and difference within 

 zoological groups is a practice which dates from before the acceptance of the evolutionary 

 theory, and is not without its convenience even in these days. Moreover, the objections 

 to diagrammatic phylogeny apply also to any form of classification which is not purely 

 empirical, and especially strongly to the attempt so to construct " keys " that they 

 shall as far as possible represent real affinities. One of the difficulties which occur 

 in such cases is prominent in the Pontoniinse. It is presented by numerous instances 

 in which classifications according to two or more different features cross one another, 

 so that alternative arrangements are suggested. When one of these arrangements has 

 the advantage of illustrating also the verdicts of several further characteristics the 

 solution is easy, but this is often not the case. I have already described an instance 



