50 ME. W. S. EOWNTREE ON THE 



CliaracinidfB seems to be rendered untenable by tlie fact that, Avhereas the resemblance is 

 closest in the case of the Erythrinoids, the actually most primitive conditions in certain 

 particulars are to be found in the highly modified herbiA'orous forms — conditions, indeed, 

 the homolo^ues of which have to be sought not amongst the Ganoids, but amongst the 

 more lowly Selachians. This fact appears also to negative the hypothesis which suggests 

 itself, that the Amia-like Erythrinoids represent the ancestral stock from which the other 

 two groups have diverged. 



Sao-emehl points out that the Citliarinoids (if I may for brevity so style the herbivorous 

 forms) present more primitive characters than the Erythrinoids in the following points : — 



(1) The greater extension of the cranial cavity towards the nasal region, and the 



associated possession of long olfactory tracts. 



(2) The less advanced ossification of the primordial skull, especially in the ethmoid region. 



(3) The simple undifferentiated socket for the hyomandibular. 



(4) The possession (in C/Marimis at least) of two " submaxillaries," ossifications which 



the author homologizes with the upper labial cartilages of Selachians. These are 

 not found in the other Characinid groups, nor in Amia, though they exist in 

 certain other Teleosts ( Gi/mnotus, Perca, some Cyprinoids). 



(5) The persistence, in Cithar'uuis and other herbivorous forms, of the epibranchial of 



the fifth arch, standing in connection with the fourth epibranchial, and serving 

 to support the accessory respiratory organ. A similar vestige has been described 

 in a Olupeid. In the other Characinids, and in Amia, it is unknown. 



With regard to the Carnivorous Characinidye (other than the Erythrinoids), Sagemehl 

 considers that the cranial conditions are entirely such as might conceivably have been 

 derived from the Citharinoid skull, but that the jaw -apparatus and the relations of the 

 hyomandibular suggest rather a connection with the Erythrinoids. 



On the whole evidence, therefore, so far as the skull region is concerned, the conclusion 

 to be drawn seems to be tliat the carnivorous (non-Erythrinoid) forms present the most 

 advanced conditions of development, whilst the Erythrinoids and Citharinoids are more 

 primitive, but that these two latter groups do not show any nearer approach to one 

 another ; in fact, that the three groups must have branched out independently from a 

 common stock. 



The African form Sarcodaces, however, classed with the non-Erythrinoid carnivorous 

 group, presents, as Sagemehl insists, most striking detailed resemblances to the Ery- 

 thrinoids in its cranial and facial characters, though separated from them by the 

 possession of an adipose second dorsal fin, and by the absence of teeth on the palatine. 



Moreover, the assumption that all the herbivorous forms are to be classed together — 

 that is, that the herbivorous habit has only originated once within the limits of the 

 family — would appear to be scarcely warranted by the examination of the single genus 

 ( Citharlnus) described by Sagemehl, though, it is true, he refers at times to others. 



The three groups or subfamilies marked out by that writer can, therefore, it seems to 

 me, be only accepted as, at naost, a provisional basis of classification. 



My own observations upon this most interesting and perplexing family of fishes have 



