82 Mr. 0. T. Regan on the 



the head ; epiotics meeting behind the 

 supraoccipital ; no epipleurals ; 2 or 3 



pectoral pterygials 30 6. Lophloidei. 



§ No post-teiuporal ; pectoral arch attached to the vertebral 

 column 31. Opisthomi. 



The synopsis given above is a modification and amplifica- 

 tion of one which formed part of a paper on teleostean clas>i- 

 fication read to the section of Systematic Zoology of the 

 International Zoological Congress at Boston in 1907, and 

 which will no donbt be published at some future date. 



Later on I hope to give more detailed accounts of the 

 anatomy and classification of some of the less known group=<, 

 but witiiin the limits of the pr^'sent paper only a few brif;f 

 notes on some points which need elucidation are possible. 



ISOSPONDtLI. 



In some external cliaracters Retropinna is intermediate 

 between Osmerus and Prototroctes. R:tropinna, Salanx, and 

 Microstoma are Argentinidte which have no mesocoracoid. 

 The Argentinidsp, Ilaplochitonidge, and Galaxiidae are 

 extremely similar in osteology, dentition, and in the absence 

 of oviducts, and are undoubttdly closely related. 



It is possible to maintain the order Isospondyli, with the 

 addition of the Haplochitonidaj and Galaxiidae, by taking 

 into consideration the mouth-structure, the maxillary entering 

 the gapp- to a greater or less extent (almost excluded in 

 Haplochitonidje), and the unpaired ethmoid. As thus 

 defined, the Haplomi, Inionii, and Microcyprini are excluded. 



The cretaceous Enchodontidje fall into the division Stomia- 

 toidt'i ; they agree with tlie Stomiatidfe in the structure of 

 the skull and of the mouth. 



The Kneriidje, known to me from external characters only, 

 show considerable resemblance to the Gonorhynchidse, and 

 may pertain to this order. 



Heteromi. 



Boulenger has pointed out tliat the Lipogenidae are inter- 

 mediate in fin-structure between the Halosauridge and Xota- 

 canthidje. These three families constitute the order Hete- 

 romi, from which I would exclude the Dercetidse, in my 

 opinion probably belonging to the Iniomi, and the Fieras- 

 feridfe, which are without question specialized Brotulidae. 



In skeletal characters Balosaurus and uSotacanthus agree 

 in that the orbito-rostral part of the cranium is elongate, the 

 parietals meet, opisthotics, basisphenoid, alisphenoids, and 



