598 THE AMERICAX NATURALIST. [Vol. XXXVI IL 



result of leaving aberrant forms in the group nearest to them is to 

 destroy all possibility of exact limitation. It is at the best impossible 

 to frame exact definitions when transitional groups of all grades 

 exist. A few orders can be isolated and defined among the bony 

 fishes, but the group as a whole cannot be divided into orders or 

 suborders or any other categories which shall have even approxi- 

 mately equal value. 



There is no sharp line separating the Isospondyli or Salmon- 

 herring series from the Ganoids, from which they are derived. From 

 the IsospondyU or Malacopterygii, as Dr. Boulenger calls them, all 

 other bony fishes seem to be descended. But the line between the 

 Isospondyli and the Haplomi is a narrow one. The Haplomi have 

 lost the mesocoracoid bone, and as it is abortive in many species 

 (Synodontidae, Myctophidae) hitherto attached to the Isospondyli, 

 these are placed by Boulenger among the Haplomi. Among the 

 Haplomi, on the other hand, are placed the spiny-rayed Percopsidae 

 because these have not lost the air duct characteristic of the Iso- 

 spondyli and their allies. 



The eels are divided into two suborders, —the Ostariophysi 

 remain together as one. The Heteromi are made to include Derce- 

 tidae, Halosauridae and Fierasferida. 



The transitional types with abdominal ventrals, the air duct and 



differ from each other in definition although comprising different 

 categories of families. The first of these is the suborder of Catos- 

 teomi, comprising the Selenichthyes (Lampris), the Hemibranchii, 

 the Lophobranchii, and the Hypostomides. The present reviewer 

 sees no reason for associating Lampris with this group. Its suborder 

 (Selenichthyes) is worthy of independent recognition. The others 

 certainly belong together, and almost as certainly are degenerate or 

 specialized allies of the other transitional group of Percesoces. 



In Boulenger's scheme, the Percesoces include not only the Athe- 

 rinidae Sphyraenidae-Mugilidai series, but also the Scombresocidae, 

 the Polynemidae, the Chiasmodontidse, the Tetragonurida, the Stro- 

 mateidae, the Ophiocephalidae, and the AnabantidiB. As thus arranged, 

 the group is defined only by the imperfect attachment of the ventral 

 fins to the thoracic arch. It cannot be exactly defined and its mem- 

 bers are very divergent. It is fair to say, however, that the fault lies 

 with the fishes, not with the classifier. The transitional elements be- 

 tween soft-rayed and spiny-rayed fishes are all still extant, and their 

 intergradations defy classification. For this reason. Dr. Gill once 



