706 THE TETRAODOXTOIDEA GILL. 



tidw to constitute a single group of his family called Oymnodontes. He 

 subsequently made another retrograde step and reverted to the Liu- 

 mean conception of the genus, combining all in one genus, Tetrodon. 



Meanwhile the work of Hollard remained unknown or neglected. In 

 1884, Gill recognized this work, utilized his material for a systematic 

 revision of the group, elevated the Linujean and Giintherian genus 

 Tetrodon {Tetraodon) to superfamily rank, and divided it into three 

 families distinguished by osteological characters. In the framework 

 thus reared, Jordan and Edwards, in 1887, intercalated all the species 

 of American Tetraodonto idea, hnt degraded the group to family rank 

 and the families to subfamilies. 



Jordan and Edwards's memoir is excellent and the first one in which 

 scientific principles were applied at the same time generally and in de- 

 tail to the consideration of the group. The range of the genera, so far 

 as the American species are concerned, appears to be natural and their 

 nomenclature nearly correct. There are some questions involved, 

 however, which merit reconsideration and I propose to now pass in re- 

 view all the genera that appear to be well established. 



III. 



1758) TETRAODON. (Linnteus. 



Jordan and Edwards have reached the same conclusions as I for- 

 merly did respecting the application of the name Tetraodon, but by a 

 different route. I should not have considered it necessary- to dis- 

 cuss the question involved if a fallacy had not found way in their 

 argument which might be repeated in various other cases. Under the 

 circumstances, a review of their argument may be useful, I first 

 give their own words : ^ 



The name Tetraodon first appears as a generic term, so far as kuowu to us, in Has- 

 selquisi's Travels in Palestine, edited by Liniueiis, in 1757. We have not examined 

 this work, but our impression is that it is binomial in form, and that the name Tetra- 

 odon is here associated only with Tetraodon fahaka, the Tetraodon lineatus of the Sys- 

 tema Natura*, and a member of the group called Arothron. 



If this work of Hasselquist be taken in consideration, the name Tetraodon must be 

 assigned to the i»resent group. 



But the couventional starting point of binomial nomenclature is later, and in the 

 Systema Naturie, Liun:i'us includes all the si)ecie8 of the present family known to 

 him, in the genus Tetraodon. 



We must then consider the later attempts at restriction of the group. 



In 1839, Swainson made the first attempt at generic division. 



Retaining the name Tetrodon for the bulk of the species (including our genera 

 SphoToides and Tetraodon), he separated from it Lagocephahis, Leiodon (or Leisomus), 

 Cirrhisomus, and Canthigaster (or Psilonotus). 



The first and last of these were well defined. The others, Leiodon (based on the 

 absence of prickles) and Cirrhisomus (based on the presence of cirri), rest on charac- 

 ters of no systematic importance. Under the generic name of Tetrodon, four species 

 are 'ueutioned a-( types. Three of these belong to the Arothron group ; the other is a 

 Chelo nodon. But none of them are Linna^an species, although one of them {testudl 



