284 BULLETIN OF THE BUREAU OF FISHERIES. 



by the broad naked preorbital, a character well shown in ( ninthf-r's plate. The l)lue or w hitish spnts 

 are generally conspicuous, especially on the sides of the head. 



We can not see that Pomacenlrxis vitianus Sauvage differs in any respect from PomaceiUrus lividuii. 



Life colors of various specimens were noted as follows: 



(1) From Pago Pago. Violet-black, with three obscure black cro.ss-bands; eye bluish; a violet 

 stripe below eye to snout; a violet shade on opercle; a faint dark blotch at base of pectoral; no spot, 

 either black or white, in axil of dorsal; fins all black. 



(2) Same locality. Black body and fins; axil with a black spot extending on base of pectoral; 

 anal edged with violet-blue; tins all black. 



(3) From same locality; specimen called i'limna (white tail). All black, unspotted; axil and 

 broad bar across base of pectoral black; edge of dorsal, anal, and caudal dull whitish. 



(4) From Apia. Grayish black mottled with sooty gray; a round black opercular spot; no other 

 markings; ventral black. 



(5) From Apia. Dull olive brown, each scale posteriorly with a black spot; dorsal dusky, the 

 soft dorsal dull yellow at base, with a curved black bar behind it; anal similarly marked; caudal dull 

 yellow at base; pectoral dull orange; ventral brownish. 



(6) From Apia; fins low. All grayish black; base of dorsal, anal, and caudal golden washed; 

 axil dark. 



833. Pomacentrus devisi Jordan & Scale, new name. South Seas. 



Puinaniitrua irijasrmtitf De Vis. Proc. Linn. Soc. N. S. \V. 1S84. -I.'.-J. South Seas; mime preoccupied. 



834. Pomacentrus onyx De Vis. South Seas. 



Pumacciitrus onyx De Vis, Proc. Linn. Soc. N. S. W. 1«84, 4.51, South Seas. 



ABTJDEFDUF" ForskM. 



(Ghi]>hko<lon Lacepede; Sleyastcs Jenyns; Chrijsiptera Swain.son; Euschistlodus Gill; Parngh/phi- 

 doihiii, Glyphidodontops, Hemiglyphidodon, and Amblyglyphidodon Bleeker. ) 



835. Abudefduf sordidus (Forskal). Ulavapua. Samoa; Hawaii; Tahiti; Eaiatea; Bonham I.; 



Fate and Makatea (Seale); East Indies. 

 This species, recognizable by its large size, by faint dark cross-bands, and the black spot below 

 the axil of the dorsal, is common throughout the tropical Pacific. It is abundant both at Hawaii and 

 Samoa, but is found among lava rocks only, not about the coral reefs. A specimen from Apia in life 

 had six dark cross-bands and the usual spot at base of last dorsal ray. 



o We have referred the question of the propriety of the use of Abudefduf instead of Gtyphisodon, and of the rejection of 

 Ahuhamrur, Farer, and other quasi subgeneric names of Forslial, to our friend Dr. Leonard .Stejneger. From a letter dated 

 November 23, 1904, we quote his view of the matter, with which we fully concur: 



" Forskal, we are informed, based his nomenclature and systematic arrangement principally on Linnaeus. 10th ed., 

 probably the only systematic work he had with him. It is therefore of importance to examine into the "subdivisiones" 

 of genera which Linnaeus employed. He has four kinds, for instance: 



1. Pe}-ca. p. 289. 3. Salmo, p. 308. 

 X. Pinnis dorsalibns 2 distinctis. X. TruttEe. 



XX. Dorso monopterygio. etc., etc. XX. Osmeri. 



2. Cimex. p. 441. XXX. Ooregoni. 



a. Apteri. 4. Gryllm, p. 42.i. 



b. Scutellati. X. Mantis. 

 r. Coleoptrati. etc. XX. Acridia. 



XXX. Bulla, etc. 

 Only the fourth kind of subdivision corresponds exactly to our subgenus, but the names of subdivision 3 are by common 



The question as to Forsk^'s names then resolves itself to which of these categories they are to be referred. If they 

 are used as subgenera, or in a subgeneric sense like category 3, we must accept them. 



It seems then to me that Abudefduf comes within category 4, Both Chwtodon and Acanlliiiru^ are in the singular, and 

 Abudefditf is in no manner distinguishable from them. 



On the other hand, Perca dentibus Louti. Daba, Scams derUibus Sidjan, Abu djubbe, Harid, fall under category 1. 



The subdivisions of Scisena (p. 44) seem to me to be very much of a similar character, or, rather, they correspond 

 almost exactly to the modem way of subdividing "eommode" such large genera according to groups of species, which we 



