﻿42 4 



group of related forms. For this we might possibly retain the name Hemibranchii, 

 but this name is certainly somewhat misleading; it is due to Cope (6) and was 

 intended to denote that the forms thus indicated are remarkable for a certain 

 weakness and incompleteness in the branchial apparatus. This is true, however, 

 of not a single one of the forms which Cope (and after him Boulenger) included 

 under the Hemibranchii, and applies only to the Solenostomidce and Syngnathidœ, 

 which he did not include in this group. Whether now this natural series of 

 forms "Hemibranchii" (-4- Gastrosteidæ and Aulorhynchidæ) -j- "Lophobranchii" (Sol- 

 enostomidæ and Syngnathidæ) should be maintained as a suborder, and where it 

 should be placed eventually in the system, I shall not discuss in the present com- 

 munication, nor shall I enter here into the reasons why I remove the Gastrosteidæ, 

 Aulorhynchidæ and Pegasidœ and place them with (or near to) the Scleroparei; I 

 believe, in fact, that but little is gained by discussing the systematic position before 

 the structure of the forms in question is better known than at the present moment. 

 The truth is, indeed, that the descriptions of the structure of most of these fishes 

 are defective, for some naturally more defective than for others; but even for such 

 common forms as the sticklebacks and the pipe-fishes there are several features of 

 considerable importance which have escaped attention or have been misunderstood *. 

 In successive later communications I hope to give a description of the principal 

 characteristics of the structure of all the forms in question, at least of their osteo- 

 logy; in the present paper I shall deal exclusively with the two genera Amphisile 

 and Centriscus. I may however just mention here already, that this whole commu- 

 nity of forms shows a number of characteristics in the skeletal structure of the 

 head, which so far as I know are not found united in any other fishes whatsoever; 

 namely: 1. the parietals and opisthotics are wanting; 2. the pterotic (squa- 

 mosal) reaches ventrally to the base of the cranium, articulating there with the 

 basioccipital or also with the parasplienoid, and thus excludes the exoccipitals and 

 prootics from meeting; 3. the snout parts of the cranium (ethmoid and vomer) 

 are extremely prolonged forming a "beak"; 4. the under margin of this beak is 

 closely bound in the whole of its length with the whole of the upper margin of 

 the part of the suspensorium lying in front of the hyomandibular; in this way an 

 extremely characteristic tube is formed, which supports in front the true mouth 

 parts; 5. the palatines are short and only connected with the vomer (in Aulosto- 

 miim alone they are likewise in touch with the anterior end of the ethmoid); 6. 



* In the Biologia Centr. Americana. I'isces, just published (Februarj' 1908) Regan (pp. X — XI), 

 after having excluded tlie Hyposlomidcs and Selenichthyes from the Catosleomi Blgr., which suborder 

 he finds "unnatural and indefinable", says that "tlic remainder, whicli corresponds to the Hemibranchii 

 of Smith Woodward, is still a heterogenous a.ssemblage which I find incapable of definition, and includes 

 three well-marked but probably related groups wliich should, in my opinion, be given subordinal 

 rank." These are 1) Thoracostei (= Gastrosteidæ, Aulorhynchida\ Fistiilariidw and Aulostomidœ ; ;.') 

 Solenichtbyes (nom. nov.) (= Anipliisilidœ and Centriscidœ) ; 3) Lophobranchii (= Solenostomidæ and 

 Syngnathidæ). As stated above 1 do not at all agree in placing Gastrosteidæ and Aulorynchidæ together 

 with Fistulariidœ-Aulostomidœ, but these matters I shall discuss in a later paper. 



