GREGORY: FISH SKULLS 229 



II 



Hemibranchii 

 A B 



Macrorhamphosoidea Aulostomatoidea 



Macrorhamphosidae Aulostomatidse 



Centriscidae Fistulariidae 



Amphisilidae 



III 

 Lophobranchii 



Solenostomidae 

 Syngnathidse 



The contrasts in the shoulder-girdle between groups I and II are rather radical and 

 tend to support the views of Jungersen and Starks that we have to do with two distinct 

 orders of tube-mouthed fishes of possibly diflFerent origins. Nevertheless both groups have 

 a dermal plate (wrongly called "interclavicle" by early authors but shown by Swinnerton 

 (1905) to be a neomorph in this group), which fuses with the lower border of the coracoid. 

 Macrorhamphosus apparently has the least specialized pectoral pterygials, from which those 

 of Fistularia and of Gasterosteus may have specialized in opposite directions. The lateral 

 line plates over the shoulder-girdle in Aulorhynchus in Division I, according to Starks, are 

 homologous with those in Macrorhamphosus of Division II. 



In short, while there is doubtless a deep gap between Divisions I and II, there seem to 

 be too many indications of ultimate community of origin to justify referring one branch 

 to the scorpsenoids and another to a distinct order of quite unknown relationships. After 

 considering the matter with some care, I incline to the opinion that the characters cited 

 by Starks in favor of relationship of the entire series to the synentognaths and Percesoces 

 outweigh the points of resemblance to the scorpsenoids cited by Jungersen. In this con- 

 nection we may cite the following passage from Swinnerton (1902, pp. 580-581): 



"To the best of my knowledge, the nearest approach to this order [Thoracostei] among 

 living fishes is made by the Scomberesocidas. Indeed, so close is this approach that on a 

 consideration of the head skeleton alone one would be almost obliged to place Belone in 

 the same sub-order with Gasterosteus. Give its cranium an arched instead of a flattened 

 roof, replace its alisphenoid by overlapping frontal and parasphenoid processes, shorten 

 the premaxillae and mandible to a normal length, elongate the symplectic still further, and 

 it would be extremely difficult to find any feature of importance in which the two crania 

 differed, for in the B'lone all the roofing bones are sculptured; in spite of its lowly affinities, 

 its opisthotic is absent; the ethmoid, though more cartilaginous, is of the same type; the 

 branchial apparatus is an exact replica of that in Gasterosteus in the number and nature of 

 the basibranchials, in the number, shape, and proportional size of its pharyngobranchials, 

 and in all other features except the fusion of the vestigial elements of the fifth arch. Again, 

 the hyomandibular is of the same shape, though its articulations are more generalized; the 

 metapterygoid is equally reduced; one pterygoid line alone is present; the palatine is small, 

 edentulous, and lacks a maxillary process; finally, it presents the acrartete condition. 



"The similarity is so great that one may say with considerable truth that the little 

 stickleback is but a slightly specialized Belone. 



