GREGORY: FISH SKULLS 



213 



mandibular rami are slender, loosely united at the symphysis and connected by a broad 

 distensible membrane (Tate Regan, 1912c-, p. 348). Nearly all the usually associated 

 bones have disappeared; there are no palatopterygoids, operculars, branchiostegals, while 

 the gills are much reduced and placed far behind the jaws. 



If it were not for the last character and for the fact that the gulpers pass through a 

 'leptocephalus' larval stage, one might regard them as degraded derivatives of some long- 

 jawed gonostomid like Cyclothone. But detailed comparison of the skulls does not support 

 this hypothesis. Also the 'leptocephalus' of the gulpers is very unlike those of the isos- 

 pondyls and more like those of certain eels. Thus it would seem at first sight that the 

 gulpers must have been derived from some big-mouthed eel type, as Boulenger (1910, p, 

 603) suggests. In a phylogenetic system of classification the gulpers would in that case 

 be placed in the order Apodes, instead of standing by themselves in the order Lyomeri. 



iig.nuchae m spin. dors . ,, 



msp.Lrfnt 



Fig. 94. Gastroslomus. Data from Nusbaum-Hilarowicz. 



On the other hand, Tate Regan {I9\2b) has made out a strong case for the derivation 

 of the gulpers from some of the predaceous Iniomi, such as Synodus, the skull of which 

 seems to afford an ideal starting-point for the peculiar specialization of the skulls of Sacco- 

 pharynx and Gastrostomus. The crania of these forms are progressively widened and made 

 more solid for the attachment of the enormously long jaws and pharynx (Fig. 94). They 

 are also much shortened anteroposteriorly as a result of the forward displacement of the 

 eyes, shortening of the snout and forward and upward thrust of the hyomandibular. The 

 skull of Cetomimus regani as described by Parr (1929, pp. 24—27) might indeed make a 

 still better starting-point for that of Gastrostomus if we assumed only that the final shorten- 

 ing and widening of th-e neurocranium took place pari passu with the excessive elongation 

 of the jaws. According to Regan (1912f, p. 349) Saccopharynx is piscivorous but Gastro- 

 stomus probably feeds mainly on small invertebrates. 



