most other Sphoeroides . Colomesus has 19 vertebrae, as 

 does S. annulatus, while formosa has a more specialized 

 reduced number of 17. In Colomesus medial prongs of the 

 prootic are present representing the remains of the dorsal 

 roof of the myodome, although these are less well- 

 developed in psittacus than asellus, while neither S. an- 

 nulatus nor formosa retain these remnants, a slightly 

 more specialized condition. In both S. annulatus and for- 

 mosa there are eight dorsal rays and seven anal rays, 

 while Colomesus has a perhaps slightly more gener- 

 alized number of 10 or 11 rays in these fins. 



In short, there are striking similarities between the os- 

 teological makeup of Colomesus, S. annulatus and for- 

 mosa, and the conversion of an annulatus-formosa-\ike 

 fish into one like a Colomesus would require only the 

 retention of remnants of the myodome and a few more 

 dorsal and anal fin rays along with the further anterior 

 prolongation and recurving of the sphenotics and the 

 development of a second lateral line, short segments of 

 which are present in some specimens of many species of 

 Sphoeroides, including S. annulatus and formosa. It 

 seems likely to me that Colomesus arose from the same 

 line of Sphoeroides radiation as that which gave rise to 

 annulatus and formosa from a testudineus-\ike ancestry. 

 Perhaps this happened somewhere between the an- 

 nulatus and formosa levels of organization, with 

 Colomesus becoming far more specialized than either an- 

 nulatus or formosa. Fraser-Brunner (1943) also con- 



Figure 266.— Dorsal views of skulls of: 

 left, Torquigener pleurogramma, 113 mm S 

 Australia; right, Amblyrhynchotes piosae, 

 33.8 mm SL, Australia. 



sidered Colomesus to be a derivative of a Sphoeroides- 

 like stock, but for unstated reasons, which I would guess 

 were zoogeographic. 



While Lagocephalus and Colomesus are relatively easi- 

 ly distinguished from the ancestral Sphoeroides- 

 ( including Guentheridia here and following) -like stock, 

 the relationships of the Indo-Pacific Amblyrhynchotes, 

 Torquigener, and Fugu are less clear. Fraser-Brunner 

 (1943) said that Sphoeroides has a long ethmoid, with 

 the frontals well removed from the premaxillary pedicels, 

 17 vertebrae, and no lower lateral line, while Ambly- 

 rhynchotes, Torquigener, and Fugu have the ethmoid 

 shorter, the frontals reaching far forward, close to the 

 premaxillary pedicels, 20 to 21 vertebrae, and a lower 

 lateral line. 



Sphoeroides does tend to have a longer ethmoid, a 

 longer ethmoid-vomerine region, and the frontal further 

 removed from the anterior end of the ethmoid-vomerine 

 region than in Amblyrhynchotes, Torquigener, and 

 Fugu, in most species of which the portion of the 

 ethmoid-vomerine region that is exposed dorsally is rela- 

 tively short so that the frontals do closely approach the 

 anterior end of the ethmoid-vomerine region. However, 



