ship is that at least some species of the latter have at 

 least poorly developed dermal flaps. 



In Laputa there is a full complement of three seg- 

 ments in the flexible encasing scale series and the fin-ray 

 rudiments, while small, are not minute. The dorsal lobe 

 of the pelvis is of moderate size and the dewlap just as 

 short as in Paramonacanthus. The most prominent 

 specializations of Laputa in comparison to the other 

 genera with flexible encasing scales and fin-ray elements 

 are that the forward extension of the supraoccipital has 

 been lost and that the encasing scale series is somewhat 

 reduced in size. I suspect that Laputa is on the same line 

 as that leading from Paramonacanthus to Pervagor, with 

 Laputa being a derivative of the forerunner of Pervagor 

 in which the dorsal spine had not yet migrated forward to 

 above the middle of the eye. 



In addition to Stephanolepis, Monacanthus, and Paro' 

 monacanthus, with fully developed forward extensions of 

 the supraoccipital bearing a relatively well-developed 

 crest, and Chaetoderma, with a somewhat shorter for- 

 ward extension and very shallow crest, two other genera 

 have a moderately developed forward extension and low 

 crest, a still rather generalized condition, and another 

 genus has the forward extension but no crest. These are 

 Alutera, Brachaluteres, and Psilocephalus, in all of 

 which the forward extension reaches to between the mid- 

 dle and almost the front of the eye, with the crest 

 relatively distinct in Alutera but only very poorly 

 developed in Brachaluteres and absent in Psilocephalus. 

 All three of these genera have highly specialized pelvic 

 apparatuses, with no dorsal lobe to the pelvis, no pelvic 

 fin-ray element, and either no encasing scales or a single 

 enlarged scale representing the remnant of the series 

 (some Alutera), even this being lost in large adults. 



The first dorsal spine is placed over about the middle 

 of the eye in Alutera and Brachaluteres, and only slightly 

 more posteriorly in Psilocephalus, while Brachaluteres 

 and Psilocephalus are further specialized by having lost 

 the small second dorsal spine, and Psilocephalus by the 

 great elongation of the body, increased number of verte- 

 brae and development of a long chin barbel. In fact, of 

 the genera studied here, I suspect that at least the ma- 

 jority of those without a dorsal pelvic lobe, with the en- 

 casing scale series essentially absent and with no fin-ray 

 element, form a natural group representing two diver- 

 gent lines of diversification from a more or less Alutera- 

 like ancestral group, which itself evolved from some 

 derivative of the Stephanolepis- Paramonacanthus level 

 of organization, perhaps from an early Rudarius -like 

 group that still retained the forward extension of the 

 supraoccipital, and which had the dorsal spine placed 

 over the middle of the eye and a tendency for the reduc- 

 tion of the fixed encasing scale series, such as is evident 

 in the two Recent species, R. ercodes with two segments 

 and R. minutus with only one. 



The Alutera-\ike ancestral group referred to above is 

 envisioned as being essentially like Recent Alutera, with 

 the exception of having a better developed spiny dorsal 

 fin basal pterygiophore more closely applied to the skull. 

 This ancestral group is seen as diverging into two lines. 



One line leads to Alutera and Psilocephalus by a 

 decrease in the size and closeness of association of the 

 basal pterygiophore with the skull, reduction in the size 

 of the first dorsal spine, and elongation of the body and 

 increased number of vertebrae, as found in some Recent 

 Alutera (vertebrae 20 to 23) and carried to an extreme in 

 Psilocephalus (vertebrae 29 or 30). In Psilocephalus the 

 basal pterygiophore is greatly reduced in size and only 

 closely attached to the skull anteriorly, while the first 

 dorsal spine is slender and delicate and the second spine 

 is lost, and the pelvis is reduced to a thin narrow shaft. 



The other line of evolution from the Alutera-Uke an- 

 cestral group would seem to lead to those genera with a 

 simplified pelvic apparatus (no dorsal lobe, no encasing 

 scales, no fin-ray element) but with a well-developed 

 spiny dorsal fin basal pterygiophore and a relatively well- 

 developed first dorsal spine. On this line is Bracha- 

 luteres, which retains the forward extension of the supra- 

 occipital and at least a remnant of the crest, as well as 

 the relatively low number of 20 vertebrae found also in 

 one species of Alutera, while the body depth is increased 

 rather than decreased. Part of the increased size of the 

 deep basal pterygiophore in Brachaluteres may be as- 

 sociated with this increased depth of the body and thus 

 of the muscle bands attaching to the sides of the pteryg- 

 iophore. The first dorsal spine of Brachaluteres is strong, 

 but the second spine is absent. Brachaluteres has the 

 same kind of elaboration and branching of the epi- 

 pleurals as found in some Alutera. 



Undoubtedly also on this line and closely related to 

 Brachaluteres is Paraluteres, with an equally low num- 

 ber of vertebrae, an even more reduced pelvis and neither 

 a forward extension of the supraoccipital nor a crest. The 

 basal pterygiophore, while strong and closely held to the 

 skull, is placed a little further back on the skull than in 

 Brachaluteres, and the dorsai spine is over the rear 

 rather than the middle of the eye, perhaps in conjunc- 

 tion with the resculpturing of the skull that took place as 

 Paraluteres developed its mimicry of Canthigaster valen- 

 tini, with it being advantageous for the dorsal spine to be 

 able to be laid back inconspicuously against the body 

 rather than jutting out prominently from the head. In 

 fact, the dorsal spine in Paraluteres is covered with thick 

 scaleless skin and in life is probably not capable of being 

 erected at a full 90° to the dorsal profile, even though the 

 second dorsal spine and presumedly also the locking 

 mechanism are present. 



Paraluteres probably evolved from the same ancestral 

 group as Brachaluteres, but before the second dorsal 

 spine was lost and the body depth so greatly increased. 

 From this same ancestral group from which Bracha- 

 luteres and Paraluteres evolved also probably came Oxy- 

 monacanthus and Pseudaluteres, with much the same 

 tendencies for elongation and increased number of verte- 

 brae as in the Alutera-Psilocephalus line, but with the 

 exact opposite trend with the basal pterygiophore and 

 dorsal spine, the pterygiophore remaining large and 

 closely associated with the skull, as in Brachaluteres and 

 Paraluteres. 



In Oxymonacanthus the basal pterygiophore of the 



