considered to have occurred independently in the two Uneages. However, my 

 observations do not confirm those of Ferraris, inasmuch as the presumably derived 

 condition of four infraorbitals is not uniformly present in all ageneiosids; in fact, 

 several species have five infraorbitals, and their shape and position varies somewhat. 

 In all species the lacrimal is relatively large, with two long anterior processes, a 

 short posteromedial process, and a long posteroventral process (Figs. 3-6, 9). 

 Variation among species occurs in the remaining three or four elements, and, in the 

 absence of developmental series, their homologies cannot be unequivocally 

 determined. The last infraorbital is relatively long and remote from the sphenotic, 

 and I assume that it is homologous among all species (Figs. 3-6, 11). This leaves the 

 observed variation attributable to infraorbitals 2-3 (numbered antero-posteriorly). 

 IaA.pardalis,A. n. sp., and Tetranematichthys there are five infraorbitals, with the 

 third being the smallest of the series, which, xnA.n. sp. and Tetranematichthys, is 

 reduced to a very short cyhnder that is just slightly longer than its diameter. I have 

 also observed five infraorbitals in specimens of^. atronasus,A. brevis, and A. 

 vittatus; in these three species, the fourth element is the smallest, and the third and 

 fifth surround the main posterior and ventral curve of the sensory canal. The 

 absence of a small posterior canal ossification, anteroventral to the sphenotic, was 

 considered by Ferraris (1988) to be synapomorphic for ageneiosids, although this 

 condition is homoplasious in outgroups, being found in doraidids, Asterophysus, and 

 Trachelyopterichthys (Britski 1972, Ferraris 1988). In some species, including at least 

 A. brevifilis,A. brevis, A. vittatus, A. ucayalensis, oxidA. n. sp., the second infraorbital 

 is flared at its posterior end, apparently where an anterior branch of the sensory 

 canal passes into the dentary. Ferraris (1988) considered this to be a synapomorphy 

 of Ageneiosus, but I have not observed it in either ^4. atronasus or A. pardalis. Apart 

 from obvious interspecific variation, I suspect that there may also be significant 





