FROGS OF COLOMBIA — COCHRAN AND GOIN 505 



from the north, but the material is not adequate to determine the 

 significance of this variation. There is considerable variation in 

 the degree of development of the humeral spine. Some of the males 

 have the humeral spine fairly well developed and can be distinguished 

 from G. antioquiensis by this character only with difficulty. Other 

 individuals, particularly old females, may lack any palpable evidence 

 of humeral spines. 



Superimposed over all the variations discussed above is variation 

 brought about by preservation and storage. This variation is rather 

 extensive, and some of the material is rather soft or otherwise inade- 

 quate for thorough study. 



The largest male at hand (MZUM 83652) measures 27.8 millimeters 

 in head-and-body length, while the largest female (CNHM 54722) 

 measures 33 millimeters. 



Remarks. — A search for the types of this species in the collections of 

 the British Museum produced one bottle (BM 80.12.5.201 [947.2.- 

 22.55]) containing soup and a few odd bones and another bottle (BM 

 78.1.25.16 [1947.2.22.54]) containing alcohol, the original label, and a 

 note that the specimen had been X-rayed in 1955. There are, however, 

 data concerning Boulenger's concept of this species in addition to his 

 type description and figure. Two well-preserved specimens (BM 

 1903.10.30.24, 1905.5.31.94) from Escorial, Venezuela, that were 

 identified by Boulenger as his H. buckleyi are so labeled by Boulenger 

 in his own handwriting. These specimens, which fit the original de- 

 scription and illustration perfectly, leave no doubt in our minds as to 

 the species described by Boulenger. 



Three centrolenids in the northern Andean region that have rather 

 uniform dark purple dorsa seem to have been extraordinarily confused 

 in collections. These are buckleyi, antioquiensis, and parabambae. The 

 confusion seems to have resulted in part from a lack of good specimens 

 for comparisons and in part from the widespread belief that the pres- 

 ence or absence of vomerine teeth in this group is without significance. 



After study of all available type material and direct comparison of 

 the three forms we now believe that each is distinct and that the 

 presence or absence of vomerine teeth serves as a valid character on 

 which to separate the species. Furthermore, Centrolenella buckleyi has 

 a different habitus from the others and usually can be recognized at 

 a glance. Both parabambae and antioquiensis seem to have somewhat 

 slender, tapering bodies, while buckleyi gives the appearance of being 

 wide and flattened. In preservation, at least, the midbody of buckleyi 

 usually is definitely flattened dorsoventrally and, more often than not, 

 is wider than the head. The other two species with dark purple dorsa 

 have more slender, graceful-looking bodies that are not flattened 

 dorsoventrally and that are seldom, if ever, wider than the head. 



