296 U. S. NATIONAL MUSEUM BULLETIN 2 06 



strongly developed. Spines on the hands seem to develop only in 

 frogs above 30 mm. in length, but beyond that point the development is 

 relatively independent of the size and age of the male, since a 38 mm. 

 frog (the smallest having spines in the present series) has about 30 

 spines on the first finger, while the largest male, 77 mm. long, has only 

 about 50 spines, the maximum number of about 120 spines being 

 attained by a male 62 mm. in length. There is a well-developed patch 

 of spines on the pollex in all males; these vary between 7 and 20 in 

 number without regard to the body size. The second finger has between 

 3 and 13 spines scattered on all of its phalanges, while the third finger 

 may have as many as 10 spines, although occasionally they are entirely 

 lacking here. Sometimes the spines are pale and weak; this is true of 

 the largest male, collected in Nova Friburgo, on May 9, and of a 64 

 mm. individual, collected at Recreio dos Bandeirantes, on the coast, 

 on February 18; so it cannot be due to the time of year in reference to 

 breeding season, as the species is still breeding in February. Some of 

 the spines in all the finger patches are smaller than others; this seems 

 to indicate that new ones are growing all the time, as I have shown to 

 happen in the genus Crossodactylus. The number of spines in miliaris, 

 however, is apparently far more variable than in any of the known 

 species of Crossodactylus. 



The adpressed heel may reach beyond the snout, to tip of snout, to 

 nostrU, or to between eye and nostril. The first and second fingers 

 are often very nearly subequal, while the first is sometimes slightly 

 longer. This comparison is very difficult to make in males, as the 

 distal phalanx contracts at right angles in preservative, so that it is not 

 always possible to make an accurate measurement. 



In color pattern there is no fixity as to the T-shaped mark on the 

 head, as this pattern is found without regard to geographic distribution 

 in most of the specimens at hand. Some have the T reduced to a V, 

 and sometimes even this is reduced to a dark line representing its 

 anterior margin. The spots following this along the back are some- 

 times fairly rectangular, but are often very ii-regular, sometimes 

 light-centered. The thi'oat and anterior chest are usually brown with 

 numerous small white spots. In a few specimens only the throat is so 

 patterned, whUe in one or two others the maculation extends over the 

 anterior part of the abdomen. In two specimens from Sao Paulo, as 

 already noted, the entire venter seems to be white, but as this condition 

 occurs also in one example from Rio de Janeiro, while several others 

 from Rio are only slightly spotted or pale beneath, it does not appear to 

 have any geographic significance. 



The vomerine teeth are similar in position, although their degree of 

 development differs individually. Most specimens have them very 

 heavily developed, close together, triangular in shape, and lying 



