Il6 PATAGONIAN EXPEDITIONS I PALAEONTOLOGY. 
of the known genera are tridactyl, but always retain a vestige of the fifth 
metacarpal, and are of mesaxonic symmetry, which is thus more constant and 
definite than in the Santa Cruz Typotheria, which have tetradactyl feet 
with either paraxonic or mesaxonic symmetry. The carpus is of the com- 
pletely interlocking type, the scaphoid articulating with the magnum and 
the lunar with the unciform, thus closely resembling the structure of the 
perissodactyls, notably the early rhinoceroses. On the other hand, the 
tarsus retains a highly primitive character ; the astragalus has a trochlea 
with very shallow groove, a short neck and convex head, which rests exclu- 
sively upon the navicular, being widely removed from the short cuboid, 
while the calcaneum bears a very large and prominent facet for the fibula ; 
the ento- and mesocuneiforms are coossified. The ungual phalanges, espe- 
cially that of the median digit, are broad and heavy and quite like those 
of such perissodactyls as Palceosyops in the Santa Cruz genera, and in 
the Pampean are more rhinocerotic in form. 
Classification of the Toxodonta. 
Ameghino has divided the Toxodonta into several family groups, the 
Toxodontidae, Xotodontidae, Haplodontheriidae, Nesodontidse, but, as he 
has predicted, this example will not be followed here. He complains that 
South America fossils are treated differently from those of other continents, 
saying: “Comme les faunes eteintes sudamericaines et celles des autres 
continents sont toujours jugees avec un criterium distinct, on pretend qu’il 
n’y a pas de raison pour separer les Nesodontidce des Toxodontidcz ; on en 
dira certainement autant des Haplodontheriidce, et on en fera probable- 
men t trois sous-familles de celle de Toxodontidcz ” (’07, 89). 
This complaint necessitates some remarks upon the significance of the 
family as applied to extinct organisms. Broadly speaking, there are 
among palaeontologists two contrasted methods of using the family group. 
One method, that of Cope, for example, is to apply the Linnaean concep- 
tion to the fossil forms and to treat the family purely as a matter of defi- 
nition, the fauna of each horizon being classified without reference to 
preceding or succeeding groups. The other method, first suggested, I 
believe, by Schlosser (’86) is to regard the family as a phylogenetic series 
and to include in it not merely the main stem of the series, but also such 
side-branches as are not themselves so distinct and so widely ramified as 
to constitute other families. Just how large and important a side-branch 
