228 
PACIFIC SCIENCE, Vol. 1, October, 1947 
The abundant series of G. am plus from 
the Rancho La Brea provided ample speci¬ 
mens for an adequate series from one local¬ 
ity of known age; no sub-Recent specimens 
were included. The Pleistocene tar pits may 
have operated over a period of several thou¬ 
sands of years, but of all the known avian- 
bearing horizons these are perhaps the most 
accurately restricted chronologically. The age 
factor was minimized by selecting well- 
ossified and completely fused bones for both 
the Recent and fossil series. Sexual varia¬ 
tion is of minor importance in these condors 
because it was found (Fisher, 1946: 547) 
that the sexes of the Recent species do not 
vary significantly in their skeletal measure¬ 
ments; hence, in the present study the sexes 
were lumped. With these factors causing 
variability eliminated, any variation in the 
am plus series could result only from a cer¬ 
tain inherent variability in the homogeneous 
stock, or possibly from inclusion of hetero¬ 
geneous stocks trapped in the oil at various 
times during the Pleistocene. To measure 
this variability the coefficient of variation 
was calculated for each skeletal element 
studied. 
Measurements and ratios of the several 
bones are presented in the tables. In these 
can be found some 3,000 measurements that 
were made with dial calipers. The range, the 
mean and its error, the standard deviation 
(o-) and its error, and the coefficient of 
variation (V) are given in Tables 1 to 14. 
Slight discrepancies are present in the third 
place of some of the figures because many 
of the computations were of necessity made 
with only a slide rule. The magnitude of 
the standard deviations was used to check 
the adequacy of the sample when compared 
to the observed range and mean. This check 
is based upon the fact that an observed 
range equal to 6a approximates the real 
range and that an observed range of 4a prob¬ 
ably includes about 95 per cent of the real 
range, whereas an observed range of 2a is 
about equal to 68 per cent of the actual 
range. The means were compared in various 
ways. For example, the formula d/a was 
used in comparing the means of large 
samples, and the familiar "t” test was used 
in comparing smaller samples. In a few 
instances both tests were applied to a single 
element. The value a d may be obtained in 
two ways (Simpson and Roe, 1939: 192- 
193). Formula 1: 
is designed to test the hypothesis that both 
the means and the variances are likely to be 
the same in the populations from which the 
two samples are taken. Formula 2: 
is used to determine the possibility that the 
two series differ significantly in the mean of 
a variate, and it assumes that the sample 
estimates of variance are near the true popu¬ 
lation values. Needless to say, a significant 
test from either formula indicates that the 
populations are likely to be different. Be¬ 
cause some workers prefer the first formula 
and others the second and because it is of 
considerable interest to note the varying 
results when the two formulae are applied 
to the same data, both formulae were used 
in this study, and the results of all the sig¬ 
nificance tests are given in Tables 15 and 16. 
In addition to this statistical analysis, a 
careful qualitative comparison was made of 
each element available for the two species. 
This later comparison, however, proved 
valueless, for no skeletal elements other 
than those of the skull could be identified 
on the basis of qualitative characters alone. 
Thus the present paper is essentially a study 
of absolute and relative measurements, their 
variance, and their importance in identifying 
