than with proximity to Guatemala. In rebuttal, it may 
be stated with certainty that none of the samples were 
selected because of the altitude of their origin. That the 
high average knob number for Guatemala is not seriously 
in error is shown by Reeves’ comparison of his samples 
with those of Mangelsdorf and Cameron (23), for which 
data on altitude were given. Facts presented in the fol- 
lowing paragraph indicate that the relationship observed 
between knob number and proximity to Guatemala in 
samples originating in the United States is causal. It is 
possible that some selectivity for low knob number did 
occur in the Andean region because of the prevailingly 
high altitude there, but it is especially noteworthy that 
the inclusion of the data for this region did not strengthen 
the observed relationship but rather weakened it. It is, 
therefore, reasonably clear that the samples were taken 
from altitudes at random and that altitude was not per 
se responsible for the relationship observed. This conclu- 
sion is, of course, based in part on the assumption that 
there is no positive relationship between low altitude and 
proximity to Guatemala. If such a relationship does 
exist, it is not detectable in relief maps, and, in fact, it 
almost certainly is non-existent. All of these facts are 
given, and some of them are strongly emphasized, in the 
literature cited on this topic by Randolph. It is unfor- 
tunate that Randolph overlooked these additional facts 
when he raised his objections. 
Longley (18), working with corn of the United States, 
showed that there exists a pronounced relationship be- 
tween knob number and proximity to the Mexican bor- 
der, even when only the high altitude states of Montana, 
Idaho, Utah, Colorado, Arizona and New Mexico are 
taken into consideration. Or, if all of the states from 
which he took samples, regardless of altitude, are con- 
sidered, there is even then a relationship. Longley rec- 
[ 366 ] 
