leathery glumes, while in the more recent levels in these 
areas there is a sudden influx of Tripsacoid cobs. Some 
of these closely resemble derivatives from maize-teosinte 
hybrids in having highly lignified, crateriform lower 
glumes which diverge at right angles from a highly lig- 
nified rachis, distichous arrangement of spikelets, and 
occasional single spikelets resulting from a partial or 
complete reduction of the second, or pedicellate spikelet 
(Mangelsdorf and Lister, 1956). This sudden appearance 
in maize of as many teosinte characters as one can get 
from hybrids of maize and teosinte cannot be explained 
as resulting from parallel or random mutation as Ran- 
dolph claimed. In addition, the evidence of prehistoric 
maize-teosinte heterosis, as revealed by the present study, 
suggests how teosinte germplasm has been maintained 
during recent evolution in maize. 
Conclusions from the extensive experimental work 
which has been done on the relation of teosinte to modern 
maize have a direct bearing on interpreting the archaeo- 
logical material described in this paper. The nature of 
teosinte germplasm has been described by Mangelsdorf 
and Reeves (1939) as consisting of multiple factor seg- 
ments distributed on several maize-like chromosomes, 
these segments having similar effects on the morpholo- 
gical characters which distinguish the ears of these spe- 
cies. It is also known that the various teosinte varieties 
themselves have both qualitative and quantitative differ- 
ences in germplasm affecting the teosinte characters 
(Rogers, 1950). This teosinte germplasm may be involved 
in heterosis, for its effects on maize may be beneficial when 
heterozygous and deleterious when homozygous (Man- 
gelsdorf, 1952). Furthermore, maize inbreds in Texas 
vary in their capacity to be improved by teosinte intro- 
gression, as demonstrated by Reeves (1950). This vari- 
ability also occurs in Mexican maize, since races grown at 
[ 119 ] 
