318 ANNUAL, REPORT SMITHSONIAN INSTITUTION, 1919. 
maintained. More people would mean greater pressure for frequent 
clearing and planting of the land. Areas that became too grassy for 
cultivation would be considered worthless, and probably would not 
be protected against the fire. There would be a gradual extension of 
the grass-covered areas and a corresponding reduction of the lands 
that could be cleared and planted. Even cultivated lands and slopes 
that have been improved by a regular system of terracing are invaded 
by a coarse bunch grass (Epicampes) in the high plateaus of Gua- 
temala (pi. 15, fig. 2.) 
PERIODS OF REFORESTATION. 
After a district has been abandoned so that the woody vegetation 
is allowed to grow and grass fires become less frequent, there is a 
gradual return to forest conditions, though the advance of the woody 
vegetation into a grass-covered area is a slow process and subject to 
frequent setbacks as long as people remain to start fires. But even 
the recurrence of fires may not keep the forest from making slow, 
gradual gains at the expense of the grasses. Some kinds of trees 
have thick, nonconducting, slow-burning bark or other protection 
against fire, and are able to compete with the grass and finally to 
overcome it. The fires kill most of the young trees, but a few sur- 
vive, with accidental protection of stones or ant hills. The grass is 
thinner around the trees and the fires gradually become less de- 
structive. Finally, when there are enough trees to shade out the 
grass, genuine forest conditions are reestablished (pi. 14, fig. 2). 
In many parts of Central America it is difficult to find any 
primeval forests or any that are old enough to represent the original 
condition before clearing began. The forests that are found in 
swamps, deserts, and rocky, precipitous places, too rough for clear- 
ing and planting by the native methods, may represent the only 
original growth. Even places that are difficult of access may be 
drawn upon for supplies of firewood or for making charcoal. That 
there are any large areas of truly virgin forest growth in Central 
America has still to be shown. Even in the rainiest districts of the 
eastern lowlands of Costa Rica the clearing of heavy forests for 
banana plantations has resulted in the discovery of abundant pre- 
historic remains. 
During the long era of prehistoric development of agriculture and 
civilization in tropical America many periods of agricultural occu- 
pation may have alternated with periods of abandonment and re- 
forestation, which would account for the presence of different kinds 
of ancient pottery and stonework in the same districts, (PI. 9, fig. 1 ; 
pi. 10, fig. 1.) Though the modern Indians of eastern Guatemala are 
much afraid of caves, the ancient inhabitants used them generally as 
