E d 
ake a ARCHEOLOGY IN BRITISH GUIANA Zot 
cultivation. The excessive size of some former clearings suggests that 
the Taruma followed the same practice as the present-day Wai Wai 
in progressively extending the field (figs. 80, 83-85). Where the land 
area is sufficient, the Wai Wai field is first cleared next to the new 
house. As the yield begins to decline, the adjacent area is planted. It 
is only after suitable land in the immediate vicinity has been ex- 
hausted that a more distant location is farmed. This stage may be 
accompanied by the moving of the village if the distance is great 
enough. That the Taruma followed a similar practice is suggested by 
the fact that field clearings with habitation refuse tend to be larger 
than those not associated with villages. Only one independent field 
clearing reached 200 by 100 meters, whereas nine of those adjacent to 
villages were as large as this or larger, and only four were smaller. 
A puzzling factor in a number of the sites is the thickness of the 
sterile layer at the surface. This was frequently 8 cm. All of the 
sites are on high places, wel] above maximum flood level, which rules 
out the possibility of silt deposition subsequent to habitation. All 
are on the summit rather than the slope of the hill, eliminating sheet 
erosion as the explanation. This sterile overlay is present sporadically 
in sites throughout the sequence, rather than the early ones alone, 
indicating some factor other than time is responsible. Inquiry among 
botanists and geologists in the British Guiana Government service 
brought no clear-cut solution to this problem. Although it would be 
of interest to know the cause, the answer is not essential since the re- 
sponsible factor apparently is unrelated to the Taruma Phase culture 
or its antiquity in the upper Essequibo area. 
Seriation of the various strata cuts reveals that a number of the 
sites were occupied more than once. The large discrepancy in seriated 
position between Cut 1 and Cut 2 at Sites E-3, E-7, E-9, and E-16 is 
accountable only by this interpretation (fig. 101). If the whole site 
had been simultaneously inhabited, the pottery type frequencies in 
the cuts should overlap as they do at E-19. Further evidence in 
favor of the conclusion that reoccupation is involved comes from the 
spatial relationships of the cuts at the sites listed above. They are 
at, opposite ends of what was interpreted in the field as a single area 
of habitation refuse, except in the case of E-16, where the presence of 
two separate refuse areas was recognized. The ceramic seriation 
shows that the latter represent two different occupations of the site 
rather than the remains of two contemporary house structures. In 
the other three sites, the locations of the first and second occupations 
appear to have overlapped to some extent, giving a continuous sherd 
area. If a single excavation had been made in the center, it should 
have revealed this situation by an unusually wide spacing between 
successive levels of the cut. Two examples of this are provided by 
