ee ARCHEOLOGY IN BRITISH GUIANA 239 
TABLE I.—Durations of Taruma Phase sites derived from the sherd refuse 
accumulation formula 
Total Number of Total Number of 
sherds Estimated | occupations sherds Estimated | occupations 
Site perl-by1-| duration | implied by Site peri-by1-| duration | implied by 
meter cut in years Seriation meter cut in years seriation 
or test of levels or test of levels 
E-9, Cut 1__- 1, 490 128,8 2 || E-17, Cut 1.- 241 20,8 1 
E-19, Cut 2_. 1, 210 164, 5 _ 2 || E-16, Cut 2. 199 17,2 1 
E-30, Cut 1. 8 77,2 3 || E-15, Cut 1. 193 16.7 1 
E-24, Cut 1. 763 66, 1 1 || E-3, Cut 1.__ 180 15, 6 1 
E-1, Cut 2... 762 66, 0 3 || E-23, Test__- 161 13,9 iT 
E-10, Cut 1._ 717 62,0 2 || E-22, Test A. 150 12,9 1 
E-13, Cut 1. 592 51,2 2 || E-14, Test._- 139 12.0 1 
E-7, Cut 2__- 582 60, 3 1 || E-4, Test_--- 106 9,1 1 
E-1, Cut 1... 526 45,5 1 |) E-16, Cut.1.- 97 8,4 1 
E-3, Cut 2__- 524 45.3 1 || E-29, Test__- 88 7.6 } 
E-6, Cut 1___ 454 39, 2 1 |] E-5, Cutgl. -- 45 3,9 1 
E-7, Cut 1__ 441 38, 2 2 || E-18, Test__- 44 3.7 1 
E-9, Cut 2___ 300 25.9 1 || E-11, Test_-- 35 3.0 1 
E-19, Cut 1. 287 24,8 2 || E-28, Test..- 28 2.4 1 
E-21, Test_.- 276 23.8 1 
The problem of computing the total duration of the Taruma Phase 
from the individual site durations is complicated by the fact that 
many of the villages were partly or wholly contemporary, judging 
from the interdigitation of levels of the various cuts on the seriation 
chart (fig. 101).. Such repetition must be eliminated from the data 
used to arrive at the total duration of the Phase. Several attempts 
were made to do this by inspection, and one of the results corresponded 
closely to the historically attested duration of around 200 years (see 
pp. 263-264 for historical details). However, the procedure was too 
arbitrary for objective description and the calculation could not be 
repeated readily by another person. An effort was made to devise a 
more systematic method of making the selection of levels to be in- 
cluded in computing total duration by measuring ceramic change. 
Although levels representing contemporary occupations should ideally 
show identical pottery type frequencies, in actual fact some variation 
always exists. In using the Taruma Phase data for computing the 
duration of the Phase, chronologically significant change must be dis- 
tinguished from this normal variation during a single time period. 
Since there was no way of determining in advance how fast the pottery 
was changing, and consequently how much difference in percentage 
frequency of the major pottery types reflected how much time differ- 
ence, the problem was approached experimentally in the following 
way: 
Beginning at the bottom of the seriation chart (fig. 101) and work- 
ing upward, the length of the bars representing the frequency of 
Kalunye Plain was measured, and the intervals where the difference 
between the bars reached 5 percent were recorded. Since there is 
a slight fluctuation in the trends from level to level, the curve some- 
times had to be artificially smoothed in order to eliminate local dis- 
tortion, but this problem was not severe. The levels included within 
each interval were considered to be contemporary. The sherd density 
