eae ARCHEOLOGY IN BRITISH GUIANA 11 
sherds on this basis, recognizing such widely divergent temper ma- 
terials as cariapé, crushed potsherds, crushed shell, sponge spicules, 
crushed steatite, mica, or less distinct differences such as fine sand 
and coarse sand. 
When temper variations are not present or do not show a consistent 
trend of change through time, the problem is to find some feature or 
combination of features that will. Under Tropical Forest conditions, 
in contrast to those in the Andean and Mesoamerican culture areas, 
the surface erosion of the sherds is always sufficient to make surface 
treatment unusable. Consequently, a choice of a primary basis for 
classification must be made from the characteristics manifested by 
the paste. Aside from temper, firing differences are easiest to recog- 
nize and define with consistency. It has been our experience that 
separation of sherds into those that are completely oxidized, produc- 
ing a totally orange, tan, or brown cross section, and those that are 
incompletely oxidized, leaving a gray core of variable width, fre- 
quently provides a basis for seriation. After the primary features 
of the sherd types are established, the type categories are refined and 
other features are sometimes discovered to be additionally significant 
in showing change through time, such as rim shapes, vessel shapes, 
or even at times surface treatment. 
In making these classifications of plain pottery types we are not 
concerned with the significance, if any, of the type distinctions to 
the potters (in fact, we are not sure how this could be determined 
on pure archeological materials without ethnographic correlations), 
but only with the ability of the types to reflect change through time. 
If the pottery types from the successive levels of a strata cut do not 
show consistent trends through time, the first classification is aban- 
doned and another is attempted. It should be emphasized that the 
classification finally presented is not superior to other possible classi- 
fications in theory, in the criteria employed, or in terms of ceramic 
technology, but only in its ability to reflect chronology. 
During the preliminary classification all decorated sherds were set 
aside to be handled later, and the sites and levels of strata cuts were 
seriated on the basis of the plain ware trends. Subsequently, the de- 
corated sherds were examined by level and classified into types, but 
with decorative technique taking precedence over paste characteris- 
tics as the basis for classification. Theoretically, it would be feasible 
to combine the two criteria but in practice this often works to multiply 
the number of decorated types (sometimes reduced to a few sherds 
per type due to the scarcity of decorated sherds) with no compen- 
sating advantage. For example, such a classification would create 
four decorated types in the place of Akawabi Incised and Modeled 
in the Mabaruma Phase, since the same decoration was applied at 
513186—60——3 
