The method outlined above for determining the dimensions and centers of the various sectors that make up a site worked well on small sites and on sites with relatively short-term occupations. In the case of larger sites, however - in the 5 to 15 ha range - with very long-term occupations this method proved to be untenable. In order to collect detailed occupation information over the entire occupation span of these larger sites, comparable to that easily obtained for smaller sites, a program of intensive systematic surface collection was carried out. The methodology employed was the same as that I had used earlier at the site of Chiripa ([Bandy 1999d]) and, with Paul Goldstein, at the site of Chen Chen (M-1) in the Moquegua Valley of Peru ([Bandy et al. 1996]).
The surface collection was begun typically at the point of greatest
ceramic density within the site, as determined in the initial recording
of the site, as described above. Around this point, a circle with
a radius of 3.99 meters (that is a 50 m
circle) was scratched
into the surface, and all ceramics in this area were collected and
assigned an arbitrary and sequential number which I refer to as a
locus number. Locus numbers allocated for the surface collection program
were between 4000 and 4999, assigned arbitrarily, and were therefore
continuous with the locus numbers used in the various excavation seasons
of the Taraco Archaeological Project at Chiripa (0-3999). Each surface
collection locus, therefore, refers to a 50 m circular area
within a site with a center defined in terms of a master site grid.
This grid was laid out on every site in which we undertook systematic
surface collection. It was oriented to the cardinal directions (though
note that in the 1998-1999 surface collection program this grid was
laid out using a magnetic compass and without correction for declination;
North in this case is magnetic North). The central point of the site,
as defined above, was typically assigned the coordinates 1000N/1000E.
Additional surface collection loci were collected at 50 meter intervals
until the whole area of the site had been sampled (that is, until
the surface density of prehistoric ceramics had dropped below 0.1
sherds per m, equivalent to 5 sherds in the 50m circle).
Each locus, or collection unit, sampled an area of 50x50m or 0.25
ha. Thus, a 5 ha site, when surface collected, would comprise 20 collection
loci. The ceramic data resulting from these surface collections are
summarized in Appendix B.
All ceramics within each locus were collected, washed, sorted and analyzed; almost 100,000 sherds altogether. Though I would like to have collected all lithics from each locus as well, this proved to be impossible. The density of debitage in most sites, particularly Formative period ones, was such that systematic debitage collection would quickly have exhausted my resources for both analysis and curation. This being the case, I decided that only formal tools and debitage of non-local origin were to be collected; flakes and shatter of quartzite and local cherts were noted in a general way on the locus form but were not recovered. Faunal material was not collected due to the fact the rapid deterioration of bone when exposed to sunlight and to the elements and to the continued deposition of modern faunal remains on site surfaces. Metal objects of clear prehistoric origin were collected, though these were quite rare. Human skeletal material, encountered occasionally in disturbed contexts, was not collected out of respect for local sensibilities and appreciation of the evident curation difficulties. I wished to avoid associations with huaqueros or looters and therefore avoided human remains as much as possible.
All told, 420 loci were collected in the course of the work (in addition
to the 154 loci of the 1996 Chiripa surface collection programme which
were reanalyzed for the present study), comprising over 1 km
of systematically collected total site area. Even so, there were some
sites - T368 and T319 spring immediately to mind - that really should
have been systematically collected but were not due to lack of time
and money. With these few exceptions, though, every multicomponent
site in the survey area larger than 5 ha was systematically surface
collected. In addition, many smaller Formative period sites were collected.
This is so because it proved to be impossible to distinguish Early
and Middle Formative occupations without systematically collected
assemblages. Therefore all sites with Early and Middle Formative occupations
were collected - again with a few exceptions.
It is worth noting that the systematic surface collection program and analysis of the resulting materials accounted for well over 2/3 of my total time in the field. Though this method produces excellent fine-grained settlement data it must be acknowledged that it requires a quite substantial investment of time and resources.