As no aceramic habitation sites were encountered, the sole criterion used to define non-agricultural sites was the density of the prehistoric sherd scatter. I am aware of the artificiality of the site concept, and of attempts that have been made to remedy it. However, as the research progressed several things became clear which compelled me to adopt a site-based approach to settlement research. First, all previous surveys in the region had adopted such an approach, and I wanted to ensure compatibility with existing data sets. Second, it rapidly became apparent that the quantity of prehistoric material in the research area was quite extensive (almost 500 sites were eventually identified) and that without the shorthand recording method of ``the site'' it would be quite impossible to document this material with the limited resources at my disposal. Third and finally, it was clear that certain clusters of ceramics in the study area were much more dense than others. While the entire Taraco Peninsula is one large, low-density ceramic scatter, certain areas within it do stand out as being more dense than others. This observation was made some years earlier in my surface collection of the site of Chiripa ([Bandy 1999d]). I finally decided to use a density of 0.1 sherds per square meter over an area of at least 100 square meters as the minimum site definition, the same site definition criterion I had earlier decided upon for Chiripa. While this figure is indisputably arbitrary, it does nevertheless seem to have some real significance, as the empirical evidence from the systematic surface collections will show. The smallest possible site using this definition would be 10 sherds in a 10x10m area. In actual fact, almost no sites approached the minimum, the vast majority covering at least 1000 square meters with 100 or more sherds.
When a site was encountered - that is, a concentration of ceramic
fragments in which those pertaining to a single time period attained
a density of at least 0.1 sherds per meter squared over an area of
at least 100 m
- then the point of greatest density was located.
This was used as the site center, and was located in space using a
Magellan GPS receiver (set to the South American provisional datum
of 1956). The dimensions of the scatter were then determined by pacing
North-South and East-West from this center. The area of the site is
calculated as the product of the length of these two axes. If ceramics
from multiple prehistoric time periods were present, then multiple
sectors were defined.
``Sector'' is best understood as a ``site within a site,''
and is conceptually similar to the common usage of ``component''
or ``occupation.'' Indeed, I will use all of these terms interchangeably.
A sector is an area containing a continuous distribution (at least
0.1 sherds per m) of ceramics from a single prehistoric time
period. Sometimes multiple sectors overlap, and together form a site.
A site, then, could also be defined as one or more overlapping sectors.
As will become apparent later on, the sector represents the truly
fundamental unit of my analysis, the site being simply a convenient
recording mechanism. It is important to note at this juncture that
agricultural features can be sectors, as well. Thus a scatter of Early
Pacajes sherds adjacent to a small raised field group would constitute
a single site but two sectors. These two sectors would have distinct
sizes and phase attributions, and would therefore be entirely separate
analytically.
Site recording involved location, measurements and sector definition, as described above. Additional information recorded on the site form included community, property owner, time period, a basic site typology (useful for distinguishing habitation sites from agricultural features, primarily), microenvironmental zone (as defined in the Chapter 2), local resources (springs, viewshed properties, access limitations, etc.), modern disturbance, agricultural history, ground cover and so on. A small sample of representative diagnostic sherds was also recovered, along with any exotic lithic materials or prehistoric metal objects (extremely rare) encountered. In general I tried to collect a minimum of 30-40 phase-diagnostic sherds from a given site, though this often proved to be impossible on the smaller sites.