The values in Table 1 and Figure 1 reveal some clear patterns in the data. Both the map and the tables need to be used together to properly assess the data: the map has summary data averaged over entire study areas, whereas the tables include values for various strata or environmental units. The smaller strata tend to show closer values between projects than do the project averages. In the discussion below, references are generally omitted, but can be derived from the table and references cited. Figure 2 provides an additional view of recorded site densities; each recorded site in the B.C. Archaeological Inventory (as current March 1992) is plotted as an open circle.
To begin, many of the projects surveyed in open grasslands of the B.C. Interior, classed generally in the ponderosa pine-bunchgrass or the Cariboo aspen biogeoclimatic zones (Krajina 1973), have remarkably similar values of sites per square kilometre: 14.84 (Hat Creek 1), 14.07 (Hat Creek 2) [separate survey areas], 12.86 (Clinton Ashcroft), 18.75 (Mouth of the Chilcotin), and 10.51 (Eagle Lake 1984). Values from intensive surveys are more variable, and are not directly comparable to probabilistic surveys (or each other) because a narrow strip of high potential land is often surveyed, and the width of the strip is seldom rigorously reported or evenly surveyed. This usually results in an inflated value for sites per square kilometre. Sites per linear kilometre is generally a more accurate reflection of density for the systematic surveys. Nevertheless, the square kilometre indices tend to the same central values as those in the grassland: 16.82 (Okanagan Lake), 7.13 (South Thompson, north bank), 10.00 (Pritchard - South Thompson, north bank), 3.74 (South Thompson, south bank), 2.14 (Princeton), 17.14 (Ashnola River), 3.12 (Seton/Shalath). The South Thompson and Seton values are based on wide surveys and are most comparable with the probabilistic survey data.
Subalpine areas have even higher average densities: 17.00 (Lytton/Botanie), 24.31 (Potato Mountain) and 11.25 (Cornwall Hills).
Canyon values along salmon rivers tend to be have the highest values of all: 32.7 (Ashcroft section of Valemount-Vancouver CN line, which is mainly in the Fraser and Thompson canyons), 55.33 (Kitselas Canyon), 30.00 (Chileotin Canyon). The Mouth of the Chilcotin, with 18.75, is also largely within the Fraser Canyon.
In the Kootenay region, two projects have values of 0.8 sites per linear kilometre, the same as Okanagan Lake. These densities are similar to coastal values. Values for the Slocan Valley are low, with 0.6 sites per square kilometre, and 0.33 per linear kilometre on Slocan Lake (the latter is not included on the tables or map).
Within the most northern parts of B.C., values tend to be very low, at least along areas probabilistically surveyed. The massive Liard River survey area has an average of only 0.05 sites per square kilometre, while the Stikine-Iskut has 0.02. However, downstream on the Stikine a high site density has been found near Telegraph Creek but the values cannot be calculated with the available data. The high density of this area is evident from the coalescing circles on Figure 2. A survey along a northwest trail resulted in a value of 39.36 per square kilometre (Nakina River), although this works out to "only" 1.18 sites per linear kilometre, still a high value by comparison to any coastal shoreline. The 210 km long Highland Post Trail in the upper Stikine/Spatsizi area found a density of 0.63 sites per square kilometre, but only 0.02 sites per linear kilometre. Some 16 prehistoric sites were found in the remote 15 x 7 km Metsantan-Caribou Hide study area in the Spatzizi area. Although only a density of 0. 15 sites per square kilometer, these sites create a small dark cluster on the map of all recorded sites (Figure 2).
There are often systematic differences in values between east vs. west and north vs. south shores of valleys. For example, within the interesting Okanagan probabilistic survey (which is the only survey of minor tributary streams of a major southern valley), west drainage strata have an average of 0.45 sites per kilometre square, while the east drainage is 0.20. On the Okanagan Lake systematic survey, the west shore is 0.91 sites per linear kilometre, while the east is 0.76. A very great difference between east and west shores was noted on the Slocan Lake intensive survey (Eldridge 1981), and Atlin Lake had a difference of 0.37 on the east and 0.18 on the west. In this case, the west shores are close to glaciers and may not have been preferred for camping. The north side of the South Thompson River has almost double the site density of the south, at 2.14 vs. 1.12 sites per linear kilometre. Some of this difference may be due to heavier modern development on the south side, which has destroyed many sites, but much of the difference may be the preferred siting of winter house pit villages to catch the sun. Similarly, on the Peace, the north intermediate terraces have a value of 8.38 vs 1.50 for the southern equivalent, while the safeline to valley rim is 11.67 on the north vs. 0.67 on the south.
Many interior lakes have high site densities, and islands in lakes always have higher densities, although sometimes the differences may not be significant. For example, Babine Lake has 0.39 sites per linear kilometre of regular lakeshore, vs. 0.99 for island shores, Tchesinkut Lake has respective values of 0.18 vs. 1.00, and Nazko 2 has 1.24 and 2.29.
The difference in values between different areas of the provincial Interior would be even more pronounced if isolated finds had been treated equally in each project. In areas of relatively sparse finds, such as the Dean and Liard Rivers, isolated finds of lithics were recorded as archaeological sites. On the Clinton Ashcroft project, isolated lithic finds (often projectile points on steep slopes [Pokotylo 1977:21]) were noted but not recorded as sites. If they had been, the sites per square kilometre would rise to 220.71, an astonishing figure based on a very reliable sample. It would extrapolate to some 17,657 sites in this valley, compared to, for instance, the 87 predicted in the entire Liard River pondage area. This undoubtedly reflects intensive aboriginal use of this grassland valley, but it also reflects the relatively high surface exposure and good visibility in the dry interior. It may also be related to its proximity to a major source of basalt which might make stone artifacts locally more disposable. It may merely be a good reflection of the cumulative effect of aboriginal use which is often overlooked in more challenging (from an archaeological survey point of view) environments. Coastal B. C.
This brings us to the B.C. outer coast, where intensive shoreline surveys in a challenging environment range from about 3 to 1S sites per square kilometre, a value generally similar to densities in the southern Interior. These sites exclude CMT's but there is a strong similarity between the Interior isolated find sites and CMTs. CMTs, like lithics on exposed soil, are highly visible, since most are standing trees. Values for CMT densities in areas of old growth forest near coastlines or rivers often translate to hundreds, and even thousands, per square kilometre (e.g., Arcas Associates 1986; Eldridge and Eldridge 1988; Eldridge, Kennedy and Bouchard 1989), even higher than the isolated lithics and perhaps more impressive, since all the cultural scars and stumps date to the last few hundred years, whereas the lithics could span thousands of years.
Along the coast, non-CMT site densities are best described as sites per kilometre, since virtually all the surveys have been linear shoreline traverses of narrow width. Almost all vary between 0.4 and 1.4 sites per kilometre, with higher values generally in the south. The lowest value, 0.2, occurs on the central coast near Bella Bella, but this survey concentrated on intertidal beaches and may not be comparable. Inter-observer variability may also be a problem: on the 1992 Gwaii Haanas project, the two most experienced crew leaders had 0.92 and 0.78 sites per kilometre respectively, but the least experienced crew leader had an average of only 0.22, bringing the project average down to 0.26.