Many of the problems identified while assessing early survey reports have been rectified by Archaeology Branch policy and the formal Guidelines for Archaeological Impact Assessment. Permit applications are now rigorously examined for methodological precision, and flawed survey methods will normally be queried and rectified prior to fieldwork. Systematic bias towards finding certain site types in certain locations will often be addressed during this consultation. However, there are still some areas where improvements can be made, and other initiatives can be taken.
Much of the field work now conducted in British Columbia is done as a result of impact assessments. Whether a development request is ever subject to the
archaeological referral process is often decided by local planners or other officials, who may have little idea of archaeological potential. For instance, Mackie and Eldridge (1992b:86-87) have recently identified a breakdown of the referral process at various levels of government within an area of northwestern B.C. In this case, non-professional, non-technical support staff from other ministries were making archaeological resource management decisions. Of development requests that do get referred to the Archaeology Branch, impact assessments are only required for those that the research officer feels have archaeological potential. This is, in the absence of recorded sites in the development area, an example of ad hoc predictive modelling. However, these models are generated without hard data except for information available on NTS 1:50,000 topographic maps (often seriously out of date), combined with personal experience of the area, if any.
Probabilistic methods for surveying coastlines should be assessed through field trials. CMT research has shown it is possible to archaeologically survey large inland areas and it should only require the addition of subsurface testing to combine these surveys with traditional site surveys to produce some data on inland and intertidal site densities along the B.C. coast. However, the survey intensity would have to be increased compared to CMT surveys because small buried sites are not visible from a distance the way standing CMTs are. Transects may be better suited to this survey than quadrats, given the logistics of locating survey areas in rainforests. Surveys need to be undertaken in different environmental zones and areas with different Quaternary and Recent history, as sea level curves are hugely varied.
Probabilistic surveys of CMTs, common during the mid to late 1980s, have largely been replaced by judgemental traverse surveys. While the latter may be sufficient to characterize CMT distributions and quickly collect suitably large samples of particular data, they are not as reliable as probability samples for assessing density and distribution. Interval transect surveys produce good results for estimating population parameters; they are not very good for defining the boundaries of CMT "sites". These have not been well defined in the past. If an interval transect method is used, it should be regular and be spaced close enough to interpolate site boundaries.
Surveys of coastal areas should include survey of mid and lower intertidal areas, with subsurface testing, in order to find inundated supertidal archaeological sites. These may often contain highly significant "wet" sites with plant preservation. Apland recommended nearly 20 years ago ',the intertidal zones of beaches should be examined for evidence of cultural activity even in the absence of such diagnostic features as middens, fish traps, house remains, etc." (Apland 1974:7). Although it is often not clear from coastal survey reports, it seems that this was not a standard coastal surveying practice until quite recently. It was only very recently that surveys and site assessments included subsurface testing of beaches, with the subsequent discovery of subsided and/or waterlogged cultural deposits (e.g., Easton 1991; Eldridge 1989, 1991; Eldridge, Bouchard, and Kennedy 1992; Mackie and Eldridge 1992a).
In the interior, there is a great need for additional probabilistic surveys in previously unsurveyed areas. The surveys need to employ subsurface testing. Supplementary survey with subsurface testing in forested zones near known high potential grasslands is also required, as this has never occurred. Even if prehistoric behaviour was strongly correlated with grasslands, there is a strong likelihood of the distribution of grassland and forest being drastically different in the past, not only due to climactic changes, but due to the abandonment of traditional Native burning techniques.
A point of interest concerning the prehistoric vegetation zones was the annual burning of the upper level growth by the Indians to encourage more range for the propagation [sic] of deer (V. Adrian, pers. com.). Chief Adrian states that until recent time the meadow land was 1000' lower than present due to this forest control (Wales 1974:2).
The situation is analogous to the changing sea levels and the distribution of coastal sites through time.
Several improvements to the reporting requirements should be made. A clear presentation of methods is vital. Of foremost importance is the delivery of explicit large-scale maps of areas covered by foot (and subsurface tested if applicable), areas covered by other techniques and areas not covered. A useful example is the 1982 Meares Island survey intensity map (Mackie 1983:15-17). Summary statistics of accurate size of study area and surveyed area should be reported. The length of shoreline (if applicable) broken down by different survey techniques is valuable information, but is seldom presented. Subsets of the survey area used in analysis should also be measured.
The Archaeology Branch needs to input survey coverage on 1:50,000 NTS map base and on GIS when it effectively replaces the hardcopy maps. It is critical to include surveys which found nothing. The survey coverage should be colour coded by survey intensity, use of subsurface testing, etc. The boundaries of areas where detailed potential maps have been produced should also be indicated on the maps. Large sites should be plotted as polygons.
There should be a tighter standard for site definition - that is, for deciding at what point the distance between two cultural deposits becomes great enough to determine that the deposits should be called two sites. There are instances of splitting to absurd levels - for instance, of describing petroglyph panels 7 m apart as two sites - and also of lumping with equal absurdity - for instance, of a railway construction camp, a discrete traditional Native village, and a shipwreck in the nearby river all with a single Borden number. There is often better quality information from split data than from lumped data. For instance, treating a large area with diverse features and components as a single entity may result in important data for one component not being recorded. Splitting is, in general, preferable to lumping, and the development of relational databases allows for detailed splitting without unnecessary data duplication. At present, the different site definitions make estimating site densities very difficult. Entry of field data into the CHIN database can control some of these problems, especially unnecessary splitting. However, published guidelines could address this problem, and a set of standards could be developed and universally applied. The effective use of GIS will impose new standards and will also encourage creative solutions. For instance, a single site with diverse components - such as a small shell midden with an associated area of CMTs which covers many hectares - can under the present system have only one set of dimensions and a single-point georeference. However, under a GIS running from a relational database, a "many to many" relationship can be defined, and the shell midden can be associated with one set of attributes and dimensions while the CMTs can have their own data set and georeferencing, while the two components share a single Borden number and set of common management fields.