We have recently had another opportunity to examine how scanning and sampling strategies compare with complete (detailed) survey. First of all though, what do we mean by these terms?
- Unrecorded scanning – a walkover, perhaps at 10m line separation, noting where the measured variable is anomalous but not recording the variable
- Recorded scanning – as before, but logging the measured variable in the correct place for subsequent objective study
- Sampling – use of blocks of detailed survey, positioning these according to some other information and ignoring the rest of the site
- Detailed survey (UK) – complete coverage of a site with lines 1m apart
- Detailed survey (Continental Europe) – complete coverage of a site with lines 0.5m apart
In this recent comparison we found that the contractor (a well known company) had made no allowance for up to four different shallow geological contexts at the site and therefore could not have assessed variability of response across these. Their survey design (unrecorded scanning followed by small areas of detailed survey) failed to recognise that river terrace gravels can typically contribute discrete anomaly strengths of 3 – 5 nT, similar to anomalies from archaeological features. What is worse is that although they must have seen this natural variation during scanning, their surveyors either didn’t care or were too dumb to realise that their chances of discriminating archaeological sources against this natural variation were virtually nil. We have to assume to assume dumb because their interpretation of their small areas of detailed survey went on to label these natural variations as pits of archaeological relevance. As important, however, was their failure to find a number of prehistoric enclosures that were revealed when we undertook detailed survey of the entire site.
Given that the scanning approach failed at this site, how often does it fail elsewhere? In this case the root cause of the problem is the contractor’s failure to take into account natural geological variation. Common sense would suggest that scanning over most gravel-based superficial geologies could not be recommended because of the similarity of natural and artificial anomaly strengths. Indeed, where we have made comparisons, other contractors’ failure to take into account natural variation is the most common cause of unreliable (or at worst, bogus) results.
Can scanning can be justified on the basis of scientific objectivity? In the worst case it produces nothing that can be independently assessed by peers, there being no recorded signal that can be examined. Assessment of variablity is coloured by the individual surveyor’s perception and thus influenced by experience, gait, fatigue, etc. It also presumes that the surveyor is able to perceive small enough variations and differences in those variations. It is fairly obvious that this is not achievable in all cases and although surveyors have claimed sensitivity as good as 2nT/m for a 0.5m vertical gradiometer there are many cases where natural variation exceeds this. There are also archaeological sites where anomaly strength rarely exceeds this. Finally, the scanning surveyor can only visualise what they have physically traversed over and given that a 10m line separation is common this represents a very thin sample compared with the 0.5m separation normally considered necessary for ‘complete’ mapping of archaeological features. Indeed, from these figures a 10m line separation could be considered merely a 5% assessment of the site and therefore perhaps systematic archaeological trenching would be as (un)reliable?
There are many factors that limit the effectiveness (reliability) of scanning based methodologies and few of these seem to be quantified by the contractors undertaking the work. This being the case perhaps scanning based absence-of-archaeology results are really just an absence-of-evidence? The claims made for scanning as a technique to detect archaeology appear at best dubious and in some cases perhaps intentionally misleading? Is there any need for such techniques now that prices are low and multisensor instrumentation is becoming more prevalent?
ArchaeoPhysica does not support the use of scanning based methodologies for the detection of archaeological features. We have, however, very occasionally used fully recorded (signal and position) widely-spaced traverses to delimit areas of unfavourable ground and then always with the support of geological and any other information available.