Archaeological Dowsing IV – Revenge of the Rods


Emmeline thought the “Ideomotor” must be some newfangled form of transport…


I’ve written before about the limited (but still too widespread!) acceptance of dowsing in archaeology on several occasions. Needless to say it hasn’t gone away since then. As dowsers are all too fond of telling us, it’s an “ancient” technique. I recently became aware of an attempt by  a group of amateur archaeologists convinced of its efficacy to win over their professional colleagues and raise public awareness. You can read it here.

The most superficially impressive claim therein is this;

“[Paul Daw’s] discovery, by dowsing, of anomalies between the stones in the Stanton Drew circle, Somerset, prior to their detection by fluxgate gradiometry, merited a brief note in the magazine British Archaeology, No. 111 (March-April 2010)”

This wouldn’t be the first time that non-academic archaeological publications have uncritically reported dowsing, but I wanted to investigate this further. The note in question is entitled “Geophysics finds encourage new look at Stanton Drew” and the relevant paragraph reads;

“[The archaeologists] found anomalies between the stones of the circle, which John Oswin and John Richards (BACAS) and Sermon suggest may be contemporary features. Though not revealed before by geophysics, similar features had been claimed earlier in the year by dowser Paul Daw.”

There is a subtle but important difference here. The claim implies a match between anomalies “discovered” by the dowser and those actually found by geophysical survey. In fact nothing is said about just how “similar” they were, nor how precise a match they were to the dowser’s features. Just that “similar features” had been claimed.

Now, a stone circle, oddly enough, consists of one or more circles of, er, stones. Stones require holes in the ground in order to set them. Several thousand years often leads to stones being removed, relocated, or destroyed. Thus there is a pretty good chance of finding some sort of buried hole in the ground in between those holes in the ground that are still obviously filled with stone. One can also take a pretty good guess at the likely size of any as-yet undiscovered holes in the ground based upon said above-ground stones. Finally, stones in stone circles are usually spaced in an even manner, further increasing the chances of correctly guessing the location new features. The upshot of all this is that unless we can know how closely Daw’s plotted holes matched those found by science, the claim is worthless. This is the closest thing to independent assessment of this society’s efforts that’s offered. Everything else is self-claimed and self-affirmed (as with the “test” linked in the closing paragraph of the newsletter).

The rest of the newsletter consists mainly of claims to have discovered or confirmed the suspected routes of stretches of Roman road (and features connected to them) in the area. Most are unfalsifiable, and to be fair, without professional backing dowsers often lack the means to “verify” their own work. One claims to have been cited by (presumably) an archaeologist;

“Interestingly, just a few months ago, Judy and I attended a lecture on ‘Roman Roads in Cumbria’ by a man we’d never met, and were amused to learn that ‘Andrews and Andrews’ are now officially credited with the discovery of the Kendal to Ambleside Roman route – even if we did use dowsing to find it!”

The big problem here applies to the other supposed success stories in this newsletter. These people may well have discovered a stretch of Roman road, but the extent to which they used dowsing to do so is far from clear. In fact like the other contributors, it is clear that their achievement relates to their fieldwork methodology as a whole including visual survey (simple observation of the lie of the land), field survey (measuring it with trundle wheels, ranging rods, perhaps even theodolites in order to detect changes in topography) placename research (settlement and feature names known or suspected to denote a former Roman road in the landscape), map work (using maps to determine likely routes based on contour lines, watercourses etc) and a bit of local knowledge to speed things up. Together with confirmatory excavation, this is how such features are found. We need to know precisely how the dowsing was done, and what aspects of the find it supposedly contributed to. Naturally I suspect that it played no part.

To be fair, being a local archaeological society, funding and opportunity for excavation is going to be limited, hence these guys aren’t often going to be able to confirm their fieldwork (dowsing or otherwise). Hence on page 7;

“Our results were vindicated when a gas pipeline subsequently cut across the line of this road and revealed a cobbled surface complete with two Roman hobnails on exactly the same alignment as that determined by dowsing (Wilson, 2009: 288)”.

Again, we are expected to take their word for this – the word “exactly” is not quantified. I realise that this is only a newsletter and not a journal article, but if the goal is to lure more “conventional” archaeologists into the fold, surely something other than bald assertion will be needed. A nice diagram of coinciding datapoints, perchance?

I realise you’re trying to attract paying members in order to keep your heads above water guys, but how about at least a sneak preview of any data that might actually lie hidden in a copy of your journal, archived in the basement of the local library and labelled “beware of the leopard”? Of course even some impressive-looking results would still need to be subject to peer review and then reliably replicated before we could all burn our magnetometers. It doesn’t matter how cheap and easy dowsing is if it does no better than chance.

So, though working somewhat in the dark, it seems to me that these chaps are checking existing lines along which roads are suspected to run. Roman roads are (famously) linear, but tend to respect the existing topography (at the time). So it should be no surprise that dowsers are able to plot a fairly straight route vaguely in the same alignment using nothing more than pre-existing knowledge and educated guesswork, mediated by the ideomotor effect. Without knowing how close a match they got, or how many “misses” they made, we can’t begin to assess whether dowsing has played a meaningful part. Quite apart from it having no currently conceivable mechanism behind it.

Now, if excavation confirmed a stretch of road “found” wholesale by dowsers without any prior evidence that there was one there – that would be more impressive. But you would need to undertake some sort of initial survey to find a place in which to begin looking. As a result there would still be a percentage chance of finding something, and so we would still need to know how many attempts were needed before gold was struck, as it were. It also wouldn’t be fair to expect a dowser to just go and find a road, because even a geophysicist would struggle without some kind of lead. The only fair way to test is a with a tailor-made and agreed blinded protocol, where there is definitely something there to be found, and the only means of doing so is by dowsing. Like this, for example.

In summary then, this newsletter presents strictly anecdotal evidence of a technique that is very difficult to separate from others that by necessity must be used when in the field. It’s never going to be able to persuade any critically thinking archaeologist that dowsing is worth looking at.

Guys, if you’re sincere and interested in evidence, even when it doesn’t work out in your favour, why not construct a proper test and publish the results in a future issue? Or at least give us enough information about your fieldwork “successes” to let us think for ourselves about them.



Tags: , , ,

%d bloggers like this: