Comments on significance of Undercurrents Survey of OSLE

© Bob Forrest, 1979. Published here by permission.

Figs. 1 to 4 should first be studied carefully.

Figs. 1 to 3, at least to some extent, mimic the selection of the 53 sites specially numbered in OSLE, whose alignments were subsequently analysed by Undercurrents. Some sites (see below) were not so numbered by John Michell, and so were not included in the Undercurrents survey either. These correspond to the unnumbered dots in my fig. 1. Some of the sites numbered by JM were discovered in the field on known alignments – like the lettered circles in my fig. 2.

Most important is fig. 4, where, after random nudging within the 1 km square of the ‘selected’ sites of fig. 3, the number of alignments drops – not because a deliberate nature of the original alignments, but because of the process of selectivity in building up fig. 3 (the ‘database’ for the nudging experiment).

The Undercurrents Computer Analysis of actual lines at Land’s End corresponds to my fig. 3. The computer found 51 triads, 5 tetrads and 1 pentad.

Upon computer nudging (corresponding to my fig. 4) the score fell to 34 triads and 1 tetrad.

The question arises: how far is the drop in the Undercurrents score related to the drop in going from my fig. 3 to my fig. 4?

This question hinges on how JM came to select his numbered sites, and though the situation is not quite as simple as that used by me in deriving my fig. 3, there are nevertheless disturbing elements of it there.

Chris Hutton Squire in his “OSLE – A memorandum of the sources of JM’s list” lists 16 menhirs given in West Penwith survey, but not numbered by JM. Of these 16, CHS found no prima facie reason to exclude for 14. Do these menhirs correspond to the unnumbered dots in my fig. 1?

Ditto for stone crosses – JM used and numbered 7, yet CHS lists 40 such. As a matter of fact the use of stone crosses in OSLE corresponds more to by use of lettered circles to extend known alignments, than to the omission of dots in my fig. 1. In a letter to me, dated 28 Apl. 76, JM wrote:

“Stone crosses. Here there is selectivity. It is very hard to cover these. So very many have been moved. The ones I took were those on good alignments which most probably have not been moved. You should omit these from an analysis. Phenomenal evidence, not statistical.”

– cf. My use of C in my fig. 2, to extend line 2,14,16.

In OSLE the 1 pentad uses a cross, as does 1 of the 5 tetrads, just as, in my ‘scaled-down’ model, my only tetrad uses a site C selected because it is on an existing line.

In another letter (undated, but written in mid-Feb. 1977) JM wrote:

“One or two (stones) got left out as inconvenient to visit (mostly on foot); others turned up in the field – and of course since I was starting with a theory, I was obviously more likely to find ones on already suspected alignments (although I included several not aligned).”

Of course, in my model I have included/lettered circles ONLY where they fall on existing alignments, or form alignments with already numbered dots, whereas JM says he did include discovered but not aligned stones. But the fact that some stones were included because they fell on suspected/existing lines is bound to distort significance to some extent.

According to CHS’s “OSLE - Memorandum etc.”, JM found 7 menhirs, numbered 4, 5, 7, 8, 11, 25 and 26 in OSLE. In his letter to me of 28th Apl. 1976, JM wrote: “Others were found by me (e.g. 7, 8, 11, 25, 26). I put them in if I had their positions accurately surveyed (7) or even approximately (25) but in the latter case would obviously not go into alignment question. Several others are known to me, but unsurveyed; a few others are recorded in archaeologists’ lists, but I did not include them because I do not trust their map refs. for this purpose. Looking at it all again, I do not think I was selective. True, I surveyed 8 because it was on a well marked line and there are others around I did not survey, but on the same line as 8 there are two other good stones, mentioned in the text but not given numbers.”

It it true then that JM’s selectivity is not as straightforward – blatant might be a better word – as that used in the derivation of my figures 1 to 4, but nevertheless, the presence of elements of such selectivity in the derivation of the set of numbered sites in OSLE is what makes me mistrust the ‘significance’ lent to the alignments by the Undercurrents Survey.

The point about the drop in alignment scores from my fig. 3 to my fig. 4 is that my 4 is analysing the alignments to be expected amongst 19 points, and comparing this score with that of fig. 3, which, because of its derivation, carries with it the alignments to be expected amongst 30* points but subject to the constraint that any triad/tetrad must contain at least 2 numbered dots.

*20 dots, 10 circles.

We have seen that in OSLE there are elements of subjectivity suggesting that what are quoted as alignments amongst 53 sites (corresponding to 19 points in fig. 4) are perhaps those inherited from a larger set subject to certain constraints (i.e. the Undercurrents simulation should have been done on the equivalent of my fig. 2, and not, as it was, on the equivalent of my fig. 3).

Since JM’s working methods are more complex than my selection procedures in compiling my figs. 1 to 4, all that can be [said] is that though the Undercurrents Survey may suggest deliberate alignment, it may also suggest something quite different.

Bob Forrest, Feb. 79