Published online by Cambridge University Press: 02 January 2015
The astronomical significance of Stone- henge has been the subject of intermittent debate and speculation ever since 1740, when Stukeley [I] first observed that the axis of the sarsen structure and of the Avenue pointed at least approximately to the sunrise at the summer solstice. Apart from Sir Norman Lockyer [z], however, few professional astronomers concerned themselves with this question, until the appearance in,Nature of the two articles by Gerald S. Hawkins, Professor of Astronomy at Boston University, in 1963 and 1964 . The first of these claimed the discovery of a number of additional alignments of astronomical significance, marked by pairs of stones and other features of the site; and the second suggested that the Aubrey Holes had been used as a neolithic 'computer' for the prediction of movements of the moon and of eclipses. Subsequently these theories received much wider publicity, in Britain as well as in the United States, through a CBS television programme, The Mystery of Stonehenge, which provided a superb example of partiality and tendentiousness in the presentation of an academic controversy.
Professor Hawkins has now elaborated his ideas in a book whose title, Stonehenge Decoded, leaves no doubt of his confidence in the rightness of his conclusions-a confidence explicity echoed in his text. 'There can be no doubt that Stonehenge was an observatory; the impartial mathematics of probability and the celestial sphere are on my side' (p. vii). 'I think I have demonstrated beyond reasonable doubt that the monument was deliberately, accurately, skilfully oriented to the sun and the moon' (pp. 146-7). 'I think I have put forward the best theory to account for the otherwise unexplained holes' (p. 147). 'I think there is little else in these areas that can be discovered at Stonehenge'(p. 147).
* The producer, Mr Harry Morgan, was in no way to blame. His own approach was meticulously impartial.
* Kindly communicated to me by Mr C. A. Newham. Hawkins explains this on p. 111.
** The correct probability, calculated with pencil and paper instead of a computer, is 0.000006. The arithmetic is not as ‘horrible’ as Hawkins suggests (p. 136). It takes about three minutes.
† What Hawkins has calculated (incorrectly) is the probability of getting exactly 24 significant alignments by chance out of a total of 50, where the probability of a random alignment being significant is 0.2. The appropriate statistical test, however, requires the calculation of the probability of getting by chance at least the observed number—that is, the observed number or more.
‡ Between the 14 points involved there are 182 possible sight-lines; but is it legitimate to include only one of the four post-holes near the Heel Stone (A) and exclude the other three? With all four included, there are 272 possible sight-lines. These totals may be reduced by eliminating all lines shorter than the shortest which Hawkins accepts, and those unlikely to have been used in one direction (e.g. from the Heel Stone and to the (unmarked) centre of the circle). They cannot, however, be reduced below 111 and 159 respectively. Under the conditions stated, the probabilities of getting by chance at least 23 significant alignments from these totals are 0.46 and 0.97. Both are wholly consistent with the hypothesis that the alignments claimed are accidental.