Having recently read the article

by Jeffrey Keen in the September edition of Dowsing Today, I feel compelled to challenge the conclusion of the article.Measuring the Size of a Dowsable Field

In the article, thirteen members of the Dowsing Research Group dowsed the length of a pencil drawn line, at intervals, throughout an approximately 24 hour period. The article presents the length dowsed by each member, at each measurement time, together with a graph (Figure 2, in the article), which shows the results of the average length recorded by the group and its variation over time.

The trend line drawn through the plotted data shows an apparently interesting variation over the measurement period. The effect indeed looks intriguing until one looks at the standard variation values quoted for each average value. Despite quoting the standard deviation in the data table, no consideration appears to have been given to these values, which is surprising since the standard deviation is a good guide to the uncertainty of the measurement process.

One way to conceptualise the importance of standard deviation, is to draw a line through, and centred upon each data point on the graph. The length of each line extends a length equal to twice the standard deviation for that data point. When this is done, it is apparent that the observed variation is not statistically significant. Practically all the data points lie within the measurement uncertainty of the other data points. One could draw almost any trend line through the plotted data, including a straight line.

This is not to say that there is no cyclical effect on the dowsing reaction, but the data does not support this conclusion. With this set up, a much larger number of dowsers need to be involved to improve the signal to noise.

Regards

D0w5er

Edited by I.P. 28.10.08 - corrected spelling error in subject line and several minor typos