The Munich Barn Experiments on Dowsing, Studies
Standing
Concerns Replication Status
Currently States The successful subset has never been reproduced. Enright reported that it disappeared under conventional analysis; Betz and colleagues replied that the effect persists, and Enright answered again.
J. Wagner, Hans-Dieter Betz and H. L. Koenig ran the Munich barn experiments on dowsing in 1987 and 1988, testing more than five hundred dowsers across more than ten thousand double-blind trials in which a water pipe on the floor below was moved between runs to positions set at random, unknown to the dowser or the attendant. Their 1990 final report to the German federal funding agency identified a small subset of trials with results the investigators read as better than chance. J. T. Enright's Testing Dowsing: The Failure of the Munich Experiments (1999) re-examined the same trial data with different statistical grouping and argued that subset was a product of testing many subgroups after the fact rather than a real localized effect, since no equivalent subset recovers the same pattern in later or independent data. The method on both sides is statistical reanalysis of the same trial archive, and both sides published without either withdrawing, keeping the exchange live. It is not evenly balanced: no later trial has recovered the successful subset, which is why the atlas records the underlying dowsing claim itself as debunked even while this exchange remains debated.
Source Testing Dowsing: The Failure of the Munich Experiments, J. T. Enright