Testing Tradition: A Controlled Study on Dowsing Accuracy

Dowsing is an ancient practice that employs rods or pendulums to locate groundwater and other subsurface features. Although it dates back several millennia and remains widely used today—by private individuals and, in some cases, by municipalities to locate the most easily accessible groundwater sources—dowsing relies largely on intuition and experiential knowledge rather than controlled scientific experimentation. This contrasts with hydrogeology, which emerged as a scientific discipline in the 19th century and is based on empirical observation, modeling, often complemented by other scientific disciplines, which involves significant costs. Field demonstrations of dowsing can appear convincing and are often viewed as cost-effective methods of selecting a location for drilling a new well; however, in regions characterized by sufficient rainfall and favorable geological conditions, groundwater is often ubiquitous. This raises questions about what constitutes a genuine “detection” and complicates meaningful comparisons between traditional and scientific approaches.

In addition to general groundwater, dowsers frequently claim the ability to detect underground drains or pipes. These discrete, well-defined targets provide a more suitable basis for controlled experiments and statistical evaluation. In this context, we present the final results of a controlled dowsing experiment co-designed by a multidisciplinary team of scientists in collaboration with an experienced dowsing practitioner, with the aim of ensuring both methodological rigor and acceptance by the dowsing community.

The experimental setup consisted of a grid divided into 25 cells, within which various objects were buried in selected locations. These objects included iron and plastic pipes, either empty or filled with water. When present, the water varied in composition and conditions, including saline or fresh water, as well as stagnant or flowing water. Participants were asked to scan the grid using either wooden or metal rods and to indicate the cells in which they believed an object was buried.

A total of 54 participants took part in the experiment: 27 with intermediate to extensive prior experience in dowsing, and 27 novices who received basic training before participation. Several participants completed multiple trials to assess reproducibility. The results are analyzed and discussed as a function of object type, dowsing instrument, and prior experience level, providing a quantitative assessment of dowsing performance under controlled conditions.