Some Explorations with Dowsing Techniques

Reprinted from the Journal of the American Society for Psychical Research Vihied LIV Ocha tote Mae Some Explorations with Dowsing Techniques Karvis Osis! Dowsing and related practices of motor automatism have been supposed to facilitate the efficient operation of ESP. It is assumed that many ESP impressions which remain subconscious and un- noticed may be expressed through involuntary and unconscious muscle movements. Tyrrell called them vehicles by means of which the ESP contents ride into conscious experience.? R. Tischner compared them to pipes of exceedingly small diameter in which liquids rise high above the level of a tank. He called them “the * The experiments described here were carried out when the author was a Research Associate at the Parapsychology Laboratory, Duke University. While assuming full responsib for this report, he wishes to express his appreciation for the kind counseling of Dr. J. B. Rhine, Dr. J. G. Pratt, and Dr. R. J. Cadoret. *G. N. M. Tyrrell, The Personality of Man, Pelican Books, London, 1947.

142 Journal of the American Society for Psychical Research capillaries of the unconscious” through which the ESP contents of the deeper levels of the mind filter into consciousness." In the golden age of psychical research automatisms were often employed, In regard to dowsing in particular, for example, there is the capital work of Barrett and Besterman* in which some astonishing cases are reported. McMahan also published a good survey on dowsing.® However, we still lack systematic investigations that Sere demonstrate the superiority of such automatisms over verbal ESP responses. It is very difficult to make such comparisons on the basis of experiments. To begin with, most of the dowsers whom I have met felt offended if I suggested ESP as an explanation of their work. Even when I could secure their cooperation for ESP tests, a subjective bias persisted. On the other hand, when working, say, with college students, the interest of the subjects may easily swing toward, rather than away from, automatisms because of the more dramatic procedures involved, The dowsing problem has been but a peripheral interest of mine, that is to say, only a “‘side-romance.”’ In the explorations reported here no systematic comparisons were made. The main purpose of the tests was to see if an unusually high scoring rate, which then might warrant an extensive research program, could be obtained. However, I decided to write up this brief exploratory work hoping that it might provide some useful material for systematic studies to be made by interested researchers. DOwSING FOR METAL PIPES® Through a professor of the Engincering College at Duke Univer- sity who worked part-time for a local branch of a gas company we were informed about an astonishing case of dowsing for small- size metal pipes. A superintendent of the company, manager of a crew of sixty workers, made use of the method of dowsing for pipes if the pipe-line was not well mapped out and new con- nections were needed. According to our informant, Mr. Gwaltney, the superintendent, was nearly 100 per cent successful. We there- fore contacted Mr. Gwaltney, and he was willing to take part in an experiment. He is a man of approximately fifty years of age, apparently task-oriented, well-measured in his actions, and work- manlike in his approach. *R. H er 4 i "EA, McMahan, “A Review of the Evidence for Dowsing," Journal of Para- , Vol. 19, 1955, pp, 203-227. experiment was financed by a grant from the Duke University Research Fund. Procedure of } inches in diameter and 1} feet in length were used . These target pipes were placed in a vacant lot in two ents of ten small, shallow trenches each. The trenches in a straight line spaced 3 yards apart and were wide, by 5 inches deep, by 2 feet long. They were d with boards 32 inches long, 9 inches wide, and | inch For recording purposes, each arrangement of trenches was red from | to 10. Targets selected by means of random tables were placed in two of the trenches while the re- g trenches in that particular arrangement were left empty. | the first eight trials the targets were placed in the trenches experimenter. In all subsequent trials an assistant arranged vets, and the experimenter was ignorant of their location. “assistant endeavored not to leave any clues which might ibly be detected by sensory means. He visited all of the $ in the arrangement cach time he set the targets and lifted boards in the same way. The experimenter was never ously, at least) able to detect any visible traces that might s where the targets had been placed. Gwaltney used two brass rods, cach about | yard long, the which were bent at a 90-degree angle to form handles. handles were placed within pieces of } inch copper pipe ling the rods to swing freely when the mounted handles eld in the outstretched hands of the dowser. In some of the ¢ also used rods mounted on handles having ball-bearings, decisions which determined the results were based on ons from the brass rods only. . Gwaltney approached the arrangement with slow, measured » His arms were stretched in front of him nearly parallel to ground, He stood from three to ten seconds on or near the d covering each trench. His eyes, so far as they could be ved, were almost constantly fixed on the rods. He usually went the nt from two to four times before he indicated choices. The calls were immediately recorded by the experi- ler. Then we both uncovered the trenches until we found the session varied from working on one to four a seg The experiment was terminated when the time-limit for the The subject was paid at approximately the same rate he was accustomed to receiving for overtime work. However, he seemed to be little Ege bi the money, om cng ot . mo in uae ae ment » He told us, “T want to he u folks to find out t it is that turns the rods.” =

144 ournal of the American Society for Psychical Research Results At the first session we obtained 5 hits out of a possible 8. In the next three sessions, when the experimenter did not know the the high scores continued: | out of 2, 3 out of 4, and 3 out of 8. Then the achievement level drastically declined. The first half of the experiment yielded 14 hits; the second half, only 7. The results of the 50 trials where the experimenter did not know the targets are marginally significant (P = .03), The results are given in Table I. Taste I Diaecr Hrrs Onrarsep ts Dowstna Pon Prves Conditions | Trials | Hite | P* to i ee ee OPE 5 Targets Unknown to Eepésknucner 1 ans | SO | 16 | 2S Peed beet, Ai ie ae | 58 | 21 003 *P obtained by use of Tables of Incomplete Beta-Function. The impression gained from the small amount of data obtained is that Mr. Gwaltney was able to score highly in the beginning, but subsequently his results declined rapidly. The decline effect is rather marked (from 14 to 7) but in so small a sample does not achieve significance. In addition to the above dowsing tests, Mr. Gwaltney did twenty runs of ESP card tests in two sessions, These yielded a deviation of — 2, very close to mean chance expectation. DOWSING FOR A RING To complete this account, I should like to report one other small exploration of dowsing. In the experimental room at the Duke University Parapsychology Laboratory, ten small pill boxes, placed a yard a were arranged in a row. An assistant placed a ring in one of the boxes which was chosen according to random number tables. Two subjects (who were mainly used for pendulum experi- ments) made seventeen trials with the divining rods used in the same manner as was described in the Gwaltney experiments. They scored 4 hits, When the probability associated with this result is ones by use of the Tables of Incomplete Beta-Function, Some Explorations with Dowsing Techniques 145 PenpULUM EXPERIMENTS pendulum is also customarily used as a motor automatism 1 dowsing situations. Its use is especially popular in France and Sermany where its operation is called ‘“‘radiesthesia."" Some very ting experimental work with this form of dowsing has been by R. J. Cadoret.’ e used an arrangement of 25 opaque squares in a five-by-five be known as “A"). The subject stood in front of the eT The ing a pendulum in his outstretched arm. The movement ulum was supposed to indicate the row where the es penny) was hidden under one of the squares. After ing the row, which was recorded, the subject stood in front columns on the adjacent side of the matrix and the procedure epeated. Both column and row having been selected by the um, the square in the matrix on which the indicated row d the indicated column was taken as the subject’s guess for target. The probability of hitting the square beneath which penny was to be found was 1/25. the same session Cadoret’s subjects also worked on another (to be known as “B”) of 25 large squares into which his yard had been divided. In one of the squares a penny was ed as a target. The indoor matrix (A) served as a map and ex for the outdoor matrix (B). With no penny under the squares matrix (A), the subject was given the task of indicating, by ans of the pendulum, that square in (A) which represented the are in (B) under which a penny was actually buried. ‘rom the geometrical relationship between the squares indicated the subjects’ calls and the target squares of indoor matrix (A), pret predicted the rey. between the calls and targets the outdoor matrix (B) but with only moderate success. On the other hand, Cadoret found a clustering of responses iro the actual target instead of a chance distribution spread over the matrix (B). He evaluated the distances between the lis and the targets and found some evidence for the cluster ‘effect? which would be the counterpart of displacement in card tests. _ Both of Cadoret’s problems, the clustering of calls around the target and the prediction from system (A) to system (B), might Very significantly relate to the way ESP operates. Cadoret rightly Points out that both of these findings might help in the effort to gain more control over ESP functioning by increasing its pre- dictability and to harness it for practical applications, as well. ~ FR. J. Cadoret, “The Reliable Application of ESP,” Journal of Parapsychology, Vol. 19, 1955, pp. 203-227. °F ‘Cadoret, op. eit.

146 Journal of the American Society for Psychical Research While Cadoret used two matrices (indoor A and outdoor B) to stand for systems (A) and (B), it seemed unnecessary for me to have a separate “outdoor” matrix. ‘The same 25 squares could be considered as system (A) with one kind of target, and at the same time as system (B) for another kind of target. Cadoret's experiment provided a near-distance (ND) test on his matrix (A) and a long-distance (LD) test on his matrix (B). The present experiment, however, was done in two parts, the first series in- cluding only ND trials on both systems, the second only LD trials on both systems. This provision eliminated confounding with the distance variable. Near-DIsTANCE PENDULUM EXPERIMENTS Procedure This experiment was done in the spring of 1956. We used t five-by-five arrangements of opaque plastic tiles. These squ had recessed bases which concealed the tilted. Under two of the tiles targets were hidden. of targets: money (at first mostly coins, later bills), of persons the subjects liked. The targets an assistant by means of random number t The subject stood at a distance of about columns, holding the pendulum in one parth pendulum (a small button on a silk d) in the direction of one of the columns. The subject column so indicated and the experimenter 1 the same procedure was a basis for her judgement for determining the call. This was done, however, in only a very few cases. Two selected subjects were used. Mrs. K. was chosen because of her striking spontaneous experiences. Mrs. D., the other subject, who was briefly visiting the Laboratory at that time, was in the habit of using the pendulum for spelling out messages on a circular letter-arrangement and was familiar with this technique. The experi- ment was terminated when the subjects left Durham in July of 1956. Results The data were checked for direct hits on both targets and the results are given in Table II. There were 9 hits obtained, 2.6 more times than mean chance expectation. This is a high rate of success when compared with card-calling experiments. The sample size was also allowed to use a “feeling” in her outstretched les e Explorations with Dowsing Techniques 147 too small for evaluating this result by the usual CR method. ‘The distribution might not have approximated the normal curve. “Therefore the arc-sine transformation was applied.* The CR thus obtained is marginally significant (P = .02). Taste I Dimecr Hrrs is Nearn-Distaxce Penoutum Exreniment Subjects No. of sa | Detox CR* | P Money Photo | 70 ae | +42 | j 16 Zin (| id + 1.36 ae ie fs 4 | +556 237 | 02 -* Data transformed by are-sine transformation method. The data were also scored for displacement by Cadoret’s method.’ The distances between call-square and target-square were “measured in units the length of the side of one square. The mean chance expectation and the variance were assessed by use of Greenwood’s formulae, which take the position of the call in the matrix into account. (For details, see p. 214 of Cadoret’s paper.) “The sums of the distances between target and call were very close to mean chance expectation for both subjects as seen in ‘Table III. No indication of clustering near the target was shown. Taste U1 DispLacement ts Teams or Distance nerwenn TAKGET AND Resronsn in Near-Distance Pexoutum Experiment Subjects Sum of Distances | Deviation cR I vical 160.0979 | —6.$487 i BEE D, .--<c00-00- 42.8366 + 5.2652 4 1:24 | 202.9345 | —1.2815 —=a8 Cadoret proposed the hypothesis that the way the subjects miss the target in system A will be related to the way they also miss it in m B. He assumed that the missing might average in the Bie Gicection and approximately the same distancein bothsystems. He used the correlation method for gaining information about the _ ® George W. Snedecor, Statistical Methods, (sth ed.), Iowa State College Press, _ Ames, Iowa, 1956. - * Cadoret, op. cit.

148 Journal of the American Society for Psychical Research distance and direction of missing in system A. This method still left him with a somewhat subjective choice and has been criticized on statistical grounds. I tried to escape the subjectivity of Cadoret’s method in the following way: I assessed the missing and hitting of the monetary a i in terms of distances in columns (right or left 0-4 columns) and in rows (right or left 0-4 rows). These, taken together of course, indicate both distance and direction. Then I assessed in the same way the missing and hitting on the photographs (or system B in my design). Where the same measures for each target were found, I scored a hit (c.g., row 2 left, column 3 right on money; and row 2 left, column 3 right on photograph). If the relation was not the same (e.g., row 3 right, column 0 on money; and row 2 left, column 0 on photograph), I scored a miss. There is the difficulty that the predicted direction could not always be applied to the targets near the edges of matrix (B) because they extended over the edges of the matrix. To cope with this I eliminated all trials where the edge of the matrix would interfere with the predicted call-target geometrical relationship. This reduced the number of predictable trials from 43 to 21. No hits were found. Apparently the results of this series do not support Cadoret’s hypothesis. LonG-Distance PenpuLUM EXPERIMENTS The long-distance (LD) test was begun early in December, 1954, with a series which utilized the services of a nonprofessional dowser (E.0.S.) from Missouri. He had some kind of “home-made” theory about dowsing and (as often happens), had built up a strong predilection in favor of his procedure. Also, in the spring of 1956, I wrote to all the dowsers who had had some correspondence with the Parapsychology Laboratory since 1955. Only two of the correspondents were ready for the LD test, a professional dowser in Germany (A.E.,) who was very self-confident and eager to participate; and an amateur dowser who provided two enthusiastic subjects in Colorado, his father (H.P.) and younger brother (M.P.). In the analyses which follow, the work of all four of these subjects will be treated as a unit, in spite of the lapse of time between these series, because the differences in procedure were very slight. Procedure A drawing representing the 25-square matrix (the five-by-five matrix referred to earlier) in the experimental room was sent to Explorations with Dowsing Techniques 149 5 subject. They were informed that this would be a kind of i o-dovning, and that the cee a map of the actual tile arrangement in the Laboratory. ey were to determine the “call” the same way as in the near-distance (ND) test, except that the drawing was now to become the focus for the pendulum. Each dowser also received a photograph of the West Duke Building in which the Laboratory is located. ‘ The experimenter randomized the target order according to random number tables. Targets were put under the squares and removed at appointed times, twice a day for the nonprofessional dowser in Missouri (E.O.S.), and once a day for the other dowsers in a subsequent experiment. The subjects could dowse at their convenience any time during the period the targets were exposed. The targets for E.O.S. consisted of a dime and a ring; those for the others, a coin (system A) and a photograph of a famous para- psychologist (system B). The subjects in Colorado and Germany were supposed to dowse on the same set of targets. However, the Colorado series was started later, and therefore one-fourth of the total trials in these two experiments have individual targets. The series for the Missouri subject consisted of 30 sessions; for the others, 25 sessions each. However, A.E., (in Germany), missed half a session; and the Colorado subjects, H.P. and M.P., worked ‘on only one set of targets each (photograph or coin) which aney frequently switched between them, H.P. taking M.P.’s target an vice versa. Results same checking and evaluation were applied to the LD data oer to those of ‘the ND series. All subjects but one made a negative score concerning hits on the direct target, i.¢., half of what would be expected by chance. The total deviation for the 155 trials is —3.2 and is not significant. These results are shown in Table IV. Tans IV Dieeecr Hits tx Lono-Distance Penoutum Experiment ——_— * One trial on coin missing.

150 Journal of the American Society for Psychical Research The displacement evaluations by Cadoret's method al meager results. Only E.0.S. had a deviation of some ccmhe “ten it was in the opposite direction from that of the ope: subjects (See Table V.) One can conclude that no definite relationship between the call and the target was indicated. . Taste V Displacement tx Trams or Distance Brrweex Tarorr AND se IN Lono-Distance Penputum Expeaient | Subjects and No. of Sum of inti tion Trials Distencka oes 3 E.OS. (Missouri) .| 60 | 178.2472.| + 19.3977 2.05 A E. (Germany) ... 49" 117.93 — M.P. (Colorado) | |. 28 35-4912 = et H.P. (Colorado) ... 18 37.5314 — 5.5091 Total cnecechs seus 155 369.2046 + 2.3884 * One trial on coin missing. The prediction from the response-target geometrical relationshi on system A (coin) to system B (homoaah was made in te same way as described in the ND series, except for the Colorado subjects, who made trials only in one system each. E.O.S. had 10 scorable trials with 3 hits which is six times mean expectation. This is too small a le for drawing conclusions, but by means of ‘Tables of foctuarlaes Beta-Function 2 P = .02. This gives us an indication that this subject might have had a consistent missing pattern of the type found by Cadoret If this assumption is correct, one could guess that Cadore found not a universal lawfulness, but a partial one, wi only in the data of a portion of subjects, as is the case other lawfulnesses of ESP such as displacem double task effects, etc. Discussion The explorations here described were made by means of a varicty of methods, but, in spite of this, one can easily group th tests into ND and LD categories. The long-distance or map- dowsing parts of the experiment gave no ite indications of extrachance effect. The subjects, however, seemed to be strongly motivated and self-confident about their chances of success. Perhaps they were too eager to permit the successful functioning of ESP; or perhaps map-dowsing is a poor test of psi capacity, at least for these subjects. ‘Some Explorations with Dowsing Techniques 151 The near-distance work, on the other hand, warrants more discussion, because all three subjects scored positively and at a rate. In the experiment with dowsing for pipes the score was nearly twice that expected by chance. In dowsing for the ring and in the pendulum tests the score was more than twice mean chance expectation. The dowsing for pipes and the pendulum tests approached marginal statistical significance. The chance hypothesis on all three experiments considered together may be tested by pooling the probabilities of each series (.031; .083; .018) by means of Fisher’s formula. When this is done the resulting x* is 20.0 and P = .003 (6.d.f.). (The first session with Mr. Gwaltney, where the experimenter knew the targets, was omitted.) Therefore the chance hypotheses in these tests may safely be rejected. Although the laboratory tests seemed to be properly safeguarded, one can seldom be absolutely sure in field tests. The experimenter, however, could not discover any flaws in the precautions taken in the latter, in spite of consistently watching for them. Did the dowsing give results encouraging enough for practical application of this method? I think we should be optimistic, because the dowsers hit twice as often as one would expect by mere chance guessing. It is difficult to predict just how the method would work in a natural setting. The chronological decline in rformance seemed to be very rapid. However, this might not Id true for the natural situation where fresh excitement might be provided for cach new try. The beginnings of each of the three series were quite impressive. The first three sessions in each series gave 2.5 (pipes); 3.3 (ring); and 4.8 (pendulum) times more hits than expected by chance alone, This exploration, in my judgment, provides good encouragement for more systematic research on the problem of dowsing. As far as the replication of Cadoret’s experiment is concerned, my efforts were almost a failure. I found no definite indications for displacement measured by his method. The prediction from one system to the other seemed to work with only one subject out of the four tried, but even this is only supported by statistically weak evidence. It is true that as far as the prediction from system A to system B is concerned, my method was dissimilar to Cadoret’s. Discrepancies in procedure and evaluation might also be a factor. My guess is that Cadoret may have hit upon a lawfulness characteristic of ESP for a certain type of subject, but not as a generally applicable rule. However, his hypothesis and its implications for practical ores tion are too important to be left with uncertainties. We need further experimentation.

152 Journal of the American Society for Psychical Research OBSERVATIONS In addition to the experience with my seven subjects, I have met and talked with several other professional and amateur dowsers. Most of them have constructed their own personal theories on dowsing. The theories are of a bewildering variety, and are naive for the most part, but they stick to them with all the passion of a Fundamentalist for his Bible. I have wondered if there is any association between having a “theory” about dowsing and being an effective dowser? There might be one or more. The dowsers use these theories of theirs for housing a multitude of clues which they have observed in the movements of the rods and in their own bodily feelings. Minute differences in movements are associated in their minds with the depth of the water, the amount of water, the width of the “‘stream,” soil layers, etc. My very vague guess is that a good dowser is a person who knows how to make sharp self-observations and to remember and systematize them for their use as the indicators of certain ESP contents. There seems to be a patiently and meticulously worked out language of muscular signs which acts as a channel of communication between the unconscious of the dowser and his conscious mind. It resembles symbolic dream interpretations, but the associations are more stable and_per- manently fixed. This sign language constitutes a private world which, in the main, has significance for the individual only. It is also worth-while to remember Sir Charles Sherrington’s question concerning the deeper significance of the movement of the divining rod.‘ It would be interesting for Freudians and Jungians to apply their interpretations to the ‘‘rod’’ and to the symbols “water’’ and “mother earth.” There are indications reached by depth psychological studies that the touching of genuine symbolism might release tremendous mental energies in the psyche. The dowsing situation might actualize some deep-seated symbols in the unconscious of the dowser which could provide very strong motivation for ESP to act. On the other hand, the fact that one has a theory for which he stands or falls in itself may give a tremendous motivational charge to the dowsing procedure. If these observations and rather vague inferences should prove to be true, they might open further possibilities for the control of ESP. Experimentation in this direction is much to be desired. Parapsychology Foundation, Inc. 29 W. 57th Street New York 19, N.Y. * Barrett and Besterman, op. cit., p. 275. Printed in Belgium by the St-Catherine Press, Ltd.