Chapter IV. Earlier and Minor Experiments

Extra-Sensory Perception, by J. B. Rhine, [1934], at sacred-texts.com

This is the only chapter in this book with a Roman numeral. This has been left intact in this transcription.—JBH

CHAPTER IV

Earlier and Minor Experiments

The two-fold objective of the experiments in extra-sensory perception at Duke University is, as stated earlier, first, to answer, if possible, by mathematically indisputable evidence the question of its occurrence and of its range; and, second, to further its understanding by the discovery of its relationships to other mental processes and to the essential physiological and physical conditions. In the detailed accounts of the experiments given in this chapter and those immediately to follow (5 to 8) the first of these purposes, concerned with the proof, will be given first consideration. Then in later chapters will be summarized those special experiments which were designed to help to explain rather than to prove. Those readers who are in danger of being bored by more proof for extra-sensory perception can well afford to go over to Chapter 9 (unless perchance they may be interested in the personal accounts of the principal subjects given in the immediately following chapters).

In these chapters containing the more complete reports all the available information that seems helpful to thorough understanding will be included and will be built up around the personalities of the major individual subjects. The reason for this is that in this work so much irregularity of conditions, procedure, and results is inevitable because of the great factor of human variability that it is hard to generalize over the whole range of subjects in any detailed fashion. However, in Part III which follows this series of individual studies there will be a general discussion of the main hypotheses, in which a summary will follow, and

p. 47

considerable generalization and summarization will be made from the results as a whole.

In the conduct of these experiments there has not been a carefully drawn up plan of procedure right from the start. In work of this kind it is necessary to proceed as explorers, ready to adjust plans at every turn, flexible as to methods and conditions. Only the general objectives need be kept fixed, and the means and criteria of interpretation. Often a block of work is of little value because of poor conditions of security against possible errors or deception, but if thereby there may be a chance to develop a good subject for later improved conditions, we relax the conditions and record them as they actually are. The sensitive but powerful factors of mental attitudes and moods have to be regarded with care. One can seldom proceed directly to the point, as with test-tube work. Much of what may seem to many as the incredible stupidity of the investigator is due to these limitations. On the other hand, I see now several places where I might have done much better, saved time and improved a point, if I had known more. Naturally, the progress of the research as it has advanced has altered and improved the plans for the future. We are more given to following up the best leads, as they come, than to trying to elaborate a perfect experimental plan and following it strictly.

.     .     .     .     .     .     .     .

The first laboratory procedure was that of the trance-telepathy experiments by Dr. Lundholm and myself. Our immediate purpose in this was to discover, if possible, any individual showing in the trance condition the striking telepathic capacity claimed for some of the hypnotized subjects in the 80's and 90's by French and English students in this field. We did not at this stage exclude clairvoyance as a possibility. We were using the hypnotic trance, however, merely as a presumably favorable condition. There was also the secondary purpose of measuring the total scores by the probability devices in order to reveal any minute and dispersed telepathic ability present. To state the outcome in a word, the first purpose was not satisfied by the experiment but the second did, I think, achieve a small degree of success.

The subjects were students who volunteered for the experiment. Dr. Lundholm and I together worked with 12 and, after Dr. Lundholm was unable to continue because of time limitation, I alone went on with 18 more. Six others were eliminated because of failure to attain the desired state of trance. With the 30 subjects in all we carried out a total of 1,115 tests, using three different procedures of testing for telepathic perception but with the same general hypnotic treatment.

p. 48

The hypnotic trance was induced at first by Dr. Lundholm, using mainly suggestions to relax and to become sleepy. After suggestions for muscular plasticity and rigidity had been effective, and suggested amnesia and post-hypnotic suggestion were found to succeed, the instruction was given that the subject would awaken on a given signal, would take a comfortable position in an armchair and would relax again. He would then be inattentive to general stimuli in the room but would be fully receptive to what the designated agent, Dr. Lundholm or myself, would desire him to do, and he would be in close rapport with the agent. It was suggested that he would receive impressions from the agent's mental processes directly. Further instruction was given the subject after he was seated in the arm chair, before a table.

There were a great many variations in the actual words used, the order of various tests of depth of trance and even in the instructions. We were exploring for a process, not quantitatively measuring one already known. My own hypnotic procedure, though I learned it from Dr. Lundholm, soon varied considerably from his. But these are unimportant matters at this stage.

We began by using numerals, 0 to 9, printed on cards. These were shuffled and a figure selected by the agent by a random cut. The figure was shown to the other observer and both kept the visual image in consciousness for the span of the test. On a given signal, explained to the subject after he was seated in the arm-chair, he was to try to perceive which number from 0 to 9 the agent was thinking of and call it aloud. The regular run was 10 calls to a series and, as a rule, 2 runs were made for each individual during each experimental occasion. (I see now we might well have run much longer series.) There was no distinction here between clairvoyance and telepathy, since there were "present" both the figure on the card and the thinking of it. Also there were no provisions against possible "unconscious whispering". It was planned to progress to better conditions as results warranted. The subjects’ eyes were, however, closed by suggestion. After the first 12 subjects had been tested there were, for this test, 210 trials with 30 correct, 9 above the chance expectation (i.e., np, where n equals number of trials and p the probability of success for each trial, in this case 1/10; here np=21.). This is only about 3 times the probable error and not to be taken very seriously. For the whole 530 trials made by the 30 subjects, the level of deviation is 23% above the chance expectation and has still about the same ratio to the probable error, slightly less (2.6). This is, however, barely enough to encourage further work and not enough for a conclusion.

On one of the other techniques we had poorer results, not significant in any way. This was a simple test with 2 possibilities. The purpose in

p. 49

its use was to lessen the intellectual or rational element and come nearer to pure "guessing". The subject was given instructions that when the signal (2 taps) was given he would be impelled to raise one of his hands a little, which one being automatically determined by the corresponding one which the agent held raised. The order of this left or right sequence was determined by random card selection, using cards with L or R stamped thereon. The results of the 340 trials of this group were below chance expectation by 4 points; i.e., 166. This is, however, only about .6 of the probable error and is, of course, just "chance" variation.

The third type of procedure involved a still more overt response. A circle 5 inches in diameter was cut in a piece of cardboard, with the circumference marked with pencil into octants. A thumbtack was stuck in the table at the center of the circle. The subject was told that when the signal was given his second finger, which would be resting on the tack's head, would automatically move out to the circumference to that octant which the agent was fixing his gaze upon and was "willing" him to touch. There were only 245 trials of this; it was very slow work and gave only a low positive deviation from np, chance expectation; namely, 4.4, which is 14% above np but only 1.27 times the probable error. We have, then, the following little table of results:

TABLE I

Extra-Sensory Perception in Hypnotic Trance

Lundholm and Rhine, Fall, 1930

[paragraph continues] The value of these results lay entirely in the encouragement they gave us to continue. It was clear that, unless they were unusual, the mere continuance at the same rate would soon produce considerable significance. For instance, in 5,000 trials at the rate of our 530 on numerals the positive deviation would rise to the very convincing figure of 8 times the probable error. The only way to ascertain if our preliminary tests were unusual or not was to continue.

Fortunately, we found that we did not need to use the trance, that it did not seem to help and, of course, made the work much more laborious. We continued then, using only the normal waking state. But to explain the circumstances of the change requires that we turn now to the work on pure clairvoyance which Dr. Karl E. Zener and myself were carrying on during the same period in which the research just described was being conducted.

p. 50

Dr. Zener had suggested that we try tests for clairvoyance on a large scale, using our college classes for subject material. We adopted a procedure similar to that of Miss Jephson. of enclosing cards in opaque envelopes, except that we used cards on which the numerals 0 to 9 were stamped instead of playing-cards. We chose the numerals as offering a simpler task, with a simpler problem in computation and evaluation. The 6 and 9 were distinguishable by the fact that all figures were upright and turned toward the face of the envelope with its marked corner to the right.

The envelopes were doubly sealed, once with shellac, and were given, 5 to each, to the students with instructions to choose a quiet occasion and try to guess the numeral from 0 to 9 on the card within. The record was to be made on the envelope. Out of the 495 returned envelopes there were 60 correct guesses, or a positive deviation of 10.5 or 22% above. It is only 2.3 times the p.e. Linzmayer was highest with 3 correct in 5 trials.

In the next experiment the letters of the alphabet were used. Only 170 envelopes were returned and of these only 8 were correct. This is 1.46 above the np or chance value (i.e., 170×1/26). It is 12% above chance expectation but, of course, quite negligible in probability value with this small number of trials.

Thus far the computation by the mean square formula of the total significance of the results from the different experiments (i.e. by taking the square root of the algebraic sum of the squares of the values for X, where X equals the deviation divided by the probable error) had risen only to 3.8, which is only of borderline significance. And the next three experiments added nothing to this reservoir of value. In fact, they even lowered it slightly (to 3.6).

In the three further experiments, conducted in much the same way, the symbols used were 5 simple designs: circle, rectangle, plus sign, wavy lines and star ( ). These were mainly chosen by Dr. Zener, with a view to avoiding undue overlapping, complication and difference in familiarity. They are still in use. (We have once since substituted a "heart" for the "waves" figure but later returned to the latter. I shall hereinafter call these cards the "Zener cards".) In the first use made of those there were 300 trials, and, although one subject got all five correct, the total hits were only 2 above np or chance. This is negligible. In the next series, 205 trials yielded only 35 as against 41 for np. This is 15% below chance expectation, the lowest percentage we have ever obtained in our regular experiments in a series of over 100 trials. This was, however, not significantly low at all, being only 1% times the p.e. The last series, 430 trials, yielded 82, when np was 86. This is within the p.e. itself.

p. 51

The entire 935 tests on the Zener cards came within the probable error range, on the negative side, with a deviation nearly equal to the p.e. As was mentioned above, this brought the total value of the positive deviation thus far gained in the extra-sensory perception tests down from 3.8 to 3.6. The 1,600 trials for pure clairvoyance may be compared in the following table:

TABLE II

Extra-Sensory Perception of the Clairvoyant Type
Class Experiments. Zener and Rhine.
(Autumn, 1931.)

These 1,600 trials were enough, we thought at the time, to show that there was no highly appreciable extra-sensory perception in those particular groups under those particular conditions. We turned then to another line of technique. Neither trance-telepathy nor class-clairvoyance had been impressive, though both had held out faint promises of evidence—if we went on long enough. We began scouting for special subjects by the quick and easy method of having friends, students,—anybody—call off the cards (Zener cards) in a shuffled pack placed facedown on the table before them. During the winter months I made a collection of these trial scores totalling 800, with 24 subjects, yielding 207 hits, with a positive deviation from chance expectation of 47. This deviation was our first really convincing result, since it was over 6 times the probable error and unquestionably significant. These results can be grasped most simply by many readers in terms of number of hits per 25 trials. These 800 trials averaged 6.5 in 25, while the chance average is 5 in 25.

These trials were all made in my presence, with fullest vigilance against deception, using cards that the subject had not had in his possession outside of my presence. Most of the subjects were friends or relatives or students from my classes. The subject usually looked at the back, sometimes picked the card up, but often did neither. He had no opportunity to learn the cards by marks on the backs, even supposing there to have been present a general visual hyperacuity. The chief advantage here over the tests given before was that when a subject did well he was encouraged to go on, when he did poorly he usually was not. This helped to guard the score somewhat by selecting the better scorers. For instance, I myself made only 15 calls in the 800, since I got only 2 correct in 15 trials. Mr. Mann has 105, because he got 37 correct in all, an average

p. 52

of 8.8 in 25, as against 5 for chance. But there were exceptions to this rule of selection. If one seemed to want very much to continue, hoping to improve, he was allowed to. The following table will give some idea of the distribution. The average per 25 trials is given, since it is the simplest basis of comparison:

TABLE III

Pure Clairvoyant Perception, Odd Tests, Winter 1931-32

Most of these figures cover more than one experimental occasion for the subject. We seldom ran over 20 trials per day, per subject. Mr. McLarty did; as did also Mr. Mann.

In order to avoid repeating description of conditions, I mention here the later results of the "Odds and Ends" that do not warrant individual presentation. Up to the end of the year 1932, there were 835 more odd trials at pure clairvoyance, yielding 208 successes, a gain of 41, which is 5.3 times the p.e.

Among this group were 100 trials by Dr. William McDougall, yielding chance average, 150 by Dr. D. K. Adams, also of our Department, giving 36 or an average of 6 per 25. Our greatest gain was the discovery of Cooper, who got 38 correct in 90 trials, a gain over np of 20, a deviation in itself 8 times the probable error. He has since done some most excellent work. For completeness, however, I must note that in these trials I did not myself supervise Cooper but asked another student, a friend of his, Mr. Harriman, to do it. Mr. Harriman, himself, got only 1 correct in 10, with the reverse arrangement. But if there were any doubt of Cooper's and Harriman's honesty, the further work of Cooper under supervision, reported in a later chapter, would adequately satisfy it.

We have, then, as a total of the odd clairvoyant tests 1,635 calls, yielding 415 successes, with a positive gain of 88, about 8 times the probable error. This is about an average of 6.2 hits per 25 trials. By these

p. 53

results the combined value of deviation over p.e. (X = √X²₁ + X²₂ ...) has risen from 3.6, where we left it after the trance-telepathy and the class-clairvoyance tests, to the quite respectable and undoubtedly significant figure of 8.9. This figure would, I think, satisfy any mathematician, since the odds against a chance theory are here somewhere around 100 millions to one. But, fortunately, there is no need to raise the question. This value is soon dwarfed by the towering scores of later experiments.

In much the same manner and conditions as these tests just described, Mr. Chas. E. Stuart, one of my students, carried out a series of exploratory tests, using at first the hypnotic trance and the undefined telepathy-clairvoyance technique used by Dr. Lundholm and myself. Later he adopted the waking condition and pure clairvoyance conditions we had come to use exclusively. At first, he used a set of symbols of his own ( ) but he changed later on to the Zener cards. It is especially interesting to note that Stuart obtained better results than did Dr. Lundholm and I—not with the same subjects, it is true, and any comparison is based on all the uncertainties of individual differences. But it is easy to see now that Stuart, himself a student, working with fellow students who knew him well, did not induce the restraint and self-consciousness which Dr. Lundholm and I undoubtedly did with many student subjects. Stuart's results also warrant the detail of a tabular statement. (See Table IV). For a description of the technique of the three procedures used see the experiments of Dr. Lundholm and myself above. Instead of the circle-8 test Stuart used the 5-drawings test mentioned above. It will be noted that the best results were obtained with the 5-symbol cards and this has been pretty much the general case. The values of X or deviation divided by p.e. are, in order, 4.1, 2.3, and 5.5. When combined by taking the square root of the sum of their squares, we have 7.24. Combined with our last figure of accumulated value against "chance", 8.9, we get the mean square, 11.5, as our more advanced fortification against the chance theory.

TABLE IV

Trance-Telepathy Tests, Conducted by C. E. Stuart, Spring 1931

p. 54

Following the experiments of Table IV Stuart went over to the pure clairvoyance tests, using the Zener Cards. The procedure was essentially the same as that used by me in the exploratory tests for clairvoyance, except that Stuart used to shuffle the deck after each 5 cards, unless the calls were checked only after the entire 25 were called. I myself seldom did this, since few subjects tried to keep track of the cards already called off and checked, and seldom were all of a given symbol drawn. At worst, however, the probability would very occasionally be increased to % for the last 5 calls. It was, however, a point in favor of Stuart's caution. He, himself, in running did not look at the cards; he held them behind his back in his own calling. See Table V for the results up to the fall semester, 1931.

TABLE V

Pure Clairvoyance, by Stuart as observer, Spring and Summer, 1931

The last figure given for the combined value of the ratio of positive deviation to p.e. was 11.5. Stuart's results in Table V alone contribute a value for X of 11.1. Combining these, then, in the proper manner, we arrive at 16.1, a value that renders the alternate theory of "random distribution" or "chance" still more hopelessly unacceptable.

Mr. Harvey L. Frick, a graduate scholar in our Department, had in May, 1931, just completed his Master's thesis on the subject "Extra-Sensory Cognition", in which he presented evidence from telepathic and clairvoyant drawing-tests carried out with considerable distance between the agent and himself as percipient. He undertook, then, to do some pure clairvoyant work as well. He was interested in the decline curve suggested by the work of Richet, ¹ Jephson, ² and Estabrooks, ³ and decided to run 100 clairvoyant trials per day for a time, calling suits on playing cards, and then total the results in order of 20's in the hundred. That is, he added up the results by the various 20's for all the runs made. He totalled after 9 days and secured a very striking decline curve. This meant

p. 55

900 calls, with 180 in each total for the columns of 20's. Chance expectation for 180 trials in calling the suits of playing-cards would be 45. The hits actually scored were 58, 50, 48, 38, 36. Subtracting 45 from the five totals of the hits in the serial 20's, we have left +13, +5, +3, -7, -9. The total deviation from chance for the entire 900 is not significant but the extremes of the decline are significantly separated. From +13 to -9 is a difference of 22. The probable error of the difference (p.e. diff. √p.e.²_A + p.e.²_B) is 5.5. This gives a ratio of 4 for the difference over the p.e., which is regarded as just barely significant.

Mr. Frick was urged to continue with his laborious task and he generously did. The curve, however, lost some of its smoothness and developed more irregularity. Naturally, since it must in any case be the expression of some form of mental configuration and since such configuration must be regarded as highly labile, we must not be surprised at the changes. In any case Frick's work contains an interesting suggestion. He ran a total of 3,120 trials, with a total positive deviation of 49 suits. This is only 3 times the p.e. But the internal comparisons are somewhat more significant. They can best be displayed in a table. See Table VI. Frick's 3,120 trials in clairvoyant card-guessing, interesting though they are in their decline relationship, raise the ratio of positive deviation to p.e. only slightly, from 16.1 to 16.4.

TABLE VI

Clairvoyant Perception of Playing Cards, H. L. Frick, Spring, 1931
Total Deviation from (np) chance by order of 20's

In the spring of 1931 Miss Miriam Weckesser, my sister in-law, then 15 years old, found she could do clairvoyant perception, if left alone, but did not do above chance with the telepathic condition added. She was encouraged to work at it from time to time through the succeeding

p. 56

year and totalled 1,050 trials, yielding 266 hits, which is 56 above np; this is 6.6 times the p.e. None of these were witnessed by anyone else, but are interesting for certain points. First, is the fact that she could only work when alone. The suggestion of the inhibiting effect of divided attention was a very good one and is brought under experimental treatment in Chapter 8. The second point of value is that Miss Weckesser lost her ability after those 1,050 trials. She declined through the last 475 trials made from December 1931 to June 1933, averaging only 5.8 per 25. Divided into 3 parts, her results are as follows: 1st 350, average 6.4 hits per 25; 2d 350, 6.9 per 25; 3rd 350, 5.8 per 25. At the time of decline she was offered, with a view to its effect, what was for her a substantial reward for scoring at her usual height but this had no deterring effect upon the decline. Of course, there is no least question of her honesty in my mind or I should not use the data even to this limited extent.

Again reporting largely for completeness, we should give the details more fully on Mr. A. E. Lecrone's experiment in general extra-sensory perception, not differentiating between telepathy and clairvoyance. Mr. Lecrone, a student in my class during the summer of 1931, became deeply interested in my results, but was courteously but frankly sceptical. He therefore (as one could only wish all sceptics would be spurred to do) set to work to give the question a fair test. He used the Zener cards and followed the procedure of having the agent look at the card while the subject attempted to perceive it. Mr. Lecrone's conditions were not perfect but they served after 1,710 trials to convince him of the reality of extra-sensory perception. The most important point in his work, however, is the fact that, assuming that telepathy was primarily involved, the function from Lecrone's mind to his friend's worked about thrice as well as when reversely directed.

TABLE VII

Lecrone's Experiment, Telepathy plus Clairvoyance, Summer, 1931

One more large group of data belongs in this miscellaneous collection, namely, that supervised during the year 1931-32 by Mr. J. G. Pratt, an assistant in the Department. Using the regular procedure already described for pure clairvoyance testing, with the Zener cards in packs of

p. 57

25, and with the checking done either after every 5 calls or after the whole pack of 25 was called, Mr. Pratt collected data on 10,035 trials with 15 student subjects, including himself. Mr. Pratt also supervised 1,975 trials with Mr. Hubert Pearce but these results will appear in the chapter devoted to Pearce's work. The other 10,035 trials were supposedly for exploratory purposes, although through misunderstanding they went far beyond this limit. The total yield was 2,151, only 144 above np., 5.3 times the p.e. The most interesting feature here is the fact that Pratt himself declined in his capacity for clairvoyant perception, as did Miss Weckesser. Both of these worked alone. Both, also, had strong interest in continuing and even in raising their scores. Pratt's results are partly itemized in Table VIII.

TABLE VIII

Clairvoyant Perception Tests, by Pratt as Observer, 1931-32

There are no other large "batches" of data except those about to be reported in the chapters named for the subject producing them, with the following exceptions (This laborious explanation must be given since many will want to know if anything is omitted—especially of the lower scores): for a year and a half, now, we have followed the policy of giving a new subject a preliminary test, the results not to be taken into the record no matter what they are. When the subject gets 3 hits in 10 or better, the record can be started on the next trial following but must be so designated at the time. If, during the performance for record, the score drops below a 6 in 25, it is legitimate to quit scoring for the time. These preliminary test data have been rejected. My estimate of them, from memory and my own experience, is that they were on the whole above chance-average anyhow and probably represent only a few hundred trials with those subjects who later came into good scoring. But there have been a few subjects who have "practised" for thousands of trials without getting above the chance expectation (np). No conclusion of this report would be changed or appreciably weakened by including these practise data. For that matter, no amount of failing to score above chance by any number of other individuals can seriously affect our judgment of the results of those who succeed, since an individual ability is in question.

p. 58

Also, I have lost a few small records by mislaying them. I remember them in general but cannot state them exactly. I should estimate from 300 to 500 as a liberal total for these. They were mainly data taken at odd moments with a neighbor and his wife. She fairly consistently ran above chance expectation and he ran below a great deal of the time. More I cannot recall; there are probably other lost bits, but they can in no event be of consequence here.

I have finally a number of scraps of data for record that do not fit in anywhere. Some of them are very good and some are poor. I cannot be sure, of course, that tomorrow or next year I will not find a sheet of data stuck away absent-mindedly in a book I was reading or holding at the time. There may have been through the course of conducting or directing these 90,000 tests such lapses as these. But I am fully confident that there is no batch of forgotten and unreported data that would alter the final "anti-chance" value (D/p.e.) by so much as half a unit. That is safe, and there we will leave it. The remnants are given in Table IX.

TABLE IX

Clairvoyant Perception; Odd Data, 1932-33

Someone may be interested in the cumulative value of X for all the data (23,550 trials) reported in this chapter; it rises to 22.7, a value of indisputably great significance.

None of the data reported in this chapter is essential to any single point made in this report. On every score better results under better conditions are available. Why then, the labor and expense of publication—if, indeed, the answer is not obvious? To give the reader the opportunity to see the whole of the case, in its infancy as well as later, at its worst and most doubtful levels as well as at the more striking stages; and to reassure him that no important block of facts is omitted. Also, some of the weaknesses of these beginnings one has only to read here to avoid. They may help to guide those who will repeat these tests.

Footnotes

54:1 Richet, Charles, La Suggestion Mentale et la Calcul des Probabilités, XVIII, 1884.

54:2 Jephson, Miss Ina, Evidence for Clairvoyance in Card-guessing, Proc. S.P.R. 38: pp. 223-271, 1928.

54:3 Estabrooks, G. H., A Contribution to Experimental Telepathy, B.S.P.R., Bulletin V, 1927.