Final Report to Fundação Bial

FINAL REPORT TO

FUNDAÇÃO BIAL

FACTORS AFFECTING THE RELATIONSHIP

BETWEEN HUMAN INTENTIONALITY AND

THE HEMOLYSIS OF RED BLOOD CELLS

John Palmer, Ph.D., Stephen Baumann, Ph.D. and Christine A. Simmonds, Ph.D.

TABLE OF CONTENTS

Extended Abstract 1

Project Staff 3

Introduction 4

Method 10

Results 16

Discussion 24

References 30

Appendices

Flow chart of test session 33 Information / consent form for healers 34 Information / consent form for non-healers 36 Post-Intention Questionnaire 38

EXTENDED ABSTRACT

This experiment was a conceptual replication of an experiment by William Braud (1990), which seemed to demonstrate an effect of human intentionality on the process of hemolysis of red blood cells in vitro. In the process of preparing a report for the present experiment, Palmer discovered an artifact that could account for Braud's main finding, which was significant variability of the hemolysis scores around mean chance expectation. However, there remained suggestive evidence of a directional effect of hemolysis retardation, bolstered by a significant association of such retardation with elevation in the earth's geomagnetic field (GMF) on the day prior to the test session. The present experiment adopted Braud's procedure of inducing hemolysis by adding 50 l of blood to 3ml of .425% physiological saline solution in a cuvette. The process of hemolysis was measured by placing the cuvette in a spectrophotometer that recorded the changes in the absorption of light by the specimen every 1 sec over a 1 min period. Each session consisted of 1 test and 1 baseline run in counter-balanced order. Each run consisted of 8 trials, with the intention to retard the hemolysis active on trials 4 and 5 of the test run only. For each run, a t-score was computed comparing the results of trials 4 and 5 to the other trials. Because the hemolysis scores can be heavily influenced by small variations of the exact times in the hemolysis process at which measurement commences, the t-scores were computed as the residuals around the regression line of the relationship between the start times and the change in absorption (hemolysis) over the 1-min trial. In other words, the effect of start-time variation was controlled statistically. The final hemolysis scores were the differences between the t-scores for the test and baseline runs. Lower scores on the test run relative to baseline indicated a tendency toward hemolysis retardation (the intended effect), whereas higher scores on the test run relative to baseline indicated a tendency toward hemolysis acceleration. The participants (Ps) were 20 practicing spiritual or psychic healers, who completed 2 sessions each, and 40 non-healers who completed 1 session each. The DC component of the GMF surrounding the cuvette in the spectrophotometer was systematically manipulated, such that in half the sessions the field was reduced to near 0 whereas in the other half it was close to its normal value, .5 Gauss. Ps were not aware of any trials except those on which they were to attempt influence of the blood cells. The intention trials were preceded by taped relaxation suggestions and were accompanied by soothing music and seashore sounds (optional for the healers). During the remainder of the sessions, Ps completed a revised version of the Hartmann Boundary Questionnaire (BQ) and the Spiritual Transcendence Scale (STS). Non- healers completed a post-intention questionnaire (PIQ) asking about their attitudes toward healing and their orientation toward the experimental task. The healers, on the other hand, completed an in depth interview reflecting similar themes. The hypotheses predicted (1) overall significant hemolysis retardation, (2) more retardation by the healers compared to the non-healers, (3) more retardation with the GMF present rather than absent, (4) higher global GMF for retardation sessions than for acceleration sessions 1 day before the session, (5) more retardation for high compared to low scores on the BQ and STS, (6) higher scores by healers than non-healers on the BQ and STS, and (7) more confidence in success in the hemolysis task among high scorers on the BQ and STS. The only hypotheses to received significant confirmation were 6 and 7, with respect to the STS but not the BQ. There was also suggestive support for Hypothesis 4, with the GMF 1 day before the test session being higher for retardation sessions than for acceleration sessions

(p = .081, one-tailed). Combined with Braud's finding, we conclude that there is strongly suggestive but not conclusive support for this hypothesis. We also found 2 significant post-hoc effects. Among the non-healers only, Ps 30 or younger tended to retard hemolysis while those 35 or older tended to accelerate it. For the younger group, success at retardation was significantly related to thinness of psychological boundaries on the BQ, a finding consistent with Hypothesis 5. The second finding was complicated. Considering only 1st sessions, Ps who received the test run 1st significantly retarded hemolysis if the manipulated GMF was off and significantly accelerated hemolysis if the GMF was on. However, this finding applied not to the primary hemolysis scores described above, but to the scores of the test and baseline runs combined. (The result is a psi effect because the experimenter conducting the hemolysis was blind to run order.) We consider both findings as only suggestive, pending successful replication. The results were discussed with reference to the problem of identifying the most likely sources (Ps or experimenters) and the evidence for non-intentional psi, which one must appeal to in order to explain the psi effect that included the baseline run. Assuming the results reflect genuine PK, they imply that the skills associated with a practicing healer do not improve the results over those obtained by non-healers, that the results can occur opposite to the person's intention (which could mean harming the patient in an in vivo application), and that healing might be most successful if performed on the day following higher than average global GMF activity.

PROJECT STAFF

There were some changes in the identities and roles of project staff from those specified in the original proposal submitted to Bial. Some of these changes were described in our interim report, although there have been a few additional changes since then. The proposal listed four prospective staff members and their prospective roles as follows: John Palmer, Ph.D. (Co-Principal Investigator): Design of study, supervision of study, statistical analyses, write experimental report. Stephen Baumann, Ph.D. (Co-Principal Investigator): Design and implement GMF manipulation, set up hemolysis test and train Simmonds to collect data, write experimental report. Christine Simmonds, Ph.D.: Collect hemolysis data, assist in subject recruitment and other aspects of the study (consistent with her need for experimental blinds), score personality tests, statistical analyses. Colleen Rae, M.A.: Recruit participants, conduct Session 1, supervise participants in Session 2. Anne Poole, R.N. (Registered Nurse): Collect blood samples.

The changes are as follows:

Dr. Palmer: At the end of March 2003, Dr. Palmer left the Rhine Research Center (RRC) to take a job at the University Hospital Zürich in Switzerland. At that time data collection had just begun. Prior to his departure, his main duties were to develop the detailed research protocol and train those who would be testing the participants to administer the protocol. While in Zürich his main project-related duties were to consult with the staff at the RRC as questions arose, conduct the data analyses, and write most of this report.

Dr. Baumann: Dr. Baumann's role has been essentially the same as listed above. Most of his time has been spent designing, constructing, installing, and testing the apparatus used to manipulate the geomagnetic field around the blood samples during hemolysis. Much of the development work he was able to do at his home base in Pittsburgh. He made frequent trips to Durham during the project period, during one of which the device was installed and tested. During his visits he also helped to explain to the relevant experimenters details regarding implementation of those aspects of the protocol related to the hemolysis. He also contributed to the writing of this report.

Dr. Simmonds: After Dr. Simmonds arrived, it became apparent that it would be best to have her assume the role in the test sessions that was originally assigned to Ms. Rae. The main reason for the change is that we decided to collect interview data from our healer participants during the control periods, and conducting such interviews was a particular skill of Dr. Simmonds that she had already been applying in other research. When Dr. Palmer left for Switzerland, she also assumed the role of in-house project leader. She also scored the personality tests and entered data into computer files sent to Dr. Palmer for analysis. Finally, she contributed to the writing of this report.

Ms. Rae: As noted in our interim report, Dr. Rae was replaced by Sally Ann Drucker, Ph.D., before the testing began. Dr. Drucker brought with her the advantage of past

experience in parapsychological testing. Her role in the testing was that assigned in the proposal to Dr. Simmonds, which consisted primarily of conducting the hemolysis. However, because of the delay in beginning the formal testing, this testing could not be completed before she had to leave the RRC. In fact, she was able to complete only 10 of the 80 sessions. We were very fortunate that Dr. Simmonds was able to recruit as the replacement Debra Hlavaty, MA. Ms. Hlavaty had a background of working in biological science laboratories, and this employment gave her experience working with procedures similar to the hemolysis protocol in our study. Thus her background was actually superior to that of Dr. Drucker with respect to the particular duties she was asked to perform.

Ms. Poole: Ms. Poole's role was as specified in the proposal. However, since we decided to collected blood only from a few individuals rather than from each participant, she made fewer blood draws than originally anticipated.

Finally, as a result of the administrative problems being faced by the RRC, Palmer decided in May 2004 to transfer grant funds from the RRC to a private account. The administration of the grant with respect to disbursements and bookkeeping were assumed at that time by Ms. Lauren McGlynn, who is not affiliated with the RRC. Ms. McGlynn will prepare the final budget report to be submitted to the Foundation.

INTRODUCTION

The purpose of this experiment supported by the Bial Foundation was to perform a conceptual replication of an experiment by William Braud (1990), in which volunteer subjects claiming no special psi talents were asked to retard the rate of hemolysis of red blood cells in a test tube from a remote location. Hemolysis refers to the bursting of the red blood cells, which can be induced by placing them in a solution differing from the saline solution they inhabit in the human body. The research is important because of its relevance to psychic healing. In effect, the participant is being asked to "heal" the red blood cells by retarding the destructive hemolysis process. If this physiological process can be affected by psychokinesis, or PK (the type of psi presumably in operation), it increases the likelihood that other physiological processes can likewise be so influenced. There are three advantages to studying the healing process in vitro (outside the body) rather than in vivo (inside the body), although the latter more closely resembles what happens in the "real world". First, the in vitro approach eliminates placebo effects as a possible interpretation of the results. Second, it allows for a greater degree of experimental control. Third, it eliminates risks resulting from psi-missing, which occurs commonly in psi tasks (Rhine, 1969). Psi-missing in a healing context would translate into the participant being harmed by the healing effort. The advantages of working with red blood cells are (1) it is a process that occurs reliably over a short period of time amenable to multiple experimental trials, (2) it is a basically reliable process which at the same time exhibits enough random variability from trial to trial, or is "labile" enough (to use Braud's term) to make PK a realistic possibility, and (3) the process can be measured in real time in the test environment without having to send the multiple specimens required by the design to an independent laboratory for analysis, which would have been prohibitively expensive. There is a growing body of evidence indicating that people can influence biological systems by PK. We extensively reviewed this research in our proposal for this experiment and will not duplicate that text here. Instead we will focus on a more detailed discussion of the Braud (1990) experiment that inspired our investigation.

Braud (1990) tested 32 volunteers in his study. (It is not reported if any were practicing healers.) Fourteen of the participants (Ps) donated their own blood to be used in the experiment, whereas the remaining 18 attempted to influence blood that was not their own (presumably blood supplied by the other 14). P was alone in a room separated from the room in which the hemolysis procedure would be performed by the experimenter (E). The hemolysis procedure consisted of transferring blood to a test tube containing a concentration of .425% physiological saline. After the contents were mixed by shaking, the test tube was placed inside a spectrophotometer that measures the hemolysis. When red blood cells burst, the solution becomes more transparent. The spectrophotometer charts the progress of the hemolysis by passing a light beam through the test tube and recording the amount of light that is absorbed at specified time intervals. Sixty hemolysis measurements were recorded during each trial, i.e., 1 per sec. These measurements can be graphed, demonstrating the time course of the hemolysis during the run. Each session consisted of 4 15-minute trials. Two of these trials, as determined by a random process, were experimental trials during which Ps were to concentrate on retarding the hemolysis taking place in the test room, while having the opportunity to observe a photo of healthy blood cells if they so chose. During the control trials, Ps were asked to not think about the red blood cells, and if that was impossible, to concentrate on having the hemolysis occur at the normal rate. E was blind as to which 2 trials were the experimental ones. Prior to the first trial Ps listened to a progressive relaxation and guided imagery tape over headphones. During each of the 4 trials, E performed the hemolysis procedure on either 2 or 8 test tubes. This manipulation was introduced to allow a test of decision augmentation theory, abbreviated DAT (May, Spottiswoode, & Utts, 1995b), which maintains that apparent micro- PK effects (under which rubric would fall the hemolysis task) are in fact caused by ESP. For example, the experimenter responsible for randomising the order of experimental and control trials might use precognition to assign as experimental those trials in which the hemolysis effect was going to be stronger anyway. The theory makes differential predictions as to the results in the 2 and 8 tube cases, hence the manipulation. The main analysis indicated no significant difference between performance in the experimental and control trials, nor did it make any significant difference whether Ps attempted to influence their own blood or someone else's blood. It also made no difference whether the scores were based on 2 or 8 test tubes. However, there was highly significant variability among the scores of individual subjects. Of the 32 subjects, 9 achieved independently significant scores, whereas only 1.6 would be expected by chance, assuming an alpha criterion of .05, two-tailed. Of the 9 significant sessions, 7 were in the psi-hitting direction and 2 were in the psi-missing direction. The author concluded from the excessive number of significant sessions that PK had been demonstrated in the experiment. During the preparation of this report, Palmer discovered an artifact that could explain this significant finding. The main control introduced in Braud's (1990) study was the random ordering of the experimental and control trials, an order to which E was blind. Although this control is effective with respect to directional findings, it is not effective with respect to variance or bi-directional effects of the type at issue. To illustrate, assume that E in the Braud (1990) experiment could by some means slightly influence the rate of hemolysis on individual trials. Assume further that E consciously or subconsciously made a guess for each session which 2 trials were experimental, and that the results of this guessing were purely random (no ESP). Finally, assume that E then unintentionally and unconsciously performed the hemolysis procedure such that hemolysis scores would be significantly higher in the expected experimental trials than in the expected control trials. There are six possible combinations of

trials in the experimental and control conditions, as illustrated below. (Note that the order of the trials within each E or C pair doesn't matter.) Assume that for a particular session E expected trials 1 and 2 to be experimental. (These trials are in boldface in the illustration below.) It can be seen that 2 of the 6 sessions (first and last rows) line up (bi-directionally) with E's guesses. One would expect significant results in these cases. In the other 4 cases, E is right once and wrong once, so the effects should cancel leaving a net effect of near 0. (Note that the probabilities are the same regardless of which 2 trials are selected as experimental.) Thus, the model predicts that significant outcomes should appear for one-third of the sessions (33%). In fact, Braud (1990) obtained significant results in 9 of 32 sessions, or 28% - very close to what the model specifies. The model also predicts that the number of positive and negative directional outcomes (psi-hitting or psi-missing) should be the same. In fact 7 of the 9 significant outcomes were in the hitting direction, compared to the chance expectancy of 4.5. Although the obtained ratio is encouraging, the departure from the null hypothesis of an even split is not quite significant, exact binomial p = .090, one-tailed. In conclusion, the significant result from Braud (1990) conforms satisfactorily to the artifact model and thus cannot be claimed as evidence for psi. E C 1 2 3 4 1 3 2 4 1 4 2 3 2 3 1 4 2 4 1 3 3 4 1 2

So how could E have produced higher hemolysis scores in some trials than in others? The most likely means would be to subtly vary the starting times of the measurement process, a factor that Braud (1990) was aware of himself. We quickly discovered from our research that the 60 hemolysis measurements do not decrease linearly over time. As illustrated in Figure 1, they follow a decelerating curve. This means that the rate of hemolysis is much greater at the beginning of the process than at the end. Thus it makes a big difference what stage of the process one measures on a given trial. If one starts early in the process, the rate of hemolysis over a 1 min period (reflected by the slope of the curve) will be greater than if one measures later in the process. If E started on the average at slightly different times in experimental and control trials, an artifact could be generated. Although Braud (1990) reported some effort to keep the hemolysis measurement procedure highly uniform, no data are reported on how successful these efforts were in reducing the variability of start times. We found in our data that the artifact is a quite strong one. We had 160 trials in our experiment, each involving 8 cuvettes or hemolysis measures. For each trial we computed the correlation between the absorption score at the start time with the change in the average absorption score from the first to the last 5 sec of the 1 min measurement period, the same change (hemolysis) measure Braud (1990) used. These 160 correlations ranged from +.051 to +.996, with a mean of +.703. We probably had more variability in our start scores than Braud did, because our E had to overcome the additional hurdle of closing the door of metal cage (to be described later) over the specimen before measurement could begin. However, she became quite proficient at this technique over the course of the study and some of the start-time

standard deviations were quite low. However, they would have to be nearly 0 to erase the artifact. We found that the correlations described above correlated only a modest +.251 with the start-time SDs. Also when we restricted our sample to trials with SDs of .01 (representing a range of about .03 units on the absorption scale) or less for the start times (22.5% of the total trials), the mean correlation of .703 dropped only to .510. Fortunately, this artifact can be controlled for statistically (which is what we did), but Braud (1990) reported no such adjustments.

Figure 1. Decline in absorption of light over a 1-min interval in a hemolysis trial from the current experiment.

The one piece of evidence for psi in Braud's experiment comes from a separate paper that reports significantly (p = .023, one-tailed) greater activity in the earth's geomagnetic field (GMF) on the day preceding those hemolysis sessions showing a net decrease in hemolysis (psi-hitting) in experimental trials compared to control trials (Braud & Dennis, 1989). This finding refers to the direction of psi scoring and thus the artifact does not apply. Although this report combines the results of the formal experiment described above with those from an earlier pilot study, and the 1-day-before time was apparently selected post-hoc from up to 7 possibilities (including the day of the testing), the fact remains that a significant outcome was obtained that merits a replication attempt. There is evidence from several studies that the GMF is relatively quiet during successful laboratory ESP trials (at least under certain other conditions) as well as the time of spontaneous ESP experiences (Persinger, 1989), and one

might logically expect high GMF activity to characterize the reciprocal PK process. The fact that Braud and Dennis (1989) found the effect for the day prior to testing rather than the day of testing is puzzling. A similar temporal displacement of the effect was also found by Adams (1985), in a review of the association between GMF and remote viewing trials. Like Braud, Adams also used the Ap index of GMF, whereas the studies reviewed by Persinger (1989) that showed effects on the day of testing used the AA index. This result, combined with the 7/2 split in favor of the predicted direction of the hemolysis effect among the high scoring participants, suggests that there might have been a real directional effect in Braud's (1990) data. If this is the case, the absence of a control for the effect of variation in the starting times of hemolysis measurement increases the likelihood that the results were due to experimenter psi (cf., Palmer, 1997). Our own experiment was intended as a conceptual replication of Braud (1990); there were too many differences in procedure for it to be considered a strict replication. We will now list the most important of these. First, for reasons having at least partly to do with his DAT manipulation, Braud had his Ps concentrate for 15 minutes. We think this is too long to expect a person, particularly an ordinary volunteer, to maintain the level of attention presumably required, so we markedly reduced the concentration time to 2.5 minutes (although it was presented to the Ps as 5 minutes). Second, we think it is unlikely that many Ps could succeed in not thinking about the hemolysis task during the control trials, and we consider the best control to be one in which Ps are not even aware that data are being collected. Thus in our experiment, not only were Ps not told that control trials were taking place, but during this period they were actively engaged in other tasks (filling out questionnaires or being interviewed by an experimenter about their psychic experiences) that would occupy their attention. Third, in an effort to minimize contamination of our results by DAT effects, we avoided making random decisions wherever possible. Most notably, the assignment of trials as experimental or control, as well as the status of the GMF manipulation, were determined by simple counterbalancing rather than randomly. This was done in a way that kept the E who performed the hemolysis measurements blind to the relevant designations. Fourth, for practical reasons discussed later, none of our Ps attempted to influence their own blood. Recall that Braud (1989) found that those who attempted to affect their own blood did not score significantly different from the other Ps. Our procedure allowed us to have most of the blood drawn from the same person, which presumably reduced unwanted between-session differences in blood composition (although we have no evidence that such differences are consequential to the hemolysis process.) In addition to the procedural modifications described above, we introduced two major independent variables into the design. First, we tested a sample of professional psychic healers to compare with a sample of ordinary volunteers that is likely comparable to Braud's sample. Second, in keeping with the GMF findings of Braud and Dennis (1990), in addition to correlating our hemolysis results with the general index used by Braud of the average GMF activity around the world, we manipulated the strength of the local GMF around the blood specimens while they were being lysed. Several difference between the naturally occurring GMF and our manipulation are of note. The natural GMF surrounds the research participant, and current theory suggests that psi results from the effect of this GMF on the brain of this individual (Persinger, 1989). The manipulated GMF, on the other hand, only affected the blood specimen and could not directly influence the participant. The rationale for the manipulation is that the participant might psychically recruit this energy to affect the hemolysis at the test site. It of course is possible that in the absence of a manipulated GMF the same process might occur; for example, in the

hemolysis studies discussed by Braud and Dennis (1989), participants may have used peak GMF amplitudes that happened to coincide with the test trials to retard the hemolysis. Our experiment provides a test of both possible mechanisms. We had originally hoped to manipulate variations in the GMF (the AC component), as this is the parameter that has been associated with psi in the research literature (Persinger, 1989). However, this type of manipulation proved to be prohibitively costly, so we manipulated instead the static component of the GMF. There is reason to suppose that this manipulation could still be effective. The variation of the GMF component is considered paramount because of its proven relation to physiological processes (Persinger, 1989). For interaction with the test specimen it could well be the static component that is key. Finally, we incorporated in the experiment several psychological tests that might be expected to correlate with objective as well as subjective success in the hemolysis task. The first of these is a revised version of the Hartmann Boundary Questionnaire, abbreviated BQ (Hartmann, 1991; Rawlings, 2001-2002), which is designed to measure the extent to which persons erect in themselves ego “boundaries” that tend to demarcate autonomous or semi- autonomous psychological processes, either intra-personally or interpersonally. Psychic healers often report a sense of "merging" with their clients, which sounds very much like the dissolution of psychological boundaries between the two persons. Also, since psi in general is conceived of as a function of the unconscious mind, the unblocking of unconscious implied by the notion of thin boundaries might be expected to facilitate psi (Rhine, 1953). In the only published parapsychological study we know of that included the BQ, Richards (1996) found thin boundaries to be positively related to self-rated success in a psychic training exercise geared toward helping another person with their personal problems. For similar reasons, and because healing practice is often associated with spirituality, we also have included the Spiritual Transcendence Scale, abbreviated STS (Piedmont, 1999, 2001), which is designed to measure the ability of individuals to stand outside of their immediate sense of time and place, and to view life from a larger, more objective perspective. The transcendent perspective is described as one where the person sees a fundamental unity underlying the diverse strivings of nature. The preceding considerations led to the generation of several hypotheses for our experiment, as indicated below:

H1. For the sample as a whole, hemolysis will be retarded during the concentration period as compared to control periods. H2. Healers will be more successful in the hemolysis task than non-healers. H3. Hemolysis will be retarded more successfully when the GMF surrounding the blood specimen is on then when it is off. H4. The hemolysis scores will be positively correlated with a general measure of the earth's GMF on the day preceding the test session. H5. High scorers on the BQ and STS will more successfully retard hemolysis during the concentration periods compared to control than will low scorers. H6. Healers will score higher on the BQ and STS than non-healers. H7. BQ and STS will correlate positively with subjective estimation of success in the hemolysis task.

METHOD

Participants Healers. Psychic or spiritual healers were recruited by announcements at Rhine Research Center events, word of mouth, fliers distributed in the local area, and local magazines with a focus on spirituality. It was necessary for inclusion in this group that Ps define themselves as “healers”, and members of this subsample often advertised themselves as spiritual healers and master practitioners in their particular field for a significant period of time. The healers came from a variety of backgrounds, including but not limited to Reiki healing, LaHochi healing, and therapeutic touch. The subsample consisted of 3 males and 17 females ranging in age from 33 to 71, with a mean of 49.84. Non-healers. Non-healers were recruited by announcements at Rhine Research Center events, distribution of fliers in the local area, and advertisements placed in local weekly newspapers. Six of these had undertaken introductory courses in healing methods, but none defined themselves as “healers”. The subsample consisted of 12 males and 28 females ranging in age from 15 to 63, with a mean of 41.36.

Questionnaires

Boundary Questionnaire. At 146 items, Hartmann’s (1991) original Boundary Questionnaire was considered too long and too time consuming to be included in this study. Rawlings’ (2001-2002) 46-item questionnaire was statistically derived from the original long version of the BQ. It has an alpha coefficient of .74 and correlates .88 with the original BQ. It has 6 subscales: Unusual Experiences (UE), Need for Order (NFO), Trust (Tr), Perceived Competence (PC), Childlikeness (Ch) and Sensitivity (Se). Rawlings found that these subscales do not correlate very highly with each other, but each has an alpha ranging from .65 up to .80. For each category, a thick boundary person sees marked distinctions between the elements defined by the category, whereas a thin boundary person tends to merge them. For example, thin boundary persons keeps their thoughts and feelings separate from each other, whereas in thin boundary persons they become mixed together.

Spiritual Transcendence Scale – Revised. The STS consists of 23 items broken down into 3 subscales: Prayer Fulfilment (10 items), Universality (7 items), and Connectedness (6 items). Each item has 5 response alternatives ranging from strongly agree to strongly disagree in a Likert format . Approximately half of the items are reverse scored. Normative data indicate that total scores differ according to age and gender (Piedmont, 1999). The scale has good external validity and generalises across different religious groups and a culture outside of the US (Piedmont, 2002). Spiritual transcendence has been argued to be the 6th aspect of personality in relation to the standard Five Factor model (e.g., Piedmont, 1999).

Post-intention Questionnaire. The PIQ is a 10-item rating scale developed by Simmonds that asked Ps about their beliefs regarding psychic or spiritual healing, personal experiences related to such healing, the method they used to protect the blood cells in the experiment, their state of consciousness during this effort, their motivation for the task, and the degree of success they expected. Six of the items were presented in visual analogue format, 1 was yes/no, and 3 were open ended. The PIQ was given only to the non-healers, as similar questions were included in an in-depth interview of the healers. A copy of the PIQ is appended to this report.

Lab Layout The experiment was conducted in one of two research suites at the Rhine Research Center. The suite used for the study consisted of four rooms, each of which was partly sound- attenuated with extra insulation in the walls and heavy wooden doors with rubber seals. The rooms principally used for the experiment were at opposite ends of the suite. The outer walls of these rooms were 12.75 feet apart (direct line) and separated by 4 walls. One of these two rooms, labelled the "intention room", was occupied by P and E1 (Simmonds). This is the room from which P attempted to influence the blood samples. The second of these rooms, labelled the "hemolysis room", was occupied by E2 (Drucker/Hlavaty) and housed the spectrophotometer. In this room the hemolysis of the blood was induced and measured. A third room, adjacent to the hemolysis room but not the intention room, was used for storage of materials and preparation of the physiological saline solution to which blood would be added for hemolysis. The blood draws also were done in this room. A diagram of the suite is presented in Figure 2.

Hemolysis Room

Prep. 10 Rm. ft 3.5

Intention Office

Room

7ft 3.5 in Figure 2. Layout of the research suite.

Experimental Design and Condition Assignments There were 80 sessions in the experiment. The 20 healers each completed 2 sessions and the 40 non-healers each completed 1 session. This meant that the healer and non-healer conditions each consisted of 40 sessions. Each session consisted of 2 runs, during each of which 8 consecutive 1-min hemolysis trials were conducted.1 One of these runs was labelled "test" and the other "baseline". Trials 4 and 5 of each run were labelled as "experimental" and the other trials as "control". The only difference between the test and baseline runs was that during the experimental trials of the test run Ps attempted to retard the hemolysis process, while during the experimental trials of the baseline run they did not intend to retard the hemolysis. In fact, Ps were not informed that hemolysis measurements were being made at any times other than those corresponding to the

1 In contrast to Braud (1990), as well as to our description of his study above, we prefer to designate each hemolysis measurement as a "trial" and the sum of these 8 trials as a "run".

experimental trials in the test run. (All the other trials in the experiment can reasonably be thought of as controls, although to avoid confusion we refer to what might otherwise be called the control run as "baseline".) The primary dependent variable in the study was the hemolysis scores, which reflected the decrease in the average absorption of light passing through the lysed blood samples from the 1st 5 sec of the 1-min measurement period to the last 5 sec of the measurement period. The two manipulated independent variables in the experiment were (1) the order of the test and baseline runs within the session and (2) whether the GMF surrounding the blood samples during hemolysis was on or off. E3 (Palmer), who was not involved in testing the Ps, created the counterbalanced orders for both independent variables. The within-session order of the test and baseline runs was varied according to an ABBA sequence, separately for the healer and non-healer sessions. The order of runs for the 2nd healer session was always the opposite of the order in the 1st healer session. For the order of runs, E3 placed two kinds of ESP cards, squares and stars, in sealed opaque envelopes with the session number and subsample written on the outside of the envelope. The order of the cards reflected the desired ABBA sequences. However, he did not determine which symbol was to represent test and which was to represent baseline. This designation was determined by E2 by a coin flip prior to the 1st session, after she had received the deck from E3. E2 did not reveal to E3 the result of the coin flip, because E3 wanted to be blind to the run assignments while carrying out the preliminary phases of the data analyses. E2 was told that the run assignments were random, and she was thus blind to the run order designations as well. The GMF manipulation was supposed to alternate between Ps separately for the healer and non-healer samples, but due to an error in the transmission of the assignments from E3 to E1, one setting applied to the 1st session of each healer and the other setting to the 2nd session of each healer, creating a confound between session order and GMF setting. This problem did not apply to the non-healer subsample, and the GMF settings were orthogonal to the order of run types within the session for both subsamples. The switch positions on the GMF device were only labelled "1" and "2", and during the experiment only the builder of the device, Baumann, knew which of these positions corresponded to maximum and which to minimum GMF. The device emits no detectible sound, heat, or light regardless of how it is set, so E2 could not tell in which state the device was in. So E2, as well as E1, was blind to these settings.

GMF Device A digital magnetometer (F.W Bell model 7010, Sypris Test and Measurement, Inc.) with both axial and transverse probes was used to measure magnetic fields. The ambient GMF measured in the test room was approximately 0.5 G with a near-vertical orientation. However, inside the spectrophotometer the magnetic field was highly distorted, especially when the unit was turned on and producing its own magnetic fields. To approximate the static component of the GMF, it was decided to suppress the distorted field around the blood sample for half the trials and then artificially approximate it with a pair of Helmholtz coils around the sample for the other half of the trials. To shield against extraneous magnetic fields occurring inside the spectrophotometer measurement chamber, a box was constructed out of high-permeability mu metal. The box measures 11.4 cm wide by 16.3 cm long by 11.1 cm high. It fits snugly inside the chamber, and surrounds the cuvette holder containing the blood sample. A sliding door on top allows access for placing and withdrawing samples. Small apertures on the two side walls allow the spectrophotometer light beam to travel unimpeded through the sample during measurements.

To reimpose inside the shielded chamber a magnetic field of approximately the same strength as the static GMF, paired Helmholtz coils approximately 1 cm apart were wound around the cuvette holder in the center of the chamber and each supplied with sufficient current (~0.12 A) from a constant current source (Agilent E3642A DC Power Supply) to produce a 0.5 G field in the center of the cuvette holder. A switch between the power supply and the shielded chamber allowed the current to the Helmholtz coils to be toggled off or on, thus creating inside the cuvette either a negligible field or a magnetic field of approximately 0.5 Gauss. The technician performing the spectrophotometer measurements remained blind to the effect of the switch setting. With the switch to the coils in the Off setting (1) and the door to the shielded chamber closed, the static magnetic field measured at the level of the blood sample was < 0.03 G (e.g. at least 94% attenuated), depending upon height from the bottom of the chamber. With the switch to the coils in the On setting (2) and the door to the shielded chamber closed, the static magnetic field measured at the level of the blood sample was 0.5 G + 6% in the vertical direction, depending upon height from the bottom of the chamber.

Blood Samples It was our original intention to draw blood from each non-healer and have them attempt to influence their own blood. The healers would attempt to influence the blood of the non-healers. This plan was abandoned because of the fear of contamination by AIDS and hepatitis, concerns that were not as pronounced in the late 1980s when Braud (1990) conducted his experiment. Despite the standard precautions against contamination, such as rubber gloves, and in the case of one staff member injections against hepatitis, concerns remained. It was concluded that additional assurance could be provided by requiring that our blood donors all submit to blood tests for relevant diseases prior to offering their blood for the experiment, but it was considered neither desirable nor practical to have each of the non- healers submit to this added inconvenience and possible perceived insult. Therefore, we decided to abandon the idea of having Ps attempt to influence their own blood, but instead had blood donated by 2 persons not otherwise involved in the study and who had taken the appropriate blood tests. During the course of the experiment, a registered nurse collected a single 40 ml sample of venous blood from one donor and 6 such samples from a 2nd donor. The blood was drawn into 6 ml Vacutainer tubes containing Solution B anti-coagulant. The Vacutainer tubes were stored in a refrigerator kept at 4° C. Each donor signed a consent form before their first blood draw.

Hemolysis Preparation and Measurement

Preparation of Saline Solution. Following Braud (1990), the blood was to be added to a solution of .425% physiological saline to induce hemolysis. Before the experiment, 2 large bottles of .85% NaCL (physiological saline) were ordered from Fisher Scientific Corporation. As needed, approximately equal amounts of distilled water and the .85% saline were mixed by E2 in a beaker, yielding 700-800 ml samples of approximately .425% saline. The top of the beaker was covered with plastic wrap to minimize evaporation.

Preparation for the Hemolysis Runs. Prior to each session, E2 switched on a Unico S2100 Spectrophotometer (s-meter). A small amount of the saline solution was titrated until the percentage was exactly .425, as measured by a Fisher Scientific Digital Conductivity Meter. A clean syringe with graded ml markings was then used to inject exactly 3 ml of the

saline solution into each of 16 10-mm glass cuvettes located in a rack. The plastic caps were then put back on the cuvettes. After the s-meter had been allowed to warm up for 15 min, E2 entered the test parameters into the device's memory. Following Braud (1990), the wavelength of light to which the s-meter would be sensitive was set at 660 mμ. The period of time during which measurements were taken for each trial was set at 63 sec, and the inter-trial interval was set at 2 min. E2 also selected an initial delay of 14 sec, which represented the amount of time it took, following extensive practice with the procedure, for her to mix the blood and saline and then put the specimen inside the s-meter for measurement.

Procedure for Hemolysis Runs. A few minutes before the 1st of the 2 hemolysis runs, the Vacutainer tube containing blood was inverted 8 times, uncorked, and part of its contents transferred to near the top of a 10 ml Pyrex test tube. Both tubes were immediately corked and the Vacutainer tube returned to its rack in the refrigerator. The rack containing the cuvettes with saline solution was transferred to the hemolysis room and placed near the s-meter. The Pyrex test tube of blood was moved to this same rack. E2 wore rubber gloves during all times that she was handling the blood. E2 allowed a period of 10 min following her greeting of P to allow for the orientation period. (This time interval was reduced to 5 min for the 2nd session of healers.) When this interval had elapsed, E2 set her digital timer to 2 min and transmitted the 1st of 3 beeps over the intercom to E1 in the intention room (see Test Procedure below). She then conducted the calibration test for the 1st trial by taking one of the cuvettes from the rack, wiping off any moisture that had collected on the outside of the cuvette with a piece of soft tissue, and placing it inside the s-meter with the frosted sides facing perpendicular to the light beam (so the light would pass through the unfrosted sides). She then closed the door of the s-meter and proceeded with the calibration test, the result of which provides the baselines for the test absorption values. When the computer screen indicated that the calibration test was completed, E2 removed the cuvette from its holder and placed it back in the rack, leaving the door of the s-meter open. When the clock reached 23 sec, the actual trial began. E2 immediately pressed a button on the computer keyboard that began the countdown of the initial delay, which she could follow on the computer screen. She immediately affixed a plastic tip to a Model SC-300UL Finpipette, which had been set to withdraw exactly 50 μl of liquid. The pipette was then inserted into the Pyrex test tube 1 cm below the bloodline and 50 μl of blood withdrawn and transferred to the cuvette with the saline solution. [The volumes of blood and saline in the cuvette were the same as reported by Braud (1990)]. The cuvette was recapped and inverted twice to mix the contents. When the initial delay reached 14 sec, E2 in rapid succession placed the cuvette inside the s-meter, closed the lid of the GMF device, and then the lid of the s-meter. The computer screen provided E2 with a charting of the successive absorption values, and when the 63 secs had elapsed the trial was over. E2 immediately set her clock back to 2 min, saved the file with the digitized absorption values to the computer hard drive, ejected the used pipette tip into a BD Sharps waste canister, and proceeded with her preparation for the next trial. The above procedure was used for each of the 8 trials in each of the 2 runs. The only modification was to signal E1 as to the start and stop times of the intention period for P. Thus E2 transmitted the 2nd of the intercom beeps when the digital clock recorded 23 sec prior to the start of trial 4, and the 3rd beep immediately after the 63-sec measurement period for trial 5.

Procedure for Participants Prior to the start of each session, E1 set the GMF device inside the s-meter to “1” or “2” as listed by the order generated prior to the start of the experiment by E3. The envelope containing the code for the order of the runs in the session was placed outside the intention room to be opened just prior to the start of the experimental session. Prior to the testing period, E1 and E2 spent a few moments relaxing in the intention room, affirming the importance of the study and that the study would be a success. This served to set a good subjective mood for the running of the experiment for both experimenters. E1 met P, who was usually waiting in the lounge area of the Rhine Research Center (RRC), and brought him or her upstairs to the laboratory suite, where they sat down for a few minutes in the intention room. The experimental procedure was explained to P, who was then asked to sign the information sheet / consent form. Copies of these forms for both the healers and non-healers are appended to this report. Next, P was taken into the hemolysis room to meet E2 and see the sample of blood displayed in the Pyrex test tube next to the s-meter. At this point some Ps asked whether they could handle the tube containing the blood. Ps who asked to do this were given a pair of rubber gloves for this purpose. Any questions about the general procedure were also answered at this time. Once P was happy with the general procedure, both E1 and P left the hemolysis room, closing the door behind them.

Orientation Period. On return to the intention room, E1 set a stop clock to zero to begin timing for the 10-min orientation period. The timer was used to help E1 coordinate the activities in the intention room with E2's activities in the hemolysis room, and it was closely monitored by E1 throughout the session. After P was seated in the intention room, E1 went to open the envelope indicating the order of the runs. She then returned to the intention room and described the specific details of the study to P. Ps were told that there would be two halves to the study, with a short break time in the middle, when they would be offered some refreshments. Ps were then shown a Powerpoint display illustrating a single red blood cell undergoing the process of hemolysis. It was reiterated that P's task was to attempt to stop or slow down this process using mental intention and visualization. Non-healers were told that they would be given some suggestions as to how to go about this process. Healers, on the other hand, were told that they should try to employ methods that they usually use when healing as much as possible, within the constraints of being in a laboratory situation. Ps were given an example of the signal beeps that they would hear at various times during the experiment, and they were also shown a screen shot of a healthy whole blood cell, which they were told they could have on the computer monitor in front of them during the intention period. Healers were told that the experimental set-up was flexible; for example, they could switch the computer and screen off completely if they wanted to. At this point, the sequence of events began to vary, depending upon whether the test period or the baseline period was to be first. The test and baseline periods each lasted 25 mins, separated by a 15-min break, during which time Ps were offered some refreshments.

Test Period. Ps were told that there would be 3 beeps during this part of the experiment, the 1st beep signalling the start of the preparation stage, which was timed to last 10 min. Non-healers were told that at this point they would hear some relaxation instructions (in the voice of E1), including music ("Discreet Music" by Brian Eno) and sounds of the sea in the background. The music was optional for healers. For all Ps, the recording included affirmations for success in the voice of E1. Healerss were told that the wording could be

ignored, or other words substituted in their mind, if they preferred. At the end of the recording (which lasted about 8.5 min) P’s were told that they would continue to hear music (again, optional for the healers) and that this would be followed by a 2nd beep.. The beep would indicate the start of the protection/intention phase of the study, which would last for 5 min. A 3rd beep would signal to the P that he or she should stop the intention. Prior to the 1st beep, non-healers were given a sheet containing written suggestions for visualization strategies that could be used to protect blood from the process of hemolysis. They were informed that they could also use their own visualization method or a combination of different methods. After Ps had read the sheet, they were asked to inform E1 which method they thought they might employ. To help them to focus on the task, P’s were asked to press the down arrow key on the computer keyboard to see an example of a healthy blood cell during the protection stage. Healers were told that they should employ the methods that they usually use when undertaking their healing work as much as possible within the confines of the experimental set-up. If the baseline period was first, these instructions were given at the end of the break period. At the sound of the 1st beep, E1 started the relaxation music, reset the timer to zero and left the intention room. When the intention period was over, the intercom was switched off. Non-healers then completed the PIQ, and there was some informal discussion of the experience of the experiment. Healers, on the other hand, were interviewed by E1 in depth about their subjective experience of healing, including an exploration of the methods that they employ in the real world and what they had just experienced in the laboratory situation. The interviews were tape-recorded for later transcription.

Baseline Period. The intercom was switched off during the entire baseline period and turned on again during the break if the test period was to follow. (This assured that neither she nor P would hear the beeps being sent from E2, who needed to follow the same beep protocol for each run in order to remain blind as to which run was the test run.) During the baseline period the non-healers completed the BQ and STS. Healers in their 1st session either began or continued with the in-depth interview during the baseline period, whereas in their 2nd session they completed the BQ and STS during this period.

Conclusion. The whole experiment lasted 75 minutes including the orientation period and break time. At the end of the experiment, Ps were thanked for their time and given a debrief sheet about the nature of the study and asked to tick boxes if they wanted feedback after the experiment was completed.

RESULTS

Aborted Sessions and Replacements Several sessions early in the experiment had to be aborted because of procedural errors. Four non-healers and 1 healer were replaced because of errors in measuring the hemolysis during experimental trials that made the scores meaningless, and 3 other healers completed an extra session for these reasons. One healer was replaced because the dial on the pipette controlling the amount of blood to be added to the saline was inadvertently altered, and 1 healer completed 3 sessions because they failed to follow the instructions for the intention period in 1 session. Decisions as to whether to replace a healer entirely or to have them complete an additional session were based on estimates at the time of the availability of suitable replacements. All these decisions were made by Palmer prior to his awareness of condition assignments and computation of hemolysis scores.

Computation of Hemolysis Scores The s-meter output consisted of 63 absorption values for each trial of the run. Following Braud (1990), the initial change scores were computed by subtracting the mean of the last 5 values in the 1-min trial period from the mean of the 1st 5 values. The beginning of the 1-min trial was defined as the highest value in the trial. Ordinarily this was the 1st value in the record, but on 16 trials (out of 1280) it was on a later measurement, due to a delay in getting the door of the GMF device closed. Four of the 16 were experimental trials. This problem had been anticipated, which is why 3 extra secs were recorded for each trial. In these cases of delay, the additional values from the end were simply added as needed to create a 60- sec trial. (In two runs, the highest value occurred on measurement 5, requiring one more value at the end than was available. Because there was little change in absorption values at the end of the run and the discrepancy was only 1 value, it was decided in these cases to retain the data and declare a trial of 59 sec.) The initial change scores were transformed into t-scores representing the experimental manipulation by subtracting from the mean of the 2 concentration (experimental) trials the mean of the 6 control trials and dividing the difference by the unbiased standard deviation of all 8 scores, assuming in each case unequal variances for the ts. (In the cases where 1 control trial was missing, this value was 7.) These "t-tests" are unconventional, but that is not a problem because the resulting values are simply scores and are not evaluated for statistical significance. They are labelled CH-scores. To correct for the expected confounding of the CH-scores by the initial values of the trials, corresponding t-scores were computed on these initial values, using the same formula as that described in the preceding paragraph. These t-scores are labelled I-scores. After adjustment of outliers in both the CH-score and I-score distributions (see below), a regression analysis of the CH-scores with the I-scores as predictor produced a correlation of .720, which confirmed our suspicion that the amount of increase in absorption across the run depended heavily on the starting value. The residuals from the regression reflect the increase of absorption that was not associated with the initial values. The grand mean of the CH distribution was added to each of these residuals, the results of which then became the revised change scores, labelled R-scores. Outliers in the R distribution were adjusted as well. Even after adjustment of outliers, the R-score distribution was skewed, z = -1.66, p = .096. Inspection of the histogram revealed that the skewness reflected the shape of the distribution overall, rather than the result of the (already adjusted) scores on the extreme tails. It was decided to perform a log transform on the distribution to remove the skewness. Natural logs were used, and by trial-and-error it was determined that minimum skewness could be achieved by adding a constant to each R-score. This adjustment brought the minimum value to 10 for the R-scores, resulting in a new skewness value with z = -.089. Only at this stage did J.P., who performed the above analyses and decided how to do them, break the code to determine which of the 2 runs in the session was test and which baseline. The 160 run scores were then transposed into two columns representing the scores for the test and baseline runs respectively. These scores were labelled simply E-scores. To make the E-scores easier to understand, a constant was subtracted from all of them to bring the grand mean to 0. The constant was the log score that corresponded to 0 in the distribution of the original residuals. The resulting scores were then multiplied by 100. Negative scores indicate a decrease in the rate of hemolysis and positive scores an increase in the rate of hemolysis. These are the scores that were entered into the ANOVAs to test the experimental hypotheses and explore related post-hoc effects. The E-scores correlated r(78) = +.686 with

comparable scores derived from the CH-scores. These "comparable scores" conformed to the scores used by Braud (1990).

Adjustments for outliers. Palmer considers it improper to remove outliers from a distribution merely because they are outliers, as this misrepresents the true nature of the data (Schmeidler & Edge, 1999, p.341). However, it is appropriate to adjust these scores so that parametric statistical tests can be applied. Such adjustments were made in the present data set for the CH-scores, the I-scores and the E-scores. All these analyses were done by, and decisions on how to perform them were made by, J.P. while he was blind to which runs were test and which were baseline. Examination of the histograms revealed that in most cases there were obvious outliers that frequently skewed the distributions significantly. Based on these examinations, it seemed reasonable to define an outlier as any score greater than .5 SD units from the next nearest score to the mean that was not an outlier. This latter score can be called the x-score. In order to roughly maintain the positions of the outliers relative to each other, it was decided to accomplish the adjustments by moving each outlier half the distance to the x- score. After each outlier had been so transformed, the SD of the resulting distribution was assessed and, if necessary, additional adjustments made, using the same procedure as before, until all scores were within .5 SD of the next innermost score. Of the 1760 run scores produced at various stages of the transformations (160 runs x 11 score types), only 16 had to be adjusted, .9%. Of the 160 runs, 13 required some sort of adjustment, 8.1%.

Adjustments of Other Scores

Global Geomagnetism (Ap) Scores. The Ap index is ordinarily log transformed and was presented that way by Braud and Dennis (1989), so we chose to apply the same transformations to the Ap scores as we did to the hemolysis scores. We decided to examine the Ap scores for the day preceding the day of test of testing, as this is where Braud and Dennis found their effect, but we also analysed the day of the test, as this is the day one would expect on logical grounds to be the most relevant. These measures will be referred to as Ap(- 1) and Ap(0) respectively. A histogram of the Ap(-1) scores revealed the expected strong positive skew, but in addition there were 3 extreme values on the right tail. These scores must have corresponded to some cosmic disturbance, such as a solar storm. Their values were 119, 154, and 161, whereas the highest score in the main distribution was 50. Employing the same strategy described above for the hemolysis scores markedly reduced the AP(-1) scores to the low-mid 50s, while retaining their place at the right tail of the distribution. Reducing all the Ap(-1) scores by 1 and applying a natural log transformed created a distribution with a trivial positive skew, z = .773. The Ap(0) scale was a similar story. Again there were 3 extreme outliers with values of 140, 161, and 181, the next highest score being 52. Following adjustment of outliers by the same method as above and application of a natural log transform to the revised Ap(0) scores minus 1, a final distribution with a trivial positive skew, z = .862, was obtained.

Personality Test Scores (BQ and STS). Finally, we decided to apply the adjustment to the 2 remaining major predictor variables in the experiment, the Boundary Questionnaire and the Spiritual Transcendence Scale, the same treatment as described above. The BQ was negatively skewed with 3 outliers in the direction opposite the skew. When these outliers were adjusted inward the skewness was actually increased to an

uncomfortably high level, z = -1.76. A natural log transform on the resulting BQ distribution reduced the skewness to z = -1.13, which we considered acceptable. The initial STS distribution had a significant negative skew, z = -2.39, but there were no outliers. A double natural log transform (taking the logs of the 1st set of logs) resulted in a distribution with the skew acceptably reduced, z = -1.40.

Description of Analyses The primary (planned) analyses were as follows: (1) An ANOVA for the combined 80 sessions was a 2 x 2 x 2 x 2 mixed design. The within-session factor was hemolysis and consisted of 2 levels, test and baseline (runs). The 1st between-session factor was group, with 2 levels (healer and non-healer); the 2nd was run order, with 2 levels, test-first (T-B) and baseline-test (B-T), and the 3rd was GMF manipulation, with 2 levels (on and off). (2) A regression analysis for the 1st sessions of all Ps, with group and run order as the between-P predictors. The dependent variable was hemolysis, defined this time as the difference between the E-scores for the test and baseline runs (test - baseline). These scores will hereafter be referred to as hemo-D scores. (3) An ANOVA for the healers was also a 2 x 2 x 2 design. GMF was the first within-P factor. The 2nd within-P factor was hemolysis. The between-P factor was run order. (4) An ANOVA for the non-healers was again a 2 x 2 x 2 mixed design, with hemolysis as the within-P factor, and run order and GMF as between-P factors. For correlational analyses, the hemo-D scores were used as the criterion variable. Predictors were sex, age, total scores for the Boundary Questionnaire (BQ) and the Spiritual- Transcendence Scale (STS), scores for each of the 7 quantifiable items on the Post-Intention Questionnaire (PIQ), and the 2 global GMF indices [Ap(-1) and Ap(0)]. Correlations were computed separately for the total sample as well as for each group, run order, and manipulated GMF condition. Secondary (unplanned) analyses were performed to follow-up and clarify significant effects from the planned analyses. Statistically significant results were defined as p < .05, two-tailed, except for the relationship between hemolysis scores from the E trials and Ap(-1), where it was p < .05, one- tailed. This is because this analysis involves a replication attempt of the significant finding reported by Braud and Dennis (1989). In keeping with their methodology, for these analyses the criterion variable was the Ap scores, and the predictor variable was hemolysis "hitters" vs "missers", defined as whether the hemo-D scores were negative or positive. All analyses were performed with SPSS 12 software.

Tests of Hypotheses

H1. H1 was tested by examining the hemolysis main effect in the analysis of all 80 sessions. The mean test score was +.837 and the mean baseline score was -.353, yielding a hemo-D score of +1.190, in the direction of hemolysis acceleration, opposite the prediction of hemolysis retardation. However, the main effect was not significant, F(1/72) = 1.07, p = .305. Thus, H1 is not supported.

H2. It was decided that the fairest way to test H2 was to compare the 1st sessions of the 20 healers and the 40 non-healers. The healers had a test mean of -.552 and a baseline mean of -1.777, yielding a hemo-D score of +1.225. The non-healers had a test mean of +2.010 and a baseline mean of -.068, yielding a hemo-D score of +2.078. Thus, the healers accelerated the hemolysis slightly less than the non-healers, in support of the prediction.

However, the healer effect was not significant in the regression analysis, t(56) = 0.95, p = .348. Thus H2 is not supported. We also decided to look at the performance of the healers separately, which allowed us to include their 2nd sessions. Healers in their 2nd sessions had a mean test score of -.121 compared to a baseline mean of .500, resulting in a hemo-D score of -.742, in the direction if hemolysis retardation. However, the hemolysis main effect was not significant, F(1/18) = 0.50, p = .826. This reversal of the session 1 result reduced their mean hemo-D score across both sessions to +.483, compared to +2.078 for the single session of the non-healers. The decline in the hemolysis retardation between sessions was not significant, as reflected by the hemolysis x session main effect, F(1/18) = .347, p = .563.

H3. The GMF hypothesis was tested separately for healers and non-healers, because it was manipulated between sessions in the former case and within sessions in the latter. For the non-healers, when GMF was off the mean for the test run was +.930 compared to -3.479 for the baseline run, leaving a hemo-D score of +4.409. When GMF was on, the mean in the test run was +3.090 compared to +3.342 for the baseline run, leaving a hemo-D score of -.025. The difference between the 2 hemo-D scores is +4.434, indicating greater acceleration of hemolysis in the GMF off condition, contrary to the prediction. However, this difference, as represented by the GMF x hemolysis interaction in the ANOVA, was not significant, F(1/36) = 1.72, p = .199. Thus H3 is not supported for non-healers. For the healers, when GMF was off the mean for the test run was -.522 compared to -1.777 for the baseline run, leaving a mean hemo-D score of +1.255. When GMF was on, the mean in the test run was -.121 compared to +.500 for the baseline run, leaving a mean hemo- D score of -.621. The difference between the 2 hemo-D means is +1.876, again indicating greater acceleration of hemolysis in the GMF off condition. However, this difference, as represented by the GMF x hemolysis interaction in the ANOVA, is not significant, F(1/18) = 0.35, p = .583. Thus H3 is not supported for healers. The significance of the 2 groups combined can be calculated using a Stouffer Z test, which was performed by converting to zs the one-tailed equivalents of the p-values reported above for each group, adding them together, and dividing by the square root of 2.The Stouffer Z resulting from this process is -0.39, p = .548, two-tailed. Thus H3 is not supported for the combined groups.

H4. Following Braud and Dennis (1989), H4 was tested by dividing the hemo-D scores for each session at 0, referring to the group with negative scores as hitters and those with positive scores as missers. The dependent variable was then the log-transformed Ap scores for the day preceding the test session [Ap(-1)]. The mean Ap(-1) score for the hitters was .906 compared to .782 for the missers. This result, although in the predicted direction based on Braud and Dennis (1989), was not quite significant, t(78) = 1.42, p = .081, one- tailed.

H5. This hypothesis was tested by computing correlations between the adjusted BQ or STS scores and the hemo-D scores for the combined groups. For the BQ, the result was r(58) = -.071, p = .590. For the STS, the result was r(58) = -.067, p = .612. Thus H5 is not supported.

H6. H6 was tested by computing the difference between the means of the BQ and STS for the 1st sessions of the healers and non-healers. The overall raw-score mean of the BQ was 88.16 (SD = 17.86). The mean of the 20 healers was 84.38 (SD = 15.94) compared to

90.05 (SD = 18.65) for the 40 non-healers. This result indicates thinner boundaries for the non-healers, opposite the prediction. However, the difference, based on the adjusted BQ scores, is not significant, t(58) = 1.37, p = .177. The overall raw-score mean for the STS was 94.97 (SD = 11.00). The mean of the healers was 105.15 (SD = 4.02) compared to 89.88 (SD = 9.77) for the non-healers, t(57.8) = 8.87, p <<. 001, using the adjusted STS scores. Thus H6 is strongly supported for the STS. The variance on STS was also significantly lower for the healers than for the non-healers, F = 7.16, p = .010, by Levene's Test. For the overall sample, the BQ and STS were uncorrelated, r(58) = .050, p = .705. However, they were correlated positively to a significant degree for the healer group, r(18) = .345, p = .029, with a nonsignificant reversal for the non-healers, r(38) = -.204, p = .387. The difference between these correlations approaches significance, z = 1.935, p = .053.

H7. H7 was tested by correlating the BQ or STS adjusted scores with the responses of non-healers to the question on the PIQ asking them to estimate their degree of success in retarding the hemolysis. For BQ, the result was in the predicted direction but not significant, r(38) = .200, p = .228. Thus H7 is not supported for the BQ. For the STS, the result was highly significant in the predicted direction, r(38) = .597, p << .001 Thus H7 is strongly supported for the STS.

Other Analyses In this section we describe the significant results obtained from analyses of hemolysis scores that were not tests of the experimental hypotheses. Readers can assume that variables mentioned under "Description of Analyses" above and not mentioned in this section were not significantly related to hemolysis scores.

Age. The only psychological or demographic variable to correlate significantly with the hemo-D scores for the 1st sessions was age. Age data were not available for 1 healer and 1 non-healer. The mean age of the remaining 58 Ps was 44.14 (SD = 12.58). The mean of hemo-D for the 1st sessions was +1.794, t(59) = 1.24, p = .220. It was found that older Ps were more prone to accelerate the hemolysis than younger Ps, r(56) = .294, p = .025. The healers in the sample had a mean age of 49.84 compared to 41.36 for the non- healers, t(53.5) = 2.98, p = .004. However, the ages of the non-healers were significantly more variable than those of the healers (SDs 13.51 vs. 8.05), F = 11.32, p = .001 by Levene's test. The correlation between age and hemo-D differed markedly for the two groups. Non-healers conformed to the pattern for the entire sample, r(37) = .461, p =.003. However, the correlation reversed for healers, r(17) = -.276, p = .253. The two correlations are significantly different, z = 2.60, p = .010. So that the results of the 2 groups are balanced with respect to GMF, it is desirable to also compare the correlation for the non-healers to the correlation for healers that is based on hemo-Dsum, which combines the 2 GMF conditions for healers. This correlation, r(17) = -.096, p = .695, still differs significantly from the correlation for the non-healers, z = 1.99, p = .046. Thus, the proper conclusion is that the positive correlation between age and hemolysis acceleration applies only to non-healers. The question arises as to how exactly to interpret this correlation for the non-healers. The fact that the overall mean was moderately positive suggests that it should be interpreted uni-directionally, viz., acceleration of hemolysis by the older Ps and no effect by the younger Ps. Dividing the sample at the median of 43 years seems to bear this conclusion out. Among the older Ps, there was a strong acceleration effect, M = +5.506, t(20) = 2.46, p = .015.

Among the younger Ps, the result was near chance, M = -1.026, t(19) = 0.33, p = .745. However, the difference between the groups did not reach significance, t(37) = 1.72, p = .093. An examination of the histogram for the ages of the non-healers revealed bimodality, with 2 normally distributed subgroups. A younger group (N = 12) with a range of 15-30 years, and an older group (N = 27) with a range of 36-63 years. A scatterplot of the age/hemo-D relationship also revealed a slight non-linearity. When the non-healer group was split at 33 instead of 43, the results were markedly different. The means of the older and younger groups were now virtually identical in magnitude (5.861 vs. –5.635). The mean for the older group remained significant, t(26) = 2.77, p = .010, but the mean was not quite significant for the younger group because of the smaller sample size, t(11) = 1.75, p = .108. However, the difference between the 2 groups is now comfortably significant, t(37) = 3.00, p = .005. Moreover, the correlations between age and hemo-D within each of these groups are quite low (.157, .193) compared to the much higher correlations with the median split (.516, .394). The point of all this is to show that the split based on the shape of the age distribution gives a better representation of the significant effect represented by the original correlation than does the median split. The consequence is to favor a bi-directional interpretation as opposed to a uni-directional one. In other words, non-healers over 35 tended to accelerate the hemolysis and non-healers under 31 tended to retard it. The age/hemo-D correlations for non-healers and healers combined were similar for the two run orders. For Order A it was r(26) = .422, p = .025, while for Order B it was r(28) = .243, p = .196. On the other hand, GMF appeared to affect the relationship. For GMF off, the correlation was virtually nonexistent, r(36) = .030, p = .856, whereas for GMF on it was quite high, r(18) = .670, p = .001. These correlations differ significantly from each other, z = 2.64, p = .008. This result could be misleading, however, because all of the GMF-on scores in this analysis were from non-healers, for whom a strong positive correlation had been demonstrated overall. Insight into this situation can be provided by examining the age/hemolysis correlations for healers and non-healers separately. For non-healers, age/hemo- D correlations were r(18) = .670, p = .001 with GMF off and r(17) = .269, p = .266 with GMF on. These correlations do not differ significantly from each other, z = 1.54, p = .124. For healers, the correlation with GMF off is r(17) = -.276, p = .253, and r(17) = .167, p = .493 with GMF on. In contrast to the non-healers, the correlation was more positive with GMF on than with GMF off, z = 1.30, p = .194. As the results for the 2 groups cancel each other out, it is best to interpret the apparent mediation of the age/hemo-D correlation by GMF as an artifact of its mediation by healer status. Finally, we asked if the 6 non-healers who had some background in healing would score more like the healers or the other non-healers with respect to age. The Pearson correlations between age and hemo-D for these 6 Ps was +.781 compared to +.448 for the other 33 non-healers. Clearly, the non-healers with some healing background followed the pattern of the other non-healers.

GMF and Run Order. In the original ANOVA for non-healers, a significant main effect for GMF, F(1/36) = 6.44, p = .016, is superseded by a significant GMF x run-score interaction, F(1/36) = 4.58, p = .039. Note that these hemolysis scores are based on the combined results of the test and baseline runs. They will hereafter be referred to as hemo-T scores, with T representing total. Despite the fact that these scores do not incorporate the run- order control, the interaction is a psi effect because E2 was blind to the run order. An examination of the overall hemo-T distribution revealed that it was significantly skewed, z = 2.58, p = .010. An examination of the histogram revealed that this skewness was

due to one outlier on the positive tail. When this score was adjusted by the method described in the previous section, the skewness was no longer significant, z = -.058. The mean of hemo- T for all 80 sessions is -0.118, t(79) = 0.18, p = .858. As illustrated in Figure 3, the interaction is such that the effect of GMF on the hemo- T scores is restricted to the run order in which the test run was given first. This run order will hereafter be referred to as Order A, and the opposite as Order B. For Order A, the mean hemo-T score was -3.292 with GMF off, t(9) = -2.53, p = .032. With GMF on, the mean hemo-T score was +3.899, t(9) = 2.53, p = .032. The difference between these two means is highly significant, t(18) = 3.56, p = .002. For Order B, the mean for GMF off was +.744 and for GMF on +1.447. Neither mean differed significantly from 0, nor did they differ significantly from each other.

Figure 3. Hemo-T scores as a function of GMF and run order for non-healers.

For healers, the corresponding GMF x run-order interaction was not significant, F(1/18) = .821, p = .377. However, it was in the same direction as for the non-healers. For Order A, the mean hemo-T score was -2.412 for GMF off and +.178 for GMF on. Neither mean differed significantly from 0, nor was the difference between the means significant, t(9) = 1.41, p = .192. For Order B, the mean was +0.083 for GMF off and +0.201 for GMF on. As with the non-healers, the biggest difference was for Order A (test run first).Using the Stouffer Z method, the result of the GMF x run-order interaction for the two groups combined is associated with Z = 2.08, p = .038, two-tailed. The hemo-D scores for the healers corresponded more closely to the scores for the non-healers with GMF off as compared to GMF on. The weak result for GMF on could be due to the fact that all these runs occurred in the Ps' 2nd sessions. Given the prevalence of decline effects in psi data, it is possible that there was simply not enough psi present in the 2nd sessions to support a relationship with an independent variable. It was decided at this point to see if any of the psychological or demographic variables might discriminate the hemo-T scores in the 2 GMF conditions for Order A of the 1st

sessions. The one variable that was found to do so was the BQ, which correlated significantly with the hemo-T scores in the GMF off condition, r(18) = -.451, p = .046. The direction of the effect indicates that Ps with thin boundaries contributed most to the retardation of hemolysis occurring in this condition. With GMF on, the correlation was close to chance: r(8) = .127, p = .727. The two correlations do not differ significantly from each other, z = 1.37, p = .170 Finally, a check was made to see if the hemo-T scores could predict the Ap index. Sessions with hemo-T scores less than 0 were classified as hitters and those with hemo-T scores greater than 0 as missers. The mean of Ap(-1) -- day before testing -- for hitters was .881 and for missers .891, t(38) = 0.08, p = .934. For GMF off sessions, the mean Ap(-1) was.877 for hitters and .836 for missers, t(18) = 0.27, p = .793. For GMF on, the mean Ap(-1) was .888 for hitters and .920 for missers, t(18) = 0.16, p = .873. The results were no more exciting for the day of testing. The mean Ap(0) was .812 for hitters and .848 for missers, t(38) = 0.23, p = .816. For GMF off sessions, the mean Ap(0) was.822 for hitters and .905 for missers, t(18) = 0.43, p = .675. For GMF on, the mean Ap(0) was .812 for hitters and .818 for missers, t(18) = .003, p = .976. Clearly, the Ap index has nothing to do with the results for hemo-T.

Variance. As discussed in the introduction, there are problems with the hemolysis variance scores reported by Braud (1990). However, it remains possible that these scores reflected a psi effect, and thus it would be of interest to learn of there were any variance effects in the present experiment. The one legitimate way we could think of to test for variance effects in our data was to compare the variance of scores around mean chance expectation in the test and baseline runs. For all 80 sessions, this ratio was not significant, F(78/78) = 1.36, p = .354, and the higher variance was in the baseline run. The subsample in our study most comparable to Braud's sample consisted of non-healers for whom the GMF around the blood specimen was minimized. Although the variance this time was higher for the test run, the ratio was still not significant, F(18/18) = 1.63, p = .618. Thus there is no evidence of a variance effect in our data.

DISCUSSION

None of the first 4 hypotheses were confirmed by the data. The overall hemolysis scores were slightly in the direction of hemolysis acceleration (Hypothesis 1). The acceleration was slightly less in the healer sample than in the non-healer sample, and this was true regardless of whether both sessions or just the 1st session of the healers were included in the analysis (Hypothesis 2). However, the differences were not significant. As for Hypothesis 4, the acceleration was greater with the GMF around the specimen on as opposed to off, but again the difference was not significant. The closest we came to confirming a parapsychological hypothesis concerned Hypothesis 3, the attempted replication of Braud and Dennis's (1989) finding that global GMF as measured by the Ap index was greater in hitting sessions (the hemolysis score in the retardation direction) than in missing sessions (the hemolysis score in the acceleration direction). The one-tailed p-value was.081. The replication hypothesis could be tested alternatively by computing a mini-meta-analysis of the 2 experiments, which can be accomplished by converting the one-tailed p-values to zs and then computing a Stouffer Z. The result was Z = 2.40, z = .008, one-tailed. So by this criterion the replication was successful. Although this criterion is not unreasonable, we prefer the more conservative criterion of one-tailed significance for the replication attempt. In our view, the best way to

characterize the current status of the hypothesis is that the results are strongly suggestive of its validity. If the effect is real, we find it odd that it should be significant only on the day prior to the test session. From a parapsychological perspective, this result seems to imply PK in the future over a 1-day time span, for which we are aware of no independent evidence. A more likely explanation is that the GMF parameters reflected in the Ap index reliably predict the presence of some other, presumably physical, factor on the subsequent day, which factor influences psi in real time. Adams (1985) suggested solar flares as the origin of such a factor, as GMF activity is elevated the day after these flares. This implies that some consequence of solar flares that registers on earth at the time of the flare (light?) might play the causal role. Our "data snooping" yielded 2 significant post-hoc effects. The first was a positive correlation between age and the primary hemolysis difference scores, such that Ps 35 or above seemed to accelerate the hemolysis and Ps 30 or below seemed to retard it. The 2nd was a complex effect that applies to the combined hemolysis (hemo-T) scores of both the test and the baselines (in contrast to the primary hemo-D scores, which reflect the difference between the test and baseline run). The finding was restricted to 1st sessions of Ps who received the test run first. Hemolysis accelerated if the GMF around the specimen was on and it slowed if the GMF was off. Anytime there is data snooping, particularly the wide-ranging data snooping in which we engaged, there is always concern for type-1 errors due to the multiplicity of analyses. We did not employ Bonferroni corrections, because our analyses were so statistically interdependent that the correction would be illegitimately conservative to the point of being meaningless. In our view, the best approach is to require that in order to be considered conclusive any new finding, whether predicted or not, must be significantly confirmed in a replication attempt, using the unadjusted alpha criteria. In the case of multiple replication attempts, the conclusion can be based on meta-analysis. (There are some complications that must be considered in applying meta-analyses, but a discussion of these is beyond the scope of this report.). For those who would like to draw a conclusion from the original finding, we recommend an approach derived from the replication criterion itself. Specifically, the criterion alpha level would be the product of the alpha criteria for the initial experiment and the *replication respectively, using the two-tailed value for the former and the one-tailed value for the latter. By this criterion, the product would be, in two-tailed terms, .05 x .10 = .005. Thus any finding of p < .005 (unadjusted) would be considered established. Both the correlation as described above between age and hemolysis (p = .003) and the interaction between GMF and hemolysis, again as discussed above but restricted to the non-healer sample (p = .002)2, meet this criterion. However, our preference is still to consider both findings suggestive pending replication. Age is rarely reported in the literature as a correlate of psi. In a review of studies up to the mid 1970s, Palmer (1978) found some evidence that children score better on ESP tests than adults, but no evidence of differences among adults of different ages. Winkelman (1981) found that PK scores increased with age among a group of children up to 15 yeas old. Moving up the age scale, Novomeysky (1984) found that the capacity for color telepathy was better in people aged between 14 and 25 but decreased with age. Finally, in a strictly adult sample, there was a significant positive correlation between age and ESP scores in the ganzfeld

2 This analysis would also meet the criterion if it were applied to the results of healers and non-healers combined t(38) = 3.42, p = .001. However, this analysis was not considered quite appropriate because it involved two scores from each healer and only one score from each non-healer. In Results, the effect for healers and non- healers combined was determined by a Stouffer Z , the p-value of which, .042, would not meet the criterion.

(Morris, Dalton, Delanoy, & Watt, 1993) although this effect was for only one of the experimenters (Delanoy), and is reported as an experimenter effect. It is conceivable that the age-hemolysis effect might be associated with menopausal changes in females, which if true would lead one to expect that women were more responsible for the effect than men. The mean difference of hemo-D scores as a function of age was in fact twice as great for females as for males (.139 vs. .060) and was independently significant for females, t(25) = 3.08, p = .005. However, an ANOVA of the hemo-D scores with sex and the age category as predictors yielded no significant interaction between age and sex, F(1/35) = 0.916, p = .345. Also, such a physiological hypothesis would not account for the reversal of the age-hemolysis relationship among the healers. A psychological explanation is potentially more promising. Of the other psychological variables examined in the experiment, only 1 correlated significantly with age. On the PTQ, older non-healers (M = 56.44) were significantly more likely than younger non-healers (M = 31.37) to view the source of any healing on the 100-point rating scale as being inside themselves than an outside agency, t(35) = 2.96, p =. 006. However, this source variable bore no relationship whatsoever to the hemo-D scores, r(36) = -.015, so it clearly did not mediate the age-hemolysis effect. E1 (Simmonds), a young female, reported feeling more at ease with Ps closer to her own age, which could have differentially affected the comfort or motivation of the older and younger Ps. But why would the same not apply to non-healers? We can think of no other plausible explanations for the age-hemolysis effect, a circumstance which increases the appeal of the type-1 error hypothesis. Almost as puzzling is the differential effect of the GMF manipulation on the combined test and baseline hemolysis scores when the test run was 1st. The fact that the baseline run contributed to the effect is not that surprising, because other evidence exists in the literature that PK can occur in the absence of awareness of the target system (e.g., Berger, 1988; Stanford, Zenhausern, Taylor, & Dwyer, 1976.) The fact that the effect was restricted to the test-first condition is consistent with E1s observation, solicited by Palmer before she was told of this outcome, that Ps' motivation for the task tended to be dampened when the test run was delayed. On the other hand, E1 also noted that this tendency was more noticeable for the healers, whereas the effect was more prevalent in the non-healers. It may be that the interview was a more convincing and enjoyable filler than questionnaire completion in the case of the healers, which reduced this effect of trial order . The differential effect of the GMF manipulation is harder to account for. One might speculate that it is the natural tendency of the GMF to accelerate processes such as hemolysis- - that is at least what the AC component of the field seems to do in the case of the EEG (Persinger, 1989) – and when the Ps recruited this energy, it simply followed its natural tendency, despite the contrary intention of the Ps. In the absence of such energy, the Ps either recruited some other energy source without these accelerating tendencies or no energy source at all to retard the hemolysis. In any event, whereas the AC component of the GMF around the person seemed to facilitate hemolysis retardation in our experiment, the DC component of he field around the specimen seemed to thwart it. This finding contradicts Braud's (1990) results; one must assume that a standard GMF field was present around the specimens in his experiment, yet hemolysis retardation was more prevalent than acceleration in his data. However, even if this effect was psi-mediated, it could have been due to psi-mediated variation in the start times of the measurements rather than a physical effect on the specimens.

Hemolysis and Personality In this section we discuss the results from Hypotheses 5 – 7. In contrast to Hypothesis 5, there was no relationship between high scoring on BQ or STS and success at the hemolysis task for healers or non-healers. In a post-hoc analysis, however, the BQ was found to contribute to retarding the rate of hemolysis where the GMF condition was low, and where Ps received the half-run first. This corresponds to a situation where motivation was high, the cognitive style of the person intending was thin boundaried, and there was no potential geomagnetic influence on the blood. Although the GMF would not directly affect the physiology of the intender, it is possible that this could be an environment in which thinner boundaried people function in an optimal psychic manner. Hypothesis 6 received partial support in this study. In contrast to expectation, healers did not score significantly higher than non healers on the BQ. In fact, the trend was actually found to be in the opposite direction, with non healers scoring non significantly “thinner” than healers. In line with Hypothesis 6, it was demonstrated that healers scored higher than non-healers on the STS scale. This is an interesting finding, as many non-healer participants also seemed to adhere to a spiritual world view (as reflected in the negative skew in the data). One of the reasons for the inclusion of spiritual transcendence as a variable in this study was the observation that healers are often considered to be spiritual individuals, even defining themselves as such (offering “spiritual” healing in their advertisements). Although there has not yet been any formal analysis of the interviews with the healers, many spoke of their personal spiritual development and had a focus on healing “with love” and for the benefit of humanity and “the greater good”. Healing seemed to have become part of their entire life. It is of interest to note that spiritual transcendence has been related positively to mental health measures. for example, connectedness is related to lower levels of stress, while Prayer fulfillment is related to adaptive coping styles (Cooper, 2003) As such, healers, might be considered to be a well adjusted group of people, and the subjective experience of undertaking healing and feeling successful may be related to good mental health. Where scoring on the two personality variables was correlated, some interesting observations were noted. Surprisingly, overall, the variables were unrelated, despite their supeficial similaries. The finding that the variables were related positively for healers, but negatively but non-significantly for non-healers might indicate differences in uses of boundaries, or the types of boundaries predominant in both cases. Preliminary observations from the interviews indicate that healers often have thin boundaries but these have been developed and are controlled and flexible. Boundaries may be selectively thinned for employment in the context of their healing and spiritual pursuits. Thinner boundaries in the non-healer group may simply reflect a perceptual-cognitive style, independent from their spiritual pursuits. In contrast to Hypothesis 7, the BQ did not correlate with subjective success in retarding the rate of hemolysis. This was somewhat surprising, given that Richards (1996) had previously found a correlation between scoring thinner on the BQ and subjective success at an ESP task. This may reflect the subjective differences between feeling that one is being psychic to feeling that one is involved in affecting a physical process. Although it is often suggested that ESP and PK are part of the same process (e.g., Rhine, 1947; Roe, Davey, & Stevens, 2003) they are different subjective experiences and their correlates may also be very different. To date, the literature on personality correlates of both PK and psychic healing remains rather sparse.

If one looks at Richards’ (1996) work, there were in fact only four boundaries which related to perceived success at the interpersonal psi task. These relate to the experiential and emotional types of boundary and were, sleep, unusual experiences, thoughts and feelings and sensitivity. The use of the shorter form of the BQ may have affected the outcomes in this study, as it may not have maximized the experiential and emotional aspects of boundary thinness. As such, future work addressing biological PK/healing should perhaps employ these same four subscales from the original longer form of the BQ (Hartmann, 1991). In support of our prediction, the STS was related to subjective success in retarding the rate of hemolysis in the non healer group. The task was presented to most participants as a “healing” task rather than a PK task, although some participants were recruited employing fliers describing the study as addressing “mind over matter”. The nature of the task was to protect the blood cells, which many people approached as a meditative practice or spoke of sending love to the cells. As such, the subjective feeling of success at the task might be increased if such strategies seemed to be actually working. The STS contains several items addressing meditation and connection, which gives us some insight to the correlation. Spiritual transcendence may form part of a general worldview that also allows for the influence of a physical system with mental intention. In support of this hypothesis, paranormal beliefs are often associated with spiritual beliefs (e.g., Kennedy & Kanthamani, 1995) and both mystical experiences and paranormal beliefs loaded onto the same factor) in Thalbourne’s original factor analysis identifying the Transliminality construct (Thalbourne & Delin, 1994).

Experimenter Psi In any psi study one must look at the possibility of psi results being attributable to experimenter psi (E-psi). Identifying the source of psi is an extremely difficult undertaking that cannot be done with complete certainty (Palmer, 1997). Palmer (1997) also noted that Principal Investigators should be treated as Es for purposes of evaluating E-psi effects, whether or not they actually tested the Ps. This means that there are 5 individuals who need to be considered as potential psi sources in the present experiment: E1 (Simmonds), E2 (Hlavaty/Drucker), PI1(Palmer) and PI2 (Baumann). The best sources of clues to the psi source are (a) the relative motivation and presumed capacities of the candidates to produce the effect in question, (b) the non-psi-based knowledge of the values of the independent variables for each session, and (c) the existence of plausible correlations between psi scores and individual-difference measures given to the Ps but to which the Es/PIs were blind. With respect to motivation, E1, PI1 and PI2 were aware of Braud and Dennis's (1989) finding about the relation between the Ap index of GMF and hemolysis and presumably would like to see it replicated. None of the Es/PIs were motivated to achieve any of the post- hoc effects uncovered, except in the general sense of wanting to see some psi in the data. On the other hand, the only findings the Ps were presumably motivated to achieve were those involving hemolysis reduction (as opposed to acceleration). As for explicit knowledge, none of the Es/PIs knew the Ap values for any of the test days. E1 and E2 both met the Ps before the session and thus could infer their approximate ages before the data were collected. All the Es/PIs knew that there were test and baseline runs, but only E1 knew at the time of testing which was which. None of the Es/PIs knew whether a given session was to be GMF high or GMF low. One relevant individual-differences correlation was discovered, namely the tendency of Ps with thin boundaries to be more successful than those with thick boundaries in retarding hemolysis with the manipulated GMF low (if the test run was 1st). This result is particularly noteworthy, because Hypothesis 5 predicted that thin-boundary Ps on the BQ would show

more hemolysis retardation than thick boundary Ps. Although this hypothesis was not confirmed in the strict sense, it makes some sense to view the hypothesis more broadly as predicting thin boundary Ps to be more successful wherever in the experiment significant psi happened to manifest, regardless of whether that is where it was expected to manifest beforehand. In this more general sense, one could say that Hypothesis 5 received partial confirmation as a result of this finding. In any event, the presence of this correlation makes it more likely that the Ps rather than the Es/PIs were responsible for the GMF-hemolysis interaction. A few words about the apparent abilities of the Es/PIs are also in order. Both E2s feel that they have psi talents and in fact both expressed some concern about possibly influencing the results with their own psi. None of the other Es/PIs claim unusual psi abilities or have demonstrated such, nor are they among the handful of parapsychologists generally considered to be exceptionally psi-conducive experimenters. Finally, the odds of E-psi being responsible for the results involving the hemo-D scores would be modestly enhanced if the significant results with these scores were shown to be due to the baseline run as opposed to the test run. Ps would presumably be focussing their attention on the test run, whereas Es might find it easier to allocate at least some of their psi to the baseline run. Such an effect on baseline scores was recently found in a pre-stimulus response experiment using GSR (May, 2004). The authors attributed the results of this study to an experimenter effect mediated by DAT. In our data, there were three effects involving hemo scores worthy of discussion: (a) the relation of hemo-D scores to the Ap index, with hemo-D as the predictor, and (b) the relation of hemo-D scores to age, restricted to non- healers, and (c) the relation of hemo-T scores to manipulated GMF in Session 1 when the test run was first. For all sessions, as reported previously, the mean hemo-D score was +.837 for the test run compared to -.353 for the baseline run. Although the deviation is slightly greater in magnitude for the test run, both values are very close to chance. The age correlation with hemo-D is somewhat more clearly driven by the test run. For the test run, the mean difference on hemo-D between younger and older Ps is 7.692 and significant, t(36) = 2.62, p = .013. The corresponding mean difference for the baseline run is only half as large, 3.804, and not significant, t(36) = 1.34, p = .188. Neither the results for (a) nor the results for (b) provide any support for the E-psi hypothesis. For the hemo-T scores with the test run first (c), the mean difference between the two GMF conditions was 5.295 for the test run and 5.595 for the baseline run, virtually identical. But both are also significant: test: t(38) = 2.12, p = .041; baseline: t(38) = 2.67, p = .011. Although the significant effect for the baseline run in could be construed as supporting the E-psi hypothesis, in our view this finding is superseded by the mediation of the BQ, as described above, which argues more strongly for a P-based effect.

The Differential Effect One aspect of the data that stands out in this study is that the effects tended to involve a contrast between the acceleration and retardation of hemolysis, rather than retardation versus no effect. The acceleration and retardation cancelled each other out, leaving a net result close to chance. Such cancellations are actually quite common in micro-PK experiments. The two major contributors to this literature both use designs in which Ps are instructed to aim for a high score or a low score; they succeed in both tasks, but the net result of all the trials combined is close to chance (Jahn, Dunne, & Nelson, 1991; Schmidt, 1970). This cancellation has led to some vigorous debates about whether such data can be better explained by decision augmentation theory (DAT) than by PK (Dobyns & Nelson, 1998; May, Utts, & Spottiswoode, 1995a). The most apparent way for DAT to yield this kind of cancellation is for Es to use ESP to assign above-chance and below-chance sessions a-priori to the appropriate

experimental conditions (e.g., high-aim or low-aim). In our study, we attempted to avoid this problem by using counterbalancing rather than randomisation to assign our sessions to conditions. Although it is still theoretically possible for DAT to apply in such circumstances, we think the chances are remote. Using a standardized method of counterbalancing drastically reduces the degrees of freedom one has in assigning sessions to conditions compared to what one finds with random assignment, and there is no independent evidence that ESP can function in the presence of such rationalistic constraints. Thus we do not believe that our findings can be attributed to this particular application of DAT, which is the one that would produce the cancellation we find noteworthy. Cancellation in the broader sense appears in ESP data at well, where it has been named the differential effect (Rao,1965). It has even been suggested that there might be some general homeostatic principle, heretofore unrecognised by physics, that forces psi to cancel itself out in the interests of preserving some kind of equilibrium in nature (Pallikar, Boller, & Bösch, 2000). Our data provide no great additional insights into this differential effect, except to once again demonstrate it.

Implications for Healing If we assume that the effects we have uncovered are real and due to PK on the part of the Ps, what implications do they have for psychic or spiritual healing? First, they demonstrate that practicing healers produce such effects no more strongly or reliably than do ordinary volunteers of the type tested by Braud (1990). Second, the results with the GMF manipulation, which included effects in the baseline runs, suggest that the effect can be produced in the absence of specific intention or effort to produce them. It could be that many of our Ps would like to have seen an effect in the baseline run had they known about, so it still possible that such a broad desire is necessary for an effect to take place. The most sobering implication of our data derives from the evidence of hemolysis acceleration. Translated into healing terms, this means that healers could unintentionally "mis-direct" their PK to make an illness worse rather than better. Psi-missing is a well known feature of psi, so one should not be surprised to see it apply to PK that occurs in the context of healing. Finally, if the finding with the Ap index continue to hold up, it may suggest that healing should be performed on days when the global GMF is relatively high, but not outside the normal range. Finally, it should be noted that the failure of healers to show exceptional ability in this experiment says nothing about their abilities to perform the different kinds of healing tasks involved in their practices.

REFERENCES

Adams, M. H. (1985). Variability in remote viewing performance. Possible relationship to the geomagnetic field. Proceedings of Presented Papers: The Parapsychological Association 28th Annual Convention, Vol. 1, 453-461. Berger, R. E. (1988). Psi effects without real-time feedback. Journal of Parapsychology, 52, 1-27. Braud, W. G. (1990). Distant mental influence of rate of hemolysis of human red blood cells. Journal of the American Society for Psychical Research, 84, 1-24. Braud, W. G., & Dennis, S. P. (1989). Geophysical variables and behaviour: LVIII. Autonomic activity, hemolysis, and biological psychokinesis: Possible relationships with geomagnetic field activity. Perceptual and Motor Skills, 68, 1243-1254.

Cooper, E.A. (2003). An investigation of the relationships among spirituality, prayer and meditation and aspects of stress and coping. Dissertation Abstracts International: Section B: The Sciences and Engineering, 64, 1484 (accessed from PsycINFO Database Record) Dobyns, Y. H., & Nelson, R. D. (1998). Empirical evidence against decision augmentation theory. Journal of Scientific Exploration, 12, 231-257. Hartmann, E. (1991). Boundaries in the mind: A new psychology of personality. New York: Basic Books. Jahn, R. G., Dunne, B. J., & Nelson, R. D. (1991). Engineering anomalies research. Journal of Scientific Exploration, 1, 21-50. Kennedy, J.E. & Kanthamani, H. (1995). Asssociation between anomalous experiences and artistic creativity and spirituality. Journal of the American Society for Psychical Research, 89, 333-343. May, E. C., Utts, J. M., & Spottiswoode, S. J. P. (1995a). Decision augmentation theory: Applications to the random number generator database. Journal of Scientific Exploration, 9, 453-488. May, E. C., Utts, J. M., & Spottiswoode, S, J. P. (1995b). Decision augmentation theory: Toward a model of anomalous mental phenomena. Journal of Parapsychology, 59, 195-220. Morris, R. L., Dalton, K., Delanoy, D. L., & Watt, C. (1995). Comparison of the sender/ no sender condition in the ganzfeld. In Proceedings of Presented Papers of the Parapsychological Association 38th Annual Convention, pp. 244-259. Novomeysky, A. (1984). On the possible effect of an experimenter’s subliminal or telepathic influence on dermo-optic sensitivity. PSI Research, 3, 8-15. Pallikari, F., Boller, E., & Bösch, H. (2000). Jungian synchronicity sheds light on the micro- PK mechanism? Proceedings of Presented Papers: The Parapsychological Association 43rd Annual Convention, 210-222. Palmer, J. The challenge of experimenter psi. European Journal of Parapsychology, 13, 110- 125. Persinger, M. A. (1989). Psi phenomena and temporal lobe activity: The geomagnetic factor. In L. A. Henkel & R. E. Berger (Eds.), Research in parapsychology 1988 (pp. 121- 156). Metuchen, NJ: Scarecrow Press. Piedmont, R. (1999). Does spirituality represent the sixth factor of personality? Spiritual transcendence and the five-factor model. Journal of Personality, 67, 985-1013. Piedmont, R. L. (2001). Spiritual transcendence and the scientific study of spirituality. Journal of Rehabilitation, 67, 4-14. Piedmont, R. (2002). Cross cultural generalizability of the Spiritual Transcendence Scale in India: Spirituality as a universal aspect of human experience. American Behavioral Scientist, 45, 1888-1901. Rao, K. R. (1965). The bidirectionality of psi. Journal of Parapsychology, 29, 230-250. Rawlings, D. (2001-2002). An exploratory factor analysis of Hartmann's Boundary Questionnaire and an empirically-derived short version. Imagination, Cognition & Personality, 21, 131-144. Rhine, J.B. (1947). The reach of the mind. New York: Sloane Rhine, J. B. (1969). Psi-missing re-examined. Journal of Parapsychology, 33, 136-157. Rhine, L. E. (1953). Subjective forms of spontaneous psi experiences. Journal of Parapsychology, 17, 77-114. Richards, D. G. (1996). Boundaries in the mind and subjective interpersonal psi. Journal of Parapsychology, 60, 227-240.

Roe, C. A., Davey, R., & Stevens, P. (2003). Are ESP and PK aspects of a unitary Phenomenon? A preliminary test of the relationship between ESP and PK. Journal of Parapsychology, 67, 343-366. Schmeidler, G. R., & Edge, H. (1999). Should ganzfeld research continue to be crucial in the search for a replicable psi effect? Part II. Edited ganzfeld debate. Journal of Parapsychology, 63, 335-388. Schmidt, H. (1970). A PK test with electronic equipment. Journal of Parapsychology, 34, 175-181. Thalbourne, M. A., & Delin, P. S. (1994). A common thread underlying belief in the paranormal, creative personality, mystical experience and psychopathology. Journal of Parapsychology, 58, 3-38 Winkelman, M. (1981). The effect of formal education on extrasensory abilities. Journal of Parapsychology, 45, 321-336.

Test 15 Test Preparation Orientation phase minute phase Debrief stage Block A break Block B

B-A trials (Control 3. P is collected from first) 7a. Completion of 8 a. At beep 1, the RRC lounge area & 6a. When 10 minutes questionnaires/ relaxation seated in the intention is over, timer is reset interviews instructions and room. The procedure is to 0. Questionnaires continue. music are listened to/ explained and consent are Refreshments own method of forms signed completed/interviews provided. General preparation 1. E1 set GMF are begun, the beeper discussion, undertaken (for the switch is switched off followed by full healers) for around explanation of the 10 minutes. 4. P is taken to intention stage. c. Beep 2 sounds and hemolysis lab to see Beeper is switched healing intention is 2.E1 and E2 blood sample where the A-B trials (Experimental on.. sent for 5 minutes Debrief relax in intention will be sent. first) d. A 3rd beep signals sheets are intention room 6 b. at Beep 1, the timer that the intention provided, is reset to 0. Relaxation phase is over. general instructions and music are e. PIQ is completed feedback 5. E2 and E1 both time listened to/ own method of by non healers and 10 minutes (or 5 if preparation undertaken healers are 7b. healer session 2). (for the healers) for interviewed until the Refreshments E1 finds out study around 10 minutes. study is over (65 provided, order, and gives full c. Beep 2 sounds and mins from start ) description of study to healing intention is sent General P, including e.g., of a for 5 minutes discussion. beep, and picture of a d. A 3rd beep signals that Questionnaires blood cell undergoing the intention phase is over. are presented. 8b. After 15 mins, (40 mins hemolysis. e. Beeper switched off Interviews from start) questionnaires e. PIQ is completed by continue. the are completed/interviews non healers and healers beeper is continue , until the study is are interviewed switched off. 15 mins timed over (65 mins from start )

Information sheet and consent form for the Rhine Research Center Healing Project

Thank you very much for showing interest in our study into healing!

What is the purpose of this study?

We are looking to see whether people can use the power of the mind to influence a chemical reaction in human blood in a test tube. We are attempting to repeat an earlier experiment in which such an influence was successfully demonstrated.

Potential benefits of this study

The purpose of this experiment is to find out if the results of our group of participants as a whole demonstrate a psychic process, or psi. If the healing effect can occur successfully in these controlled conditions, this may have medical relevance for psychic healing effects inside the human body.

Potential risks of taking part in this study

We consider that the procedure is unlikely to cause you any adverse effects and demonstrates little risk for you. Participants will have no physical contact with the blood samples.

What is involved in this study?

We are looking for 20 healers to take part in this study. If you agree to take part, you will be asked to sign this consent form. The study requires that you come to the Rhine Research Center for two times in total. Each session will last approximately one hour. In each session you will be attempting to influence the blood of a different donor. The process you will be attempting to influence will be that of hemolysis, or the destruction of red blood cells in a test tube.

What is hemolysis?

Blood is added to a saline solution of a weaker concentration than that in your body. This causes the contents of cells to burst over a time course of two to three minutes. As this happens, the solution becomes increasingly transparent. The extent of hemolysis in the blood cells can be measured by placing the test tube into a device called a spectrophotometer. This assesses how transparent the solution has become.

What is your task?

You will be seated in a closed room several feet from the room containing the blood sample and spectrophotometer. Within the constraints of our experimental protocol we would like you to use whatever method of influence that is customary for you, and which you feel comfortable with. Even with this proviso, we recognize that our task is likely to differ from the way that you are accustomed to working, and thus our experiment is not intended as a test of your individual ability as a healer. Therefore, we hope that you will not feel any ‘pressure to perform’; on the contrary, we would like to encourage you to approach the task more as a fun exercise, which, nonetheless, has relevance to medicine and healing. Prior to the intervention period, you will have the option of hearing some taped relaxation instructions which may help you to focus your attention. Then, you will be given a signal, at which time you should attempt to mentally retard the process of hemolysis, which would represent a kind of ‘healing’ of the blood cells. Although you will not be able to see the actual blood sample, to help your concentration you will have the option of seeing a graph showing the progress of the hemolysis or an example of a single blood cell bursting. You will receive your results after the experiment is completed! In addition to the healing session, we will also ask you to fill out one or more short paper and pencil psychological tests. We would also like to interview you about your experiences as a psychic/spiritual healer during your sessions.

Researchers

This study is being conducted by Dr. John Palmer, Dr. Steve Baumann, Dr. Christine Simmonds and Dr. Sally Ann Drucker at the Rhine Research Center. The work is funded by a grant from the Bial foundation in Portugal. For more information about the RRC, please go to our website www.rhine.org or phone 919 309 4600. If you have any questions or queries please contact Dr. John Palmer, [email protected] , or Dr. Christine Simmonds, [email protected] or ext. 212.

Informed consent

Even if a healing effect is achieved overall, we will not be collecting enough data from any one person to determine if that person’s results demonstrate psi or if the person has psychic ability. It is unlikely that this experiment will cause you any adverse effects. However, you have the right to change your mind and withdraw from the study at any time. You may ask for your data to be removed from the analysis at any time (either during or after the experiment) and for all your records to be destroyed. All data obtained from you will be kept strictly confidential. We will not cite you by name in any publications or media outlets without your written permission. Feedback on the overall study findings and the results of your psychological tests will be made available to you if you desire them. Finally, you may decline to answer any questions with which you are uncomfortable, e.g., in the questionnaires.

I have read and fully understand the above and consent to take part in this study.

Name______Signature______Name of Researcher______Signature______

Information sheet and consent form for the Rhine Research Center Healing Project

Thank you very much for showing interest in our study into healing!

What is the purpose of this study?

Potential benefits of this study

Potential risks of taking part in this study

The study will involve working with a small sample of someone else’s blood. Although you will have no physical contact with blood, we can assure you that the blood will be healthy. It will also be drawn from one individual by a registered nurse (North Carolina Board of Nursing).

What is involved in this study?

We are looking for 40 participants to take part in this study. If you agree to take part, you will be asked to sign this consent form. The study requires that you come to the Rhine Research Center for one session. This will last for about an hour and a half. It will involve a review of the experimental procedure and you will be given the opportunity to ask any questions that you might have. You will then try to influence the process of hemolysis, or the destruction of red blood cells in a test tube.

What is hemolysis?

Blood is added to a saline solution of a weaker concentration than that in your body. This causes the contents of cells to burst over a time course of two to three minutes. As this happens the solution becomes increasingly transparent. The extent of hemolysis in the blood cells can be measured by placing the test tube into a device called a spectrophotometer. This assesses how transparent the solution has become.

What is your task?

You will be seated in a closed room several feet from the room containing the blood sample and spectrophotometer. This experiment is not intended as a test of your individual ability to heal. Therefore, we hope that you do not feel any ‘pressure to perform’; on the contrary, we would like to encourage you to approach the task more as a fun exercise, which, nonetheless has relevance to medicine and healing. Prior to the intervention period, you will hear some taped relaxation instructions which may help you to focus your attention. Then, you will be given a signal at which time you will be asked to try to mentally retard the process of hemolysis, which would represent a kind of ‘healing’ of the blood cells. To help you concentrate, you will be shown a graph showing the progress of the hemolysis or an example of a single blood cell bursting. You will receive your results after the experiment is completed! During the session we will also ask you to fill out one or more short paper and pencil psychological tests.

Researchers

This study is being conducted by Dr. John Palmer, Dr. Steve Baumann, Dr. Christine Simmonds and Dr. Sally Ann Drucker at the Rhine Research Center. The work is funded by a grant from the Bial foundation in Portugal. For more information about the RRC, please go to our website www.rhine.org or phone 919 309 4600. If you have any questions or queries please contact Dr. John Palmer, [email protected] or Dr. Christine Simmonds, [email protected] or ext. 212.

Informed consent

I have read and fully understand the above and consent to take part in this study.

Name______Signature______Name of Researcher______Signature______Date__/__/__

2741 Campus Walk Avenue, Building 500, Durham, North Carolina, 27705 tel (919) 309-4600

Post intention questions

Please tick the relevant box or mark your level of feeling about the question on the line as follows:

0 50 99

This person has rated a neutral response to the question. ______

1. How motivated were you to succeed at the healing task?

0 50 99 Not at all average in terms of motivation extremely motivated

2. Do you believe that mind can affect matter?

0 50 99 I do not believe that mind can affect matter I totally believe that mind can affect matter

3. Do you believe that human intention can heal?

0 50 99 Human intention cannot heal I totally believe that intention can heal

4.Have you had any personal experiences with “psychic” healing? Yes □ No □ If yes, and you would like to share an experience with us, please do so in the space below. ______

5. If healing works, how do you think it might work? ______

6. Please describe the method of healing that you used just now. ______

7. Did you think or feel that your method was working just now?

0 50 99 Not at all completely Please describe why you feel it was or was not working in the space below. ______

8. When you were trying to protect the blood cells, did you feel in any way that your state of consciousness was different from your normal waking state of consciousness?

0 50 99 Completely normal partially altered extremely altered

9. If healing works, where does the source of the healing energy come from?

0 50 99 Healing comes from me healing comes from outside of me

10. If you feel that the source is more external, where do you think healing energy comes from? Please describe in the space below ______

Thank you for completing this part of our study on healing