Content uploaded by Brian Laythe
Author content
All content in this area was uploaded by Brian Laythe on Nov 30, 2020
Content may be subject to copyright.
Journal of Parapsychology, 77
A CRITICAL TEST OF THE EMF-PARANORMAL PHENOMENA
THEORY: EVIDENCE FROM A HAUNTED SITE WITHOUT
ELECTRICITY-GENERATING FIELDS
ABSTRACT. Previous research in electromagnetic and geomagnetic elds (EMF and GMF) and their relationship to
paranormal phenomena has been performed under the theoretical assumptions of hallucination due to GMF elds. The
current study tests the possibility that nonhallucinatory paranormal phenomena are also associated with EMF/GMF
elds. EMF and GMF perturbations were examined in context of collected potential phenomena with data logging
equipment at a haunted site with no electricity. Overall results indicate that EMF and GMF elds were signicantly
greater in both magnitude and variability inside-the-location compared to outside-the-location baseline measurements.
Differences in GMF magnitude were small compared to EMF. Through correlation, EMF/GMF elds were demonstrated
to change in range and location throughout the duration of the investigation. Results involving individual reviewed
phenomena indicate that phenomena are strongly and signicantly associated with serial EMF and GMF spikes, that
both increases and decreases in EMF/GMF elds are not differentially predictive of phenomena, and that increases in
the number (i.e., duration) of serial spikes do not differentially predict phenomena.
Keywords:
is very limited. This lack of research is troublesome, as the assumption that variation in EMF activity
examining any relationship between EMF and haunting phenomena has been based on individually per-
ceived and sensed internal events that occur in a haunted locale. Indeed, the literature dedicated to EMF
-
-
that are captured with external recording devices. While literature exists that addresses EMF/GMF with
internal psychological perception, there is little work examining EMF/GMF with phenomena that occur as
-
within haunted locales in context of EMF and GMF. However, to examine any hypothetical relationship
-
Second, we address potential confounds in EMF/GMF measurement, including the variable of distance,
and assumptions underlying an EMF/GMF phenomena hypothesis. Finally, we address the considerable
limitations in the evaluation of “paranormal phenomena.”
A Critical Test of the EMF-Paranormal Phenomena Theory 213
EMF/GMF Thus Far in Haunted Locales
Previous research involving uctuation of EMF/GMF has been strongly inuenced by Michael
Persinger. The work by Persinger and his colleagues has demonstrated that low-level magnetic elds can
induce perceived “haunting phenomena” in laboratory settings (St. Pierre & Persinger, 2006). Laboratory
research demonstrated that application of 1 to 3 hertz magnetic elds applied to the parietal-temporal lobes
produced a “sensed presence” in about 80% of subjects (Booth et al., 2005). Per these researchers’ work,
the sensed presence is described as “the personal proximity of a Sentient being, a presence, or ‘another
consciousness’” (St. Pierre & Persinger, 2006, p. 1080). These elds not only seem to facilitate hallucina-
tions involving the presence of nearby entities but have also been shown to affect memory recall. Healey
and Persinger (2001) demonstrated that participants placed in a low-level EMF eld reported 3 times as
many false memories when recalling a narrative as those who were not exposed to EMF. In the context
of both meta-analysis and multiple experiments (e.g., St. Pierre & Persinger, 2006), these researchers hy-
pothesize that geomagnetic elds occurring naturally in certain locales might induce hallucinations that
can account for haunting phenomena (Gearhart & Persinger, 1986). In essence, GMF-induced hallucina-
tions may account for the bulk of what witnesses report as paranormal phenomena, and worse, may alter
memory to create nonfactual memories in haunted locales.
As a result of Persinger’s work, eld research examining haunting sites has focused primarily on the
presence of abnormal EMF and GMF elds. Several researchers have demonstrated signicantly different
degrees of both eld strength and variation between haunted and nonhaunted locales (Braithwaite, 2004,
2006; Braithwaite, Perez-Aquino, & Townsend, 2004; Braithwaite & Townsend, 2005; Maher, 2000; Nic-
hols & Roll, 1998; Roll & Persinger, 2001; Wiseman, Watt, Greening, Stevens, & O’ Keeffe, 2002; Wise-
man, Watt, Stevens, Greening, & O’Keeffe, 2003). For instance, Nichols and Roll (1998) demonstrated
EMF elds with Tri-Field meters that were signicantly greater in areas of reported phenomena compared
to locations with no reports of phenomena. Wiseman et al. (2002) found that eld strength and variance of
EMF/GMF were associated with the location where haunt experiences had been reported. Later research
by Wiseman et al. (2003) demonstrated signicantly greater variation in EMF/GMF as a whole, but not in
magnitude. Braithwaite et al. (2004), with the use of the Magnetic Anomaly Detection System (MADS)
which he designed, compared a “hot-spot” to a nonactive area. Their results demonstrated that GMF vari-
ability and magnitude were not only signicantly different in a test and hot-spot area but also that elds
signicantly varied during the investigation. In the above studies, Braithwaite and his colleagues were
also able to distinguish “pulses” or spikes of GMF that were signicantly stronger than baseline readings
in the dataset. Of more detailed interest, Braithwaite et al. (2004) demonstrated both substantial magnitude
and variability effects within a one-room area and demonstrated that these EMF elds occurred in the 8
to 10 hertz range.
In context of Wiseman et al. (2002) and the ndings of Braithwaite et al. (2004, 2006), a “eld
picture” of EMF/GMF begins to take shape. In general, variability and magnitude of GMF/EMF appear
greater within haunted locales but vary over time and location within the site. What is essential with the
above studies is that external phenomena, such as recorded sounds or video, were not examined with any
type of EMF/GMF readings. Thus, the previous research suggests that “haunted” sites show greater vari-
ation in EMF/GMF elds than “un-haunted” sites. As a result of this variation in EMF and GMF, these
elds could account for such experiences as hallucinations (e.g., Braithwaite & Townsend, 2005) that
individuals interpret as haunting phenomena.
This leaves an entire area of haunting phenomena unexplored within an EMF/GMF hypothesis.
Persinger’s research can account for almost any type of subjective (i.e., internally and personally per-
ceived) haunting experiences. Likewise, previous research has demonstrated, at least in part, that EMF/
GMF elds in haunted locales seem to t conditions that could induce a Persinger effect due to variability
in EMF/GMF elds (Braithwaite et al., 2004,). Where Persinger’s model does not t, hypnagogic episodes
(e.g., Cheyne, Newby-Clark, & Rueffer, 1999; Furuya et al., 2009; Kampanje, 2008; Sherwood, 2002) or
214 The Journal of Parapsychology
However, what about those instances when anomalous phenomena occur and are documented with
audio clips each year of allegedly anomalous phenomena. Although many of these events can be explained
if an anomalous event is captured on either audio or video, then this type of occurrence cannot be due
to a hallucination. Tradition in ghost hunting has maintained that EMF/GMF perturbation is associated
with external phenomena. The purpose of the current research is to examine this particular claim above
-
tive phenomena (i.e., recorded phenomena of human-shaped entities, captured psychokinesis, or quality
electronic voice phenomena) and EMF/GMF.
Critical Issues in Testing the EMF/GMF Phenomena Hypothesis
occur. We have divided these issues and their potential confounds into three primary experimental con-
cerns. First, we address the conditions regarding how EMF would theoretically perform in the context of
anomalous phenomena. Second, we discuss two variables related to EMF that have not been addressed
with the controversial issue of the evaluation of anomalous phenomena as “anomalous” as opposed to
environmentally explainable events.
The Hypothetical Nature of EMF/GMF in the Context of Paranormal Phenomena: Issues of Time
and Occurrence
-
-
lyze EMF/GMF and phenomena that occur in the context of time. In order to test this type of relationship,
Braithwaite et al., 2004), the current research makes no claims about the degree of variability, magnitude,
or duration of EMF/GMF expected as a result of anomalous phenomena. The research of Braithwaite et
-
tude of these spikes was not large (+/-60 nT/.6 mG) and typically occurred for no longer than 0.5 seconds.
-
nomena.
As such, to test openly an EMF/GMF phenomena hypothesis, small perturbations of EMF/GMF
-
ability or magnitude occur in relation to anomalous phenomena. Frankly, this may not be the case. It is
event, data have not been examined to verify or rule out any of these claims.
The Intervening Variables of EMF Fields, EMF Vectors and Distance, and Their Effects on
Measurement
In addition to the above issues that involve a lack of data regarding the EMF/GMF phenomena
A Critical Test of the EMF-Paranormal Phenomena Theory 215
hypothesis, it is also the case that previous research has not addressed a fundamental confound involving
all affect the readings of a receptive EMF/GMF meter. What is crucial to understand is that the registered
In a study in which anomalous phenomena are hypothesized to relate to EMF/GMF variation,
-
cult to determine, it is never clear whether or not changes in EMF or GMF are due to singular or multiple
the fact that the location of the EMF source cannot be determined, and therefore, the reading of a meter is
going to change as a function of that distance. In essence, EMF/GMF changes recorded from a meter will
As such, two important conditions exist in conducting research in which EMF/GMF is compared to
time-dependent potentially anomalous events. First is the understanding that the sensitivity of the meters
for examining small perturbations in EMF/GMF data around potential events. A small perturbation in the
into account with EMF/GMF readings illuminates an important fact about EMF/GMF data measured in
-
crotesla) of the reading are useless aside from comparing a reading to the distributions that are collected.
-
the collected EMF/GMF meter readings, but cannot provide accurate information about the actual strength
Collecting the Observable: Observable Anomalous Events, Their Evaluation and Interpretation
-
a particular captured event can be reliably viewed as paranormal or anomalous.
Research in social and cognitive psychology provides good reason for individuals to be skeptical
make certain personal events very suspect in terms of their anomalous origins.
216 The Journal of Parapsychology
-
ases. For instance, contagion effects can change memories (Merklebach, Van Roermund, & Candel, 2007),
as well as emotions and behavior (Barsade, 2002), and create stronger biases when individuals are familiar
event, even when recorded on audio or video. Likewise, research in belief perseverance (Lepper, Ross,
the face of contradictory information. Thus, cues indicating the likelihood of an anomalous event can be
easily confound the assessment of evidence.
Machado & Silva, 2007), we suggest that there is never a situation in which an anomalous event is con-
events are much more likely (in terms of probability) to be anomalous compared to other events. However,
should be used to rate each event as either very unlikely to be paranormal or very likely to be paranormal.
-
tigator will accept evidence as anomalous that can be easily explained.
The criteria should involve minimum requirements by which an investigator can rule out an event
due to environmental or psychological factors. It should also contain criteria that make an event more
persuasive in terms of its paranormal validity, such as phenomena that can be perceived as a repeated
intelligent response to an investigator. To that extent, the current research employs an Evaluative Model
-
ronmental explanations. Through the use of this procedure, phenomena can be assigned additional points
based on optional, but contributing, factors that would suggest that an event is more likely to be anoma-
lous. All of the EMPE criteria are designed to eliminate psychological and environmental conditions that
could lead to the misinterpretation of a nonanomalous event as anomalous.
Summation and Hypotheses
In summation, we propose that more analyses of haunted locations are necessary, beyond the
research that establishes EMF/GMF variation and magnitude differences within haunted locations. We
make the general claim that a detailed but open-ended study is necessary to determine how EMF/GMF
-
tive haunting phenomena, namely anomalous noises and events physically viewed via camera may have
exploratory hypotheses are as follows:
-
side the location compared to baseline readings taken outside the house.
2. As an exploratory function, we examine correlations of meters placed inside the investigation site
location.
or decreases in EMF and GMF readings.
4. We will explore any differences between increases, decreases, and variability of EMF and GMF in
A Critical Test of the EMF-Paranormal Phenomena Theory 217
relation to the occurrence of phenomena, and investigate whether the duration of a given increase
or decrease of EMF/GMF magnitude is in any way associated with anomalous events.
Method
Participants and Sample
With the help of the Ivy Tech Paranormal Organization (ITPO) and the Association for the Study
of Anomalous Field Phenomena (ASAFP), an investigation was conducted at Black Moon Manor (BMM),
a site considered to be known for haunting activity by many previous paranormal investigative groups.
. All members were briefed in protocols for the current study, and have had
previous training on data collection methods on previous investigations.
Brief History and Details of Location
Black Moon Manor was built by John C. Eastes in Hancock County, Township of Buck Creek,
-
loween attraction. The owner relates that the history of the home he obtained was told to him by a woman
who claimed to have previously lived at the location. One story contends that the manor was once used as
a house for smallpox patients during an outbreak of the disease. The current manager reports more than
200 deaths at the home, and there is an unmarked cemetery in the back of the manor where Eastes family
members are interred, including a girl named Racheal Eastes, who was 5 years old at the time of her death.
Other claims of deaths include several members of the Eastes family, a woman/girl drowning in the well,
are also reports of a child named Martha who haunts the home, along with another entity named Henry or
Larry (A. Hansford, personal communication, June 18, 2011).
Historical research uncovered partial support for these accounts. The house was a home of the
-
mented drowning of a young girl named Rachael, who was a niece of John Eastes. Similarly, another
-
ily members. Additionally, a farmhand, Henry Beckner, also lived at the location with the Eastes family
records.
Regardless of its history, Black Moon Manor was selected due to reports of and evidence for ac-
tive external phenomena that have occurred there. Reports of activity made by many investigative groups
EVPs of varying quality. The source of the activity has been attributed to both the deceased family mem-
bers and to the smallpox victims that were reported to have died at the location.
With regard to physical details of the location, the house is situated approximately one quarter
mile from a traveled road, in an area approximately three quarters of a mile in all directions away from
the nearest residential housing, thus limiting human interference. The house is a two-story dwelling of
approximately 2,000 square feet. It was determined that the building lacked electricity based on the fol-
lowing indicators:
1. There were no wires leading from the electricity pole to the building or fuse box, nor any wires
leading from the pole to the main electricity poles along the road. Investigation of the outside of
218 The Journal of Parapsychology
the building demonstrated no generator sources of electricity, and no wires above ground leading
to the house. Inspection of the basement, including the underside of the house, indicated no heavy
gauge wires or electrical connections.
2. A room-by-room investigation was conducted in search for wires, speakers, or any type of electri-
cally powered equipment that could assist in hoaxing.
including along electrical outlets so as to verify an absence of electricity. No readings or spikes
were detected.
Equipment
Generator
25 feet from the investigation site to power all equipment. Power was provided through a 12-gauge exten-
sion cord to an equipment table set up on the porch of the building.
Real Time Investigative Ghost Hunting System (RIGS) At the beginning of the investigation,
and one Natural EM meter) in locations inside the building. One additional Tri-Field 100XE and Natural
magnitude for baseline data. Tri-Field 100XE meters are reported to have a resolution/sensitivity of .2 mG
without coils (with coils, .002 mG). Natural EM meters report a resolution/sensitivity of 10 mG. Addi-
footsteps or general movement around the range of the meter.
For both meter types, the manufacturer claims that calibration is reliable and that drift of read-
ings due to loss of meter calibration over time is not possible. Placement of meters inside the structure
was dictated by the owner, who rated each room in terms of the most frequent activity experienced by
on “mains” frequency commonly calibrated for normal electricity and EMF production. In addition, Tri-
Field meters in the current study were attached to magnifying coils, which increased sensitivity to detect
very small perturbations within the 0–1 mG range. These coils do, however, convert the normally three-
axis Tri-Field meter into a single-axis meter. Therefore, readings from our EMF meters represent one-axis
assessments. Natural EM meters did not have coils and were calibrated on the magnetic setting, thus mea-
at 24 times per second
this type can produce reliable readings at this level of sampling, although a delay or suppression of magni-
tude can occur at this sample rate. As our analysis examines extreme readings of the meter, this fault in the
equipment emphasizes conservative readings, and would not confound EMF spikes in a manner similar
Recordings of these data were collected in anywhere from 2-hour to 4-hour increments and time-synced
with other equipment according to the protocol described below.
DVR system.-
red video and audio of all sites where the meters were placed. These videos were collected continuously
in 15-min intervals for each camera for the duration of the investigation. Each video channel was time-
stamped to EMF and GMF readings.
AVR units. Battery-powered digital audio voice recorders (AVRs) were also placed where both
meters and video had been placed in the location. These battery-powered units were set to “conference”
A Critical Test of the EMF-Paranormal Phenomena Theory 219
setting in order to record as much area as possible. The AVRs were time-stamped in order to determine
EMF and GMF readings when specic noises or phenomena occurred.
Protocol
Setup. Several steps were taken to minimize human contamination during the course of the in-
vestigation. First, only investigators were present on site during the time of data collection. Second, two
separate teams, with the use of Tri-Field meters, examined the house for any signs of trickery, including
wires, trap-door access, as well as sources of electrical power as mentioned above. In both examinations,
no signs of trickery or sources of electricity were present. Access to the site was only available from one
side of the house, where non-investigating members could observe entrances or exits from the building
across the duration of the investigation.
After setting up room numbers for areas of interest, one Tri-Field XE100 and one Natural EM
meter were placed approximately 3 feet apart apart from one another in the middle of each room, and then
connected to the RIGS system. At least two video cameras were placed in each area, positioned from 8 to
10 feet away from any individual meter so as to prevent EMF contamination. Battery-powered AVR units
were placed between the meters. However, in previous testing, the electrical output of these units were
shown not to register on either type of meter. The combination of audio, video and meter placement within
these four specic areas created “data collection traps” by which multiple audio, video, and EMF/GMF
readings could be sampled at the same time. Key to this strategy was a precise log of the start and stop
times of all three components based on a common time unit. The result of this procedure was that audio,
video, and EMF/GMF readings were synchronized, allowing for accurate comparison of all three within
a common time frame. This type of meter layout allowed for four hot-spot areas where data could be reli-
ably collected, and for one set of baseline meters placed approximately 7 feet outside of building. Cameras
and audio recorders were not placed with baseline meters. Note that any phenomena that occurred in other
areas were deemed interesting, but invalid for the purposes of this study.
Once the devices were set up and time-stamped, a log of the individual movements and location
of all investigators was maintained at all times. Data collection periods were of two types: interactive
and noninteractive. Noninteractive sessions involved all members removing themselves from the site, al-
lowing the meters’ audio and video to run without any human involvement. Interactive sessions typically
involved one team of two to four people interacting within the recorded environment, in order to attempt
to facilitate PK (e.g., knocking or movement) or EVPs within that location.
Classifying objective and subjective haunting events. Although many personal experiences
were common over the investigation, these events were deemed to be subjective, and not of interest to
the current study. However, evidence captured on either audio or video was held to a standard set of cri-
teria for evaluation referred to as the Evaluative Model for Paranormal Evidence (EMPE). Eight possible
points could be assigned to phenomena based on the following criteria. Points 1–3 are granted to address
the basic quality of recorded evidence criteria such as “Is the event external?” “Is the picture clear?” “Is
the audio clear?” and “Has the entire event been captured?” Points 1–3 also address common natural
phenomena that are mistaken for anomalous activity (e.g., grunts, knocks, thumps, bumps, moans, ani-
mals, dust, mist, distant shadows, or light reection that are likely due to the environment). Any of the
aforementioned events were automatically relegated to Class 1 as due to multiple environmental causes
that could not be reliably ruled out. Point 4 is granted if any reliable or probable alternative means can be
ruled out for a particular phenomenon. For example, this point was granted if other sources of video or
audio could not explain the event and/or if the event could not be re-created. Point 5 was granted when
human interference or hoaxing could be reliably ruled out (e.g., physical contact with objects, whispering,
or subvocalization). An additional point was granted for: events that had no human interaction component;
phenomena that were captured and were very clear and distinct in terms of audio or video, thus making
subjective interpretation less likely; and phenomena that appeared to be repeated, rational, and intelligent
responses to a human agent.
220 The Journal of Parapsychology
In terms of evaluation, potentially anomalous events were given a rating representing an estimate
of the likelihood that each event may or may not have been anomalous. From that rating, the quality of
evidence could be examined and understood as either: Class 1, likely to be environmental (EMPE score
1–3); Class 2, possibly environmental, but also possibly anomalous (EMPE score 4–5); and Class 3, more
likely to be anomalous (EMPE score 6–8). Events that were clearly environmental (e.g., car, bird, animal,
or investigator) or events that demonstrated the presence of additional concurrent noise in other audio
sources were labeled as Class 0. Please note that in order for EVP to be considered of quality for the cur-
rent study mandated, the noise must not have been present at the same time in any of the other audio re-
corders placed in the house. Thus, the audio event was isolated to one particular recorder, greatly reducing
the probability that the voice was due to either the investigators or the environment. Similarly, any video
event of note was examined for investigators in the principal video but also checked against additional
cameras to ensure that investigator contamination was not present.
Classication occurred in three phases. First, three reviewers did an evaluation of all captured
events and assigned them a rating according to the EMPE system. Any individual event that reached Class
2 was compared against all relevant additional audio and video sources by at least two of the three re-
viewers. Thus, error was minimized by independently performing a full audio or video comparison twice.
Reviewers had strict instructions about the acceptance of the captured phenomena as Class 3, based on
the lack of presence of additional investigators, or the lack of audio noise on other recorders. Re-creation
attempts were also performed on site prior to the designation of Class 3. As such, any event reaching Class
3 represents an event that, to the best of our knowledge, was not caused by human or environmental con-
tamination.
However, initial reviewers were aware of what events were occurring on an EMF spike. Thus, a
second group of three evaluators who were naïve as to whether or not an event had triggered a spike, and
who had not been present during the data collection period, performed a second review. During the second
analysis, the new reviewers were provided only with the core audio and video evidence and did not engage
in the full multiple camera or audio comparison process described above. Thus, the second review process
served as a simple manipulation check against possible rating bias due to spike knowledge. Congruence
between these two series of ratings was 80%. To address the differences in comparison, and to ensure that
ratings were not affected by knowledge of a spike occurring with the remaining events, an additional two
individuals who were also naïve performed a complete review, following the protocol described above, of
all comparison audio and video sources. These new reviewers gave a nal rating for any disputed event
that reached Class 2 or Class 3. Disputed Class 0 or Class 1 events maintained their original review status.
This was due to the fact that all Class 0 events had been factually ruled out as contaminated regardless
of spike, and that Class 1 events, due to multiple environmental causes, could not be safely classied as
anything reliably anomalous. Thus through this three-step process, the event dataset was fully analyzed
against all relevant audio and video sources to ensure a lack of contamination for events, and that the
ratings either corresponded to naïve review, or if in contention, were fully vetted through a naïve review.
Analysis and coding of EMF/GMF. As the goal of the current article was to be consistently
thorough, the statistical analysis approach to the EMF/GMF data is somewhat unorthodox. Again, the
reading of a meter is determined in part on the distance of one or multiple elds. Because of our inability
to determine the location(s) of the elds, and of our equipment being limited to magnitude only (and not
frequency), we chose to report magnetic data in the raw volt input of the meters themselves. We are aware
that it is traditional to convert magnetic readings to mG or nT; however, the mapping process is approx-
imate, and use of mG converted from volt input could add signicant error to the analysis. As our main
hypothesis involved the simple association of variation in EMF/GMF as a result of potential phenomena
that occurred, and our equipment is unable to give us information beyond an increase or decrease in mag-
nitude of the meter, it seemed most honest to present the data as simple volt input distributions. Thus,
standard statistical analysis is presented with raw volt input, with the understanding that variation of volt
input EMF through the Tri-Field 100XE meters represents a range of 0 to 1 mG, and that volt input of the
A Critical Test of the EMF-Paranormal Phenomena Theory 221
Natural EM meters represents a range of 0–100 mG. Again, of primary concern is a change in magnitude
that occurs in association (through time) with recorded events in the environment that are independent of
the data-logging EMF/GMF system.
Whereas baseline tests and descriptive analysis are performed using standard inferential statistics
based on a normal distribution, our hypotheses involving individual phenomena and EMF/GMF behavior
were examined differently. For the current study, three or more spikes of EMF and GMF elds within 1
second were isolated from the dataset at either 2.5 or 3 standard deviations above or below the session
mean. Each of these spikes (which represent 1/24th of a second) was time-stamped, and investigators then
examined audio and video based on the second where a series of three or more 2.5 or 3 standard deviation
serial spikes occurred within one second. Each of these data points where EMF/GMF spikes occurred
were identied as positive (e.g., increase in magnitude), negative (e.g., decrease in magnitude), or mixed
(e.g., both 3-SD increases and decreases in magnitude, indicating extreme variability). Potential phenom-
ena that occurred within the 1-second boundary were considered associated with the spike. However, for
clarity, a 0.5-second time lag was allowed for human errors in timing, and potential suppression from the
input of the analog meters at 24 samples per second. Other phenomena that did not occur within the 1-sec-
ond window were also collected and noted.
Of interest in the current dataset were the number of serial spikes (i.e., more than three spikes
within 1 second) that occurred. As such, analysis focused not on general signicance, but on the number
and count of probabilistically unlikely spikes that occurred around the time of analyzed phenomena. The
rationale for this method, although unorthodox, is one of precision. Previous research has not explored ei-
ther the degree or frequency of EMF/GMF uctuation as a function of paranormal phenomena. Thus, there
is no way to know how frequently or how strongly EMF/GMF might rise or decline, should the hypothesis
prove to be supported. Looking at the data backwards allows for a precise analysis of EMF/GMF spikes
and how often they occur. This process also prevents masking of the data by examining a mean average
where three to ve spikes over the course of a second are hidden by the overall average.
Results
Social science researchers traditionally worry about having a large enough data set to test hypoth-
eses reliably. The current research has the opposite problem. Sample sizes in some cases are so large due
to the collection of 24 samples per second over hours of time that very small mean differences and covar-
iation appear statistically signicant, but have no practical signicance. In large mean difference tests, we
addressed this by calculating the Cohen’s δ which provides an effect size statistic for the size of the mean
difference. In cases of correlation, we encourage the reader to focus on the effect size statistic and not
necessarily on the statistical signicance of the test.
Descriptive Statistics and Replication: General EMF/GMF Baseline Comparisons
In order to examine and test the overall means of EMF and GMF over a period of time, means and stan-
dard deviations for each meter are provided for each recorded session in Table 1. Baseline meters (i.e.,
meters placed 7 feet outside the location) were meters 9 and 10. In order to test whether EMF/GMF elds
signicantly differed in either magnitude or variability in the haunted location as compared to outside of
the building, independent-sample t tests, and Levene’s tests for inequality of variances were conducted
for both EMF and GMF elds across all of the sessions. These ndings serve as a partial replication of
previous research conducted by Braithwaite et al. (2004) and Wiseman et al. (2002, 2003). Results are
provided in Tables 2 and 3.
222 The Journal of Parapsychology
Table 1
Means and Standard Deviations for the 10 Meters in Each of the Four Sessions
Session 1 (n = 176,586) Session 2 (n = 165,778) Session 3 (n = 361,936) Session 4 (n = 54,019)
Meter M Min Max SD M Min Max SD M Min Max SD M Min Max SD
Ch1 T .16 .12 .20 .01 .17 .13 .21 .01 .19 .11 .26 .02 .22 .18 .27 .01
Ch2 G .01 -.18 .21 .10 .01 -.18 .21 .11 .01 -.25 .29 .14 .01 -.27 .31 .17
Ch3 T .27 .23 .30 .01 .27 .22 .31 .01 .30 .20 .37 .02 .34 .30 .38 .01
Ch4 G .01 -.03 .06 .01 .01 -.03 .06 .01 .01 -.04 .06 .01 .01 -.05 .42 .02
Ch5 T .17 .10 .22 .01 .17 .10 .22 .01 .18 .10 .25 .01 .19 .13 .25 .02
Ch6 G .01 -.03 .05 .01 .01 -.03 .05 .01 .01 -.04 .07 .01 .01 -.04 .07 .01
Ch7 T .07 .03 .10 .01 .07 .03 .10 .01 .07 .03 .11 .01 .08 .03 .12 .01
Ch8 G .01 -.01 .03 .00 .01 -.01 .03 .00 .01 -.01 .33 .01 .01 -.01 .04 .01
Ch9 T .11 .07 .14 .01 .11 .07 .13 .01 .11 .07 .14 .01 .12 .09 .15 .01
Ch10 G .01 .00 .02 .00 .01 .00 .02 .00 .01 .00 .03 .00 .01 .00 .02 .00
Note. Data displayed in millivolts. Numbers are in volt input. Approximate mapping of millivolts to milligauss for Tri-Field:
.000–.075 mV= .000–.037 mG;.075–.150 mV = .037 to .085 mG; .150-.225 mV = .085–.250 mG; .225–.300 mV = .250–1.0
+ mG. Approximate mapping of millivolts to milligauss for Geomagnetic Meters: .000–.075 mV = .000–3.7 mG; .075–.150
mV = 3.7– 8.5 mG; .150–.225 mV = 8.5–25.0 mG; .225–.300 mV = 25.0–100.0+ mG
Results indicated that EMF magnitude and variability were signicantly greater for all of the EMF
Tri-Field meters inside the haunted location in comparison to single readings taken outside. In terms of
mean differences, readings inside the location ranged from 50% to 100% greater compared to outside the
location (see Table 2). In all cases, variability inside those locations with activity spots was greater than
the variability obtained from the outside baseline EMF/GMF meters. Statistical signicance for both mean
differences and variability was well beyond p < .001, as indicated by the strength of t scores and Levene
ratios. Effect sizes provided by Cohen’s δ in all cases exceeded 1.0 thus indicating large differences be-
tween means.
GMF mean scores also differed signicantly (p < .01) with one exception: at locations with ac-
tivity compared to the single outside baseline meters across sessions. However, Cohen’s δ tests for GMF
show that, while statistically signicant, these mean differences were very small in terms of effect size
(δ = .0 to .11, see Table 3). Although mean differences were small, similar to EMF ndings, variability
for GMF was signicantly greater across all of the sample sets within all sessions in comparison to the
baseline data. In this case, variability was similar to EMF, where the signicance of GMF variability was
less than .001 in all cases.
Correlation of Meter Readings Across Sessions: Test of EMF and GMF as Separate
Measurements, and Common Sources of EMF/GMF Generation
Product moment correlations were conducted for each session between all of the meters as part of
a manipulation check to ensure that EMF and GMF meters were measuring different hertz ranges, as well
as an exploratory investigation of covarying meters as a potential indicator of a common source of EMF/
GMF eld production. These results are provided in Table 4.
A Critical Test of the EMF-Paranormal Phenomena Theory 223
Table 2
Baseline Tests: Magnitude and Variability Comparison of Tri-Field
(EMF) Meters for Each Session
Meter No. Base Mean Comp. Mean Mean
t
Mean
p
Cohen
δ
Base
SD
Comp
SD
Var.
F
Var.
p
SESSION 1 (n = 176,586)
10.109 0.165 -2113.83 .001 7.32 .006 .009 2.62 .001
3 0.109 0.268 -7188.11 .001 24.38 .006 .007 1.57 .001
50.109 0.172 -2150.21 .001 7.11 .006 .011 3.53 .001
70.109 0.068 1754.10 .001 5.79 .006 .008 1.90 .001
SESSION 2 (n = 165,778)
1 0.107 0.167 -2306.23 .001 7.84 .006 .009 2.00 .001
30.107 0.269 -6912.62 .001 24.84 .006 .007 1.40 .001
5 0.107 0.170 -2013.02 .001 7.11 .006 .011 3.30 .001
7 0.107 0.068 1548.46 .001 5.15 .006 .008 1.75 .001
SESSION 3 (n = 361,936)
1 0.114 0.195 -2485.41 .001 5.81 .008 .018 5.03 .001
30.114 0.303 -4512.60 .001 10.56 .008 .024 9.10 .001
5 0.114 0.182 -2391.22 .001 5.65 .008 .015 3.56 .001
7 0.114 .073 1991.00 .001 4.52 .008 .010 1.49 .001
SESSION 4 (n = 54,019)
1 0.121 0.221 -1981.90 .001 11.70 .005 .011 3.91 .001
30.121 0.342 -4676.90 .001 27.95 .005 .010 3.25 .001
5 0.121 0.195 -1080.50 .001 6.61 .005 .015 7.95 .001
7 0.121 0.077 857.89 .001 5.14 .005 .011 4.01 .001
Note: Numbers are in volt input. Approximate mapping of millivolts to milligauss for Tri-Field: .000–.075 mV = .000–.037
mG; .075–.150 mV = .037– .085 mG;.150–.225 mV = .085–.250 mG; .225–.300 mV = .250–1.0 + mG
As Table 4 indicates, the vast majority of relationships between meters were statistically signi-
cant, due to sample size. However, for practical interpretation, correlations greater than .30 were under-
lined, indicating where two meters shared approximately 10% or greater covariation. Correlations indicate
that Tri-Field meters did not substantially correlate with Natural EM meter readings. In the majority of
cases, the r between these meters was less than .10, and in some few instances inversely related, but below
our arbitrary .30 threshold. In other words, these correlations provided evidence that both types of meters
were registering different frequencies of the electromagnetic eld. The lack of correlation between EMF
and GMF meters suggests that frequencies of EMF were not produced in the higher end of the Natural EM
meter, nor in the lower end of the Tri-Field meter. In theory, such a scenario would produce covariation
between both types of meters if coming from a common source.
However, as more of an exploratory analysis, correlations were also examined to determine
whether common sources of EMF within the “haunted” location produced similar increases or decreases
in EMF/GMF readings. These relationships are graphed according to meter placement in Figure 1. A
re-examination of the correlation tables within each session indicates that EMF and GMF readings do
224 The Journal of Parapsychology
Table 3
Baseline Tests: Magnitude and Variability Comparison of Geomagnetic
(GMF) Meters for Each Session
Meter No. Base Mean Comp. Mean Mean
t
Mean
p
Cohen
δ
Base
SD
Comp
SD
Var.
F
Var.
p
SESSION 1 (n = 176,586)
1 .012 .013 -7.16 .001 0.01 .002 .104 2565.80 .001
3.012 .012 -10.29 .001 0 .002 .010 25.42 .001
5 .012 .012 -22.39 .001 0 .002 .010 21.53 .001
7 .012 .012 8.08 .001 0 .002 .005 5.51 .001
SESSION 2 (n = 165,778)
1 .012 .014 -7.85 .001 0.02 .002 .106 2674.39 .001
3.012 .012 -14.55 .001 0 .002 .010 25.46 .001
5 .012 .012 -22.48 .001 0 .002 .010 22.41 .001
7 .012 .012 3.78 .001 0 .002 .005 5.67 .001
SESSION 3 (n = 361,936)
1 .011 .014 -10.28 .001 0.03 .002 .141 4081.98 .001
3.011 .012 -15.89 .001 0.11 .002 .012 30.96 .001
5 .011 .012 -22.40 .001 0.11 .002 .012 29.79 .001
7 .011 .011 1.22 .22 0 .002 .006 7.52 .001
SESSION 4 (n = 54,019)
1 .011 .014 -4.13 .001 0.02 .002 .165 4776.00 .001
3.011 .012 -14.24 .001 0.08 .002 .017 52.00 .001
5 .011 .012 -6.58 .001 0.10 .002 .014 33.90 .001
7 .011 .011 3.24 .001 0 .002 .006 7.12 .001
Note: Numbers are volts input. Approximate mapping of millivolts to milligauss for Geomagnetic Meters: .000-.075 mV = .000–3.7 mG;
.075-.150 mV = 3.7 to 8.5 mG; .150-.225 mV = 8.5-25.0 mG; .225-.300 mV = 25.0-100.0+ mG
strongly correlate (or inversely relate) with their own meter type within sessions, despite the fact that pairs
of meters were placed more than 15 feet apart in separate rooms. In some instances, these relationships
are more perplexing as the meters are more than 30 to 40 feet apart. Although the precise meaning can be
debated, as Figure 1 shows, Meters 5 and 6 were always inversely related to other meters in other locations
in the house. In some cases, other meters would correlate substantially with each other, depending on the
time, and then become unrelated or inversely related, depending on the session. These relationships,
particularly when they were inverse, suggest the presence of multiple sources of EMF variation in a
house with no known articial means to create them.
A Critical Test of the EMF-Paranormal Phenomena Theory 225
Table 4
Correlations of EMF and GMF Readings Between Meters for Each Session
Session 1 (n = 176,586) Session 3 (n = 361,936)
Meter 1 2 34 5 6 7 8 9Meter 1 2 34 5 6 7 8 9
1T 1T
2G .14 2G .11
3T -.20 -.14 3T .73 -.06
4G .02 -.04 .00 4G .00 -.05 -.02
5T .08 .11 -.41 .02 5T .43 .16 .28 .01
6G -.02 .01 .02 -.62 -.01 6G -.04 .02 -.01 -.61 -.02
7T -.11 -.22 .40 -.01 -.50 .03 7T .09 -.27 .34 -.02 -.41 .02
8G .03 -.12 -.01 .37 .02 -.64 .00 8G .00 -.11 -.04 .35 .01 -.61 -.02
9TB .31 .01 .01 .02 .25 -.01 -.01 .05 9TB .68 .03 .64 .00 .44 -.02 .12 .00
10GB -.03 .02 .02 -.08 -.02 .19 .04 -.31 .05 10GB -.06 .01 -.03 -.10 -.06 .19 .03 -.33 -.03
Session 2 (n = 165,778) Session 4 (n = 54,019)
Meter 1 2 34 5 6 7 8 9Meter 1 2 34567 8 9
1T 1T
2G .16 2G .21
3T -.25 -.14 3T -.07 -.19
4G .02 -.03 .00 4G .04 -.03 .01
5T .11 .12 -.44 .02 5T .08 .20 -.43 .02
6G -.04 .02 .02 -.62 -.01 6G -.08 .00 .03 -.44 -.02
7T -.19 -.22 .39 -.01 -.48 .03 7T -.14 -.29 .32 .00 -.60 .03
8G .04 -.12 -.02 .39 .02 -.64 .00 8G .07 -.10 -.05 .23 .04 -.60 -.01
9TB .22 .01 -.02 .01 .33 -.01 .03 .05 9TB .22 .06 .05 .01 .10 -.01 -.09 .05
10GB -.03 .03 .01 -.12 -.03 .20 .04 -.31 .05 10GB -.05 .01 .05 -.07 -.06 .12 .07 -.32 .03
Note: Bold = p < .05, Underline = r > .30
Phenomena-Based Analysis: EMF/GMF and Its Potential Association to Objective Anomalous
Phenomena
In terms of phenomena, the site performed exceedingly well. Although many subjective feelings,
noises, and sensations were present during the investigation, this location produced numerous phenomena
that were difcult to explain by normal means. Some of these events were captured by our equipment,
some were not. For example, both coins and candy appeared on the oors of the room without any appar-
ent means of doing so. In addition, a milk crate both disappeared and reappeared in the basement of the
house. In both instances, cameras could not determine how these events occurred.
Likewise, two cases of apparent apparitions occurred in which a series of shadows of 7- to 8-foot-
tall people were captured on video while the house was empty. We conducted further investigation with
available light sources but were unable to re-create these shadows from any aspect of the room or house.
Childlike voices not belonging to any investigator were also captured on audio, despite the fact that there
226 The Journal of Parapsychology
Figure 1.
for EMF (T) and GMF (G) Meters in Each Session
were no children present and residential houses were not nearby. Thus, this house appeared to have genu-
ine anomalous activity occurring within it.
Examination of Events With and Without EMF/GMF Spikes in Comparison to Captured Potential
Phenomena
In order to test the hypothesis that EMF/GMF increases and decreases were associated with phe-
category of phenomena collected. Results are provided in Table 5. The expected count for events that
occurring a spike was determined by the overall ratio of spike seconds (e.g., the number of seconds dur-
assumption of the test is that if phenomena events are random then the number of events that occur dur-
ing spikes should not exceed the overall ratio of spike seconds compared to nonspike seconds during the
course of the investigation. In order to account for the fact that some events lasted more than 1 second,
the expected ratios were multiplied by 4, providing an expected ratio representing the random expected
occurrence of these spikes over 4 seconds. Please note that all events recorded across the classes of phe-
nomena were shorter than 4 seconds. However, using a 4-second anomalous event time in comparison to
the overall seconds of spikes creates a more conservative test with regard to the expected counts of the chi
-
timate. As Table 5 demonstrates, across all categories of phenomena, including ruled out and explainable
only (p = .0006 to 10-6).
A Critical Test of the EMF-Paranormal Phenomena Theory 227
Table 5
Chi-Square Tests of Evidentiality Categories by Presence or Absence of Serial Spikes
Events with Spikes Events Without Spikes
Category (Tri & Geo) O E O E χ2p
Class 0 32 8.20 9 32.80 86.34 10-6
Class 1 57 18.8 37 75.20 97.02 10-6
Class 2 10 3.60 8 14.40 14.22 .000162
Class 3 8 1.80 1 7.20 26.69 10-6
Percentage Expecteda
20 80
Category (Tri Only) O E O E χ2P
Class 0 14 1.60 6 18.40 104.45 10-6
Class 1 29 3.84 19 44.16 179.18 10-6
Class 2 30.56 4 6.44 11.55 .000675
Class 3 6 0.56 1 6.44 57.44 10-6
Percentage Expectedb
8 92
Category (Geo Only) O E O E χ2p
Class 0 18 2.52 318.48 108.06 10-6
Class 1 28 5.40 17 39.60 107.48 10-6
Class 2 71.32 49.68 27.77 10-6
Class 3 2 0.24 0 1.76 14.66 .000128
Percentage Expectedc
12 88
Note: aExpected = 1,592(spike seconds)/31,597(total investigation seconds) = 5% rounded times 4 (20%). bExpected
= 534(spike seconds)/31,597(total investigation seconds) = 2% rounded times 4 (8%). cExpected = 1,058(spike
seconds)/31,597(total investigation seconds) = 3% rounded times 4 (12%)
Examination of Positive, Negative, and Mixed EMF/GMF Spikes and Captured Potential Phenomena
In order to examine whether increases, decreases, or mixed increases and decreases (variability)
differentially predicted events, additional chi squares were conducted within each class of phenomena.
Results are shown in Table 6. Expected counts were calculated by the percentage of serial increases, de-
creases, and mixed-second events collected within the total sample. Again, the assumption of the test is
that if a particular type of spike is more predictive of phenomena, then it will exceed the expected percent-
age of its spike type from the entire sample when compared with phenomena. As Table 6 indicates, the
proportion of increases, decreases, and variability do not signicantly differ from the overall sample. As
such, serial spikes associated with phenomena do not appear to relate specically to a particular type (e.g.,
positive, negative, or mixed) of EMF/GMF spike.
228 The Journal of Parapsychology
Table 6
Chi-Square Tests of Evidentiality Categories by Types of Spikes
Positive Spikes Negative Spikes Mixed Spikes
Category (T&G) O E O EO E χ2p
Class 0 12 9.61 14 16.74 5 4.65 1.07 .58
Class 1 19 17.67 30 30.78 8 8.55 0.15 .92
Class 2 52.79 24.68 2 1.35 3.75 .15
Class 3 2 2.48 5 4.32 1 1.20 0.23 .88
Percentage Expecteda31 54 15
Category (Tri-Field Only) O E O E O E χ2p
Class 0 43.85 9 9.08 1 1.04 0.01 .99
Class 1 9 7.97 18 18.82 2 2.15 0.18 .91
Class 2 1 0.55 1 1.298 0 0.15 0.58 .74
Class 3 1 1.65 4 3.89 1 0.44 0.95 .62
Percentage Expecteda27.5 64.9 7.4
Category (Geo Only) O E O E O E χ2p
Class 0 8 6.20 5 7.58 4 3.20 1.60 .48
Class 1 10 10.22 12 12.48 6 5.26 0.13 .93
Class 2 4 2.55 13.12 21.36 2.56 .27
Class 3 10.73 10.89 00.38 0.49 .78
Percentage Expecteda36.5 44.6 18.8
Note: a Expected percentages derived from total number of positive, negative, and mixed spikes over duration of investigation.
Examination of Number of Spikes in Association with Captured Potential Phenomena
Finally, in order to determine if the overall number of serial spikes (e.g., variability) differentially
predicted phenomena, chi squares were again employed within each phenomena category. Results are
shown in Table 7. As per the previous analyses, expected counts were determined by the total sample per-
centage of the number of serial spikes (within 1 second) that occurred within the total dataset. As Table 7
indicates, across all categories of phenomena, the number of spikes that occurred on the given second of
the event and, therefore, the duration of the spike, was not a signicant predictor of events across catego-
ries with the exception of Class 2 events (p = .007). Examination of Table 7 suggests this signicant nd-
ing comes from a greater number of 5 or more serial spikes in 1 second occurring at a greater frequency
than expected.
Discussion
The goal of the current study was to examine the hypothesis that EMF/GMF changes as a function
of anomalous phenomena that occur in purportedly haunted locations. We deemed our hypotheses explor-
atory, because previous research (e.g., Braithwaite, 2004, 2006; Braithwaite & Townsend, 2005; Wiseman
et al., 2002; 2003) has examined degree and magnitude of EMF or GMF in a supposed haunted location
but has not examined objective captured evidence of potentially anomalous events in relation to these
elds. The overall ndings of the current research appear to support previous research that showed that
A Critical Test of the EMF-Paranormal Phenomena Theory 229
Table 7
Chi-Square Tests of Evidentiality Categories by Numbers of Serial Spikes
3 Spikes 4 Spikes 5 Spikes 6+ Spikes
Category O E O E O E O E χ2p
Class 0 21 22.40 7 4.80 31.80 1 2.70 2.93 .40
Class 1 41 39.90 11 8.55 1 3.36 44.95 2.58 .46
Class 2 56.30 11.35 3 0.55 0 0.78 11.99 .007
Class 3 5 5.60 1 1.20 1 0.46 1 0.69 0.84 .83
Percentage Expecteda70 15 5.8 8.7
Note: Expected percentages derived from total number of serial spikes collected over duration of investigation.
EMF/GMF elds are abnormal, or at least different, in these “haunted” locations. Also, we provided the
rst eld evidence (to our knowledge) that EMF and GMF do seem to predict potentially anomalous phe-
nomena. We address individual hypotheses below.
General EMF/GMF Magnitude and Behavior
Our examination of EMF/GMF demonstrated that eld strength in terms of magnitude and vari-
ability was signicantly greater at locations with activity compared to the single pair of outside baseline
meters. These ndings lend partial support to previous research, although Braithwaite et al. (2004) and
Wiseman et al. (2002, 2003) noted more differences in the geomagnetic elds, and found less variability
and magnitude differences with standard “mains” frequency EMF.
What is interesting is that both magnitude and variability of EMF “mains” frequency elds demon-
strated the most variability and magnitude despite a lack of EMF generating sources in the location. GMF
elds could be expected to vary simply as a function of environmental conditions of the earth, solar rays,
and other effects, but EMF is most often affected by articially created means. One explanation is that the
frequency of the elds during the investigation lay somewhere in between EMF and GMF frequencies,
thus increasing the magnitude readings of 60-Hz calibrated EMF meters. However, correlations between
EMF and GMF were essentially nonexistent, which rules this possibility out. As such, we have no expla-
nation as to why these EMF elds were greater and why they varied as they did.
EMF meters were attached to extremely sensitive coils, whereas GMF meters were not. One rea-
son why GMF may not have shown the magnitude differences that EMF did is that the GMF meters were
not sensitive enough to pick up that smaller variations in magnitude that the EMF meters could detect.
In either event, GMF still produced signicantly greater degrees of variability. However, despite smaller
differences from baseline measurements, GMF appears to predict anomalous events as well as EMF does.
From a broader perspective, the above ndings do generally support some of Persinger’s earlier
hypotheses (e.g., St. Pierre & Persinger, 2006). Even though specic frequencies of elds could not be
determined, there are denite differences in EMF and GMF elds above and beyond baseline measures.
As such, the EMF/GMF readings within a potentially haunted house suggest that these sites are a likely
place where a person’s perception could be affected.
The current research examined multiple areas within the site in comparison to baseline readings.
When the correlations are examined by session, a very intriguing nding seems evident. Correlations
across sessions demonstrated that readings in different rooms either correlated or inversely correlated
depending on location and time. The results of these correlations lead to some tentative conclusions about
the source and behavior of EMF/GMF in “haunted” locations. First, we accept the likelihood that one
eld or vector of EMF/GMF affecting two sets of meters that are physically distant from each other would
increase or decrease the readings at the same time and in the same way. If this assumption is theoretically
The Journal of Parapsychology
-
ent inside the house affecting different meters at different times. Given the quick decay of magnitude that
within the house would account for varying positive and negative relationships between meters. Degrees
location.
data. A closer inspection of Figure 1 and Table 4 demonstrates that different meter areas would strongly
correlate with other meter areas, but only at certain assessment times. At other times during assessment,
-
between meter areas would remain constant in terms of the pattern of correlations between meter areas
across all four sessions.
Thus, if this site is typical, an examination of the correlations across sessions suggests that EMF/
Rather, multiple sources of EMF of varying strength seem apparent. We would suggest that the correla-
and effecting different areas of the house as time progresses. This interpretation is at least partially congru-
However, as electricity sources were deemed nonexistent in the house, we have no explanation as to why,
Coded Analysis of EMF/GMF Spikes and Associated Phenomena
In terms of our principal hypothesis regarding the association of EMF and GMF with phenomena,
our results lend initial support to the idea that EMF and GMF are associated with captured nonhalluci-
natory phenomena that occur within purportedly haunted locations. The ratio of events occurring during
phenomena.
-
nomena, with a few caveats that will be mentioned below. Primarily, the demonstrated relationship be-
Persinger, 2006) hypothesis of GMF-stimulated hallucination. This, by no means, discredits GMF halluci-
nations, but it does provide evidence that would rule out an entirely neurological explanation of haunting
phenomena. Although the explanation of what these events represent is something we leave to others,
purportedly haunted locales, they are associated with EMF and GMF, and that internal sensation and per-
0 phenomena, which presents a quandary. Our goal for performing the analyses on Class 0 (ruled-out
events) and Class 1 (likely non-anomalous events such as bumps, thumps, and noises) was to provide
when spikes occurred over the data set. Thus, some sort of association exists between these “likely non-
anomalous” phenomena and EMF. Several potential explanations exist. First, particularly with Class 0
phenomena, many of the events were noises due to either airplanes overhead or cars passing by. Both of
A Critical Test of the EMF-Paranormal Phenomena Theory 231
elds. It may be the case that, as the meters are highly sensitive, the EM elds of these cars and planes
were being picked up. In a somewhat similar vein, Class 1 phenomena most frequently consisted of loud
bangs, bumps, and knocks that frequently occurred in the house, as well as voices that were more likely
to be investigators than EVP, or lights or dust that appeared to have odd trajectories. While interesting,
by EMPE criteria this type of phenomena was too easily explainable, and therefore relegated to Class 1.
Although better safe than sorry, some of these events may not, in fact, have been due to environmental
means, and thus the number of events categorized as Class 1 phenomena may have been inated due to
Type II error.
Regardless of the signicance of the ratios of Class 0 and Class 1 phenomena, they exist for com-
parison purposes. The remaining classes of phenomena, particularly Class 3, represent very closely ex-
amined and analyzed events that most people would interpret as “paranormal.” We are at a loss to explain
all of the Class 3 phenomena, and the percentage ratios of events occurring on spikes are highest with
this class. Events in this category included actual human-shaped shadows that we could not explain after
considerable effort, audio voices of children who were not present, sounds of footsteps and keys jingling
with investigators absent, and several instances of a male voice with a southern accent repeating what
appears to be “dee-dup.” Ratios of spikes for Class 3 events, which were much less common in the data-
set, ranged from 85–100%. However, despite other sources of EMF that may add error to the model and
false positives to the lower classications, the prediction rate of spikes to the events that were likely to be
anomalous, while not perfect, is strong enough to warrant the claim that spikes do seem to be associated
with closely examined anomalous events.
Increases, Decreases, and Variability in EMF/GMF and Length of Spike and Captured
Phenomena
Secondary analysis of EMF/GMF spikes overall was not signicant. Results showed that both
three-SD increases or decreases, as well as increases and decreases within a 1-second interval (i.e. vari-
ability), did not appear to differentially predict phenomena occurring in any of our phenomena categories.
Likewise, the number of serial spikes that occurred within 1 second (e.g., the length of the perturbation)
also did not seem to differentially predict phenomena events. However, our non-signicant ndings actu-
ally provide very telling information about EMF/GMF that, until now, has not been examined. Essentially,
nonsignicant results in these tests demonstrate that both increases and, more importantly, decreases from
average eld strength are associated with phenomena. These data suggest that any type of perturbation—
whether the eld increases, decreases, or varies positively and negatively—is a potential sign of anoma-
lous events.
Our lack of relationship between both the type of EMF/GMF spikes and the duration of spikes with
anomalous events is also useful for understanding how EMF/GMF behave in the context of anomalous
phenomena. The current research seems to suggest that very brief magnitude small spikes and longer sus-
tained spikes are equal potential predictors of anomalous activity. Practically speaking, this has relevance
to the hobbyists of ghost hunting as well as to parapsychological eld research. Meters used for either
ghost hunting or formal research must be both sensitive to small perturbations and able to sample EMF
quickly enough to detect spikes and decreases within a fraction of a second in order to test EMF as a pre-
dictor in the eld. Although speculative, the current research suggests that most of the inexpensive EMF
meters do not have the processing speed or sensitivity to capture EMF spikes that might predict quality
phenomena. It also suggests that the practice of ignoring small variations in EMF may mean missing or
not capturing a potentially anomalous event.
Weakness, Perspective, and Future Research
The methods used in the current study have some weaknesses, some of which are due to the nature
of EMF/GMF, and others which can be improved or are philosophically and scientically thorny. The most
232 The Journal of Parapsychology
obvious issue with the current research is a desperate need for replication. Although probability in many
ways protects the investigator from Type I error, the current research can make no claims that the eld and
phenomena behavior here is consistent either with the site itself or other locations. As such, our current
goal is to return to conduct additional investigations, implement software to better review evidence, and
make a comparison of other sites that have reliably demonstrated objective anomalous phenomena.
As to the association of EMF/GMF spikes in relation to phenomena, while every conceivable pre-
caution was taken with regard to the accurate estimation of the EMF/GMF equipment we used, the equip-
ment is by no means perfect. We hope to expand on our equipment in order to examine frequency or add
additional meters to areas in hopes of triangulating elds. Yet, the nature of these ndings cannot be dis-
counted by a perceived deciency in the equipment. If the reader can accept that the meters and data-log-
ging equipment employed register changes in EMF/GMF magnitude, and were applied, implemented, and
recorded in a consistent manner, then any error in the equipment cannot explain why registered spikes
(produced from data logging) would correspond to external events recorded in the environment. Even if
some internal error of the equipment was occurring, such as electrical feedback, it still would not explain
why particular feedback was occurring at a time that corresponded to the recording of external phenom-
ena.
Related to the above, general magnitude and variability readings regarding EMF/GMF may have
been different if coils had been available for all meters. Whereas three-standard-deviation spikes were
plentiful with GMF meters, overall magnitude readings and the number of spikes may have differed if
coils had been employed with GMF. Because these spikes were plentiful with GMF meters, we do not
believe that the EMF/GMF phenomena relationship demonstrated here was signicantly confounded by a
lack of coils for these meters, but we do intend to examine this possibility in future research.
Finally, and in a more theoretical context, we wish to state that the phenomena that have been cap-
tured are not intended to be considered “proof” of ghosts or haunting. Scientically and philosophically,
we prefer to make a clear statement that the captured phenomena were consistently evaluated only on their
ability to be easily explained. We leave the interpretation and personal meaning of such phenomena as
child-like female voices or shadow apparitions to the individual reader. Unfortunately, people will go great
lengths towards disavowing or accepting this type of research according to their belief systems, which in
many cases go above and beyond reasonable conclusions whether they are skeptics or believers. As such,
we feel the only reliable position to take is within the stated methods of evaluation that we employed to
determine the likelihood that these relationships exist.
Although the current research methods can always be improved, the ndings are strong enough to
demonstrate that a relationship exists between phenomena that are difcult to explain by environmental
means and changes in EMF and GMF readings, even when factoring in considerable degrees of error. A
denitive explanation for the occurrence of the phenomena, either by very unlikely but normal causes or
by supernatural means, is not something we can provide.
Conclusion
The overall research, despite some unavoidable weaknesses, provides initial evidence that video-
and audio-captured phenomena are associated with perturbations in the EMF/GMF elds of a “haunted”
location. Through careful classication of these events, and the independently measured variation in EMF/
GMF that is associated with them, we cannot help but conclude that a hallucination-based explanation for
haunting phenomena alone does not account for these ndings or phenomena. However, we openly claim
that this study represents initial ndings, and we are in the process of replicating this research at the same
location, as well as at other locations. We welcome theoretical contributions and other researchers’ interest
in these phenomena, as well as independent replication of these ndings.
A Critical Test of the EMF-Paranormal Phenomena Theory
References
between epistemic skepticism and relativism. Forum Philosophicum, 15
behavior. Administrative Science Quarterly, 47, 644–675.
Booth, J. N., Koren, S. A., & Persinger, M. A. (2005). Increased feelings of the sensed presence and
increased geomagnetic activity at the time of the experience during exposures to transcerebral weak
. International Journal of Neuroscience, 115
Braithwaite, J. J. (2004). Magnetic variances associated with “haunt-type” experiences: A comparison
using time-synchronised baseline measurements. European Journal of Parapsychology, 19
spontaneous strange experiences: The magnetic detection system (MADS). Journal of Parapsychology,
69, 151–171.
Braithwaite, J. J., Perez-Aquino, K., & Townsend, M. (2004). In search of magnetic anomalies associated
with haunt-type experiences: Pulses and patterns in dual time-synchronized measurements. Journal of
Parapsychology, 68, 255–288.
Braithwaite, J. J., & Townsend, M. (2005). Sleeping with the entity—A quantitative magnetic investigation
of an English castle’s reputedly “haunted” room. European Journal of Parapsychology, 20, 65–78.
experiences associated with sleep paralysis. Journal of Sleep Research, 8
perseverance. Social Behavior and Personality, 11, 57–60.
Estes Family History (n.d.). Retrieved from www.roots.com
Sleep and non-
. Dreaming, 19,
historical and contemporary poltergeist episodes occurred with sudden increases in geomagnetic
activity. Perceptual and Motor Skills, 62,
versus being correct. A meta-analysis of selective exposure to information. Psychological Bulletin,
135, 555–588.
Healey, F., & Persinger, M.A. (2001). Experimental production of illusory (false) memories in
reconstructions of narratives: Effect size and potential mediation by right hemispheric stimulation
International Journal of Neuroscience, 106
Journal
of Sleep Research, 17, 464–467.
effects in the classroom. Journal of Personality and Social Psychology, 50
Lorber, W., Mazzoni, G., & Kirsch, I. (2007). Illness by suggestion: Expectancy, modeling, and gender in
the production of psychosomatic symptoms. Annals of Behavioral Medicine, 33, 112–116.
American Psychologist,
62, 671–681.
Journal of Parapsychology,
64
Merckelbach, H., Van Roermund, H., & Candel, I. (2007). Effects of collaborative recall: Denying true
information is as powerful as suggesting misinformation. Psychology, Crime, & Law, 13,
The Journal of Parapsychology
neuropsychological aspects [Abstract]. Journal of Parapsychology, 62, 110–111.
bias in criminal investigations. Psychology, Public Policy, and Law, 15
and social contagion. Social Psychology, 40, 111–118.
female concept of self. Journal of Psychology, 137, 5–16.
History of Hancock County, Indiana: Its people, industries and institutions.
The poltergeist. New York: New American Library.
Roll, W. G., & Persinger, M. A. (2001). Poltergeist and nonlocality: Energetic aspects of RSPK [Abstract].
Journal of Parapsychology, 62
Roll, W. G., Persinger, M. A., Webster, D. L., Tiller, S. G., & Cook, C. M. (2002). Neurobehavioral and
neurometabolic (spect) correlates of paranormal information: Involvement of the right hemisphere and
International Journal of Neuroscience, 112
Journal of Personality and Social Psychology,
32
Sherwood, S. J. (2002). Relationship between the hypnagogic/hypnopompic states and reports of
anomalous experiences. Journal of Parapsychology, 66, 127–150.
St.-Pierre, L. S., & Persinger, M. A. (2006) Experimental facilitation of the sensed presence is predicted by
International Journal of Neuroscience, 116
College physics. New York: Worth.
Tsang, E. W., Koren, S. A., & Persinger, M. A. (2004). Power increases within the gamma range over the
frontal and occipital regions during acute exposures to cerebrally counterclockwise rotating magnetic
. International Journal of Neuroscience, 114
United States Federal Census (1880). Retrieved from http://www.ancestry.com
Wiseman, R., Watt, C., Greening, E., Stevens, P., & O’Keeffe, C. (2002). An investigation into the
Journal of
Parapsychology, 66
“hauntings.” British Journal of Parapsychology, 94
Acknowledgments
Part of this research was funded by the Student Government Association of Ivy Tech Community
College: Sellersburg, Indiana. The authors wish to recognize the ASAFP investigators who contributed
Mull. Thanks to Anthony Sams and Emily Bonden for assistance in previous versions of this manuscript.
Association for the Study of Anomalous Field Phenomena (ASAFP)
32 Beechwood Rd
Jeffersonville, IN 47130
blaythe@asafp.org
A Critical Test of the EMF-Paranormal Phenomena Theory
Abstracts in Other Languages
Spanish
UNA PRUEBA CRÍTICA DE LA TEORÍA PARANORMAL DE FENÓMENOS EMF-GMF:
INVESTIGACIÓN DE UNA CASA ENCANTADA SIN CAMPOS GENERADORES DE
ELECTRICIDAD
perturbaciones de EMF y GMF en el contexto de fenómenos potenciales recolectados con equipos
de registro de datos en una casa encantada sin electricidad. Los resultados globales indican que los
en comparación con las mediciones iniciales realizadas fuera de dicha ubicación. Las diferencias en la
magnitud de GMF fueron pequeñas en comparación con EMF. Las correlaciones mostraron que los campos
EMF/GMF cambiaron en rango y ubicación a lo largo de la investigación. Los resultados relacionados
asociados con los picos EMF GMF en serie, que los aumentos y disminuciones en los campos EMF/GMF
en serie no predicen diferencialmente los fenómenos.
French
les champs EMF/GMF changeaient en ampleur et en localisation durant tout le temps de la recherche. Les
236 The Journal of Parapsychology
German
EIN KRITISCHER TEST DER THEORIE ÜBER EMF UND PARANORMALE PHÄNOMENE: HINWEISE VON EINEM
SPUKORT OHNE ELEKTRIZITÄTSERZEUGENDE FELDER
ZUSAMMENFASSUNG: Frühere Forschungen über mögliche Zusammenhänge zwischen
elektromagnetischen und geomagnetischen Feldern (EMF und GMF) mit paranormalen Phänomenen
wurden unter der theoretischen Annahme durchgeführt, dass GMF-Felder Halluzinationen induzieren.
Die vorliegende Studie überprüft die Möglichkeit, dass nicht-halluzinatorische paranormale Phänomene
auch mit EMF/GMF-lFeldern zusammenhängen. Änderungen des EMF und GMF wurden im Kontext
möglicher Phänomene mit Geräten zur Datenerfassung an einem Spukort ohne Elektrizität aufgezeichnet.
Die Gesamtresultate weisen darauf hin, dass die EMF- und GMF-Felder sowohl in Grössenordnung
wie Schwankung innerhalb der Örtlichkeit signikant grösser sind—verglichen mit Baselinemessungen
ausserhalb der Örtlichkeit. Unterschiede in der Grössenordnung zwischen GMF und EMF waren
gering. Durch Korrelation konnte nachgewiesen werden, dass sich die EMF/GMF-Felder im Verlauf
der Untersuchung in ihrer Reichweite wie Lokalisierung änderten. Die Nachprüfung an ausgewählten
Phänomenen ergab, dass die Phänomene deutlich und signikant mit Gruppen von EMF- und GMF-
Spikes zusammenhängen, dass sowohl der Anstieg als auch der Abfall der EMF/GMF-Felder nicht zur
Vorhersage bestimmter Phänomene beiträgt und dass Anstiege in der Anzahl, d. h. in der Dauer von
Spikesserien die Ausprägung bestimmter Phänomene nicht vorhersagt.
CopyrightofJournalofParapsychologyisthepropertyofRhineResearchCenterandits
contentmaynotbecopiedoremailedtomultiplesitesorpostedtoalistservwithoutthe
copyrightholder'sexpresswrittenpermission.However,usersmayprint,download,oremail
articlesforindividualuse.
