NIC TECHNIQUES

• Document Type: CREST
• Collection: STARGATE
• Document Number (FOIA) /ESDN (CREST): CIA-RDP96-00788R001300140002-2
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 23
• Document Creation Date: November 4, 2016
• Document Release Date: April 5, 2000
• Sequence Number: 2
• Publication Date: January 10, 1980
• Content Type: REPORT

PY
r 2
d For Release 20MG@VI AG9AADA.-1"h4~~~1~~26
:41r-) 326-6200 ? Cable: SRI IN
07 CIA-RDR96-00
10 January 1980
Quarterly Progress Report
Covering the Period 1 October to 31 December 1979
SRI International Project 7560
CSR -4299
This document consists
of 29 pages.
Copy
 VV
Approved For Release 2000/08/97 : CIA-RDP96-007~8R001300140002-2
The objective of this program is to investigate the phenomenon of
The type of counter-
i
measures and factors that inhibit RV will be investigated. The work
effort will involve gathering data on specific geographic areas throughout
the world and examining research pertinent to improving the reliability
of the data obtained via the RV process.
Approved For Release 2000/01667 : CIA-RDP96-00 R001300140002-2
 Approved For Release 2000/08 07 : 9A-RDP96-00788 001300140002-2
II INTRODUCTION AND SUMI1ARY
Dr' CISSe 55 i ~ q
The purpose of this program is to provide a data bas
the area of remote viewing (RV). RV is the acquisition
and description, by mental means, of information blocked from ordinary
perception by distance or shielding, and generally believed to be secure
against such access. This includes the ability of subjects to view remote
geographical locations, even at intercontinental distances, given only
geographical coordinates or a known person on whom to target.
Investigations into the RV phenomenon at SRI International over the
past eight years have ranged from basic research with regard to proof or
the lack thereof as to the existence of the phenomenon, to
applications where existence of the phenomenon is taken as a given. The
present study, with its emphasis on application potential, leans toward
the latter--extensive proof of the phenomenon is not pursued here. A
measure of proof is provided, however, by the quality of results obtained
in' `tests carried out under double-blind conditions.
In this report we present the results of a several-month reliability
study. The purpose of this study was to delineate those factors which
appear to affect the reliability of the RV phenomenon, to develop a
methodology to minimize the deleterious effects of such factors, to test
that methodology in a training procedure involving several RVers, and to
evaluate the effects of such training on the basis of success in t
applications.
The factors affecting reliability, and the training methodology
designed to improve reliability are presented in Section III. In Section
III we also discuss the apparent impact on the training program ont
Approved For Release 2000/08/08 : CIA-RDP96-00_ZBBZP01300140002-2
 Approved For Release 2000/08/97 : CIA-RDP96-00748ROO1300140002-2
applications. The results of the study to date indicate that substantial
progress has been made. Finally, in Section IV we outline the potential
for a broad-based integration of RV phenomenA
Approved For Release 2000/ 8/07 : CIA-RDP96-007/88 R001300140002-2
 Approved For Release 2000/0q/07 : CIA-RDP96-0078PROO1300140002-2
The primary objective of this program is to provide a basis for
Of particular interest with regard to application is the use of an
abstract targeting procedure known as coordinate remote viewing (CRV),
a procedure we have had under investigation at SRI since 1972. In this
procedure the target site coordinates (latitude and longitude in e.g.,
degrees, minutes, and seconds) are relayed with no further information
to the individual who is to view the site. The remote viewer is asked
simply to proceed on the basis of the coordinates alone.
Admittedly, such an abstract targeting procedure seems without basis,
at least with regard to the present scientific paradigm. As a result we
can make no claim for the technique other than the purely pragmatic one
that it appears to work. It can only be pointed out that in psychoenergetics
research in general, the possibility of success in such a protocol is in
accord with an observed "goal-oriented" nature of the laws that appear to
govern psychoenergetic functioning.
In this section we discuss the findings of an investigation into CRV
reliability, the factors that affect it, the development of procedures to
improve it, and the results of application challenges to test it.
An investigation into the general problem of target acquisition has been
carried out an reported. See R. Targ, H. E. Puthoff, B. S. Humphrey,
and C. T. Tart, "Investigations of Target Acquisition," Research in
Parapsychology 1979, Scarecrow Press, Inc., Metuchen, NJ, 1980.
Approved For Release 2000/8/07 : CIA-RDP96-0078f R001 300140002-2
 Approved For Release 2000/08/0 :CIA-RDP96-0078 001300140002-2
A. Advances in Development of CRV Potentials
1. Background
Since the introduction of coordinate remote viewing (CRV) several
years ago, it has been apparent that CRV is often capable of yielding
highly accurate and useful data.
There are, however, several instances of failures, in which the
CRV description did not correspond to ground truth reality. Since one of
the program tasksl%is to "continue the investigations of methods to improve
the phenomena," a special study program was set up to attempt to determine
the factors that affect CRV reliability, and, to the degree possible, to
develop procedures to minimize the deleterious effects of such factors.
It was recognized at the outset that there were two facets of
the reliability problem that were of interest and would therefore have to
be addressed:
(1) Vertical Potential. Given that an individual exhibits
a demonstrable CRV ability, is it possible to develop
that ability beyond a neophyte status, that is,
increase the signal-to-noise ratio?
(2) Horizontal Potential. Does the CRV process possess
a structure sufficiently definable to imply a
meaningful construct for transfer/trainability
across a broad base of individuals, potentially
providing increased reliability on the basis of
cross correlation of multiple CRV responses?
Approved For Release 2000/0 07 : CIA-RDP96-0078 R001300140002-2
 Approved For Release 2000/08(07 : CIA-RDP96-00788F1001300140002-2
Results of the study program to date, described below, indicate progress
2. Signal/Noise Characteristics
The anatomy of the CRV phenomenon has been under intense scrutiny
at SRI in an effort spearheaded by remote viewer #002. These explorations
have centered about two areas:
(1) Observing and understanding the characteristics
of the noise.
(2) Observing and categorizing the characteristics
of the signals.
With regard to the noise aspect of the CRV channel, the process
of mapping out its characteristics has consumed a large part of a two-year
effort to isolate the factors involved. Four major categories of noise
have been delineated in this process. They are:
(1) Analytical Overlay. As the remote viewer becomes
aware of the first few data bits, there appears to
be a largely sponteneous and undisciplined rational
effort to extrapolate and "fill in the blanks."
This is presumably driven by a need to resolve the
ambiguity associated with the fragmentary nature of
the emerging perception. The result is premature
internal analysis and interpretation on the part
of the remote viewer. Example: An impression of
an island is immediately interpreted as Hawaii. To
circumvent this, a procedure for disciplined
rejection of premature interpretations and conclusions
is called for.
(2) Associational Overlay. In addition to provoking
premature interpretation, the incoming data bits
appear to stimulate pre-existing mental formations
that are associationally related to the target
material. Example: An impression of a round object
triggers an image of a favorite childhood ball. The
triggering of such associational overlays leads to
imaginative images that divert or embellish the picture
Approved For Release 2000//07 : CIA-RDP96-00788 001300140002-2
 Approved For Release 2000/08/07/ CIA-RDP96-00788R~01300140002-2
being built up from the incoming data bits. To
overcome the effects of this type of overlay,
training to recognize and discriminate against
associational images is required.
(3) Monitor Overlay. This is comprised of noise
intruding into the remote viewer's awareness
inadvertently as a result of undisciplined talk
or actions on the part of the session monitor.
Examples cover a broad spectrum, ranging from,
e.g., stimulation of sailboat images by a casual
pre-session discussion on sailing, to the subtle
reinforcement (e.g., by body language) of certain
responses that match the monitor's biases and
preconceptions as to the nature of target; in
short, any action on the part of the monitor
that degrades the remote viewer's attentiveness to
the task at hand. To bring this under control,
a standardized monitor behavior must be introduced
in which, for example, the monitor is restricted
to the use of certain standard phrases during his
monitoring of a CRV session.
Environmental overlay can be minimized by judicious
control of environmental factors, such as by providing
a quiet, dimly lit, relatively homogeneous monochrome
visual field absent of strong features and peripheral
clutter.
involve peripheral and subliminal perception of
environmental features, since, as is known from
study of subliminal perception, information not
processed at a conscious level can nonetheless
infiltrate perceptual and thought processes.
Environmental Overlay. This type of overlay has its
source in the physical surroundings of the CRV
session. Specifically, conditions of the session
chamber (e.g., obtrusive shapes, sounds, visual
highlights) are found to insinuate themselves into
the CRV response. A mundane example: an after-
image produced by a strong vertical line in the
session chamber can lead to a_predominant vertical
line in the "target" image. More esoteric examples
Although the latter overlays can be dealt with by controlling
elements in the environment of the remote viewer, the analytical and
Approved For Release 2000/08/07 : CIA-RDP96-0078PROO1300140002-2
 Approved For Release 2000108107f : CIA-RDP96-007888001300140002-2
associational "fill-in-the-blanks" overlays stem directly from cognitive
processes within the remote viewer. Our investigation of these overlay
patterns leads to a model of RV functioning shown schematically in
the reading of
Figure 1. With the application of the "stimulus" (e.g.,
a coordinate) there appears to be a momentary burst of "signal" that enters
into awareness for a few seconds, and then fades away. The overlays
appear to be triggered at this point to fill in the void. Success in
handling these complex processes apparently requires that a remote viewer
learn to "grab" incoming data bits while simultaneously attempting to
control the overlays. A strict and disciplined methodology to perform
this delicate and difficult task, involving repeated coordinate presenta-
tion and quick-reaction response, has been developed and is presently
being confirmed with four remote viewers; #002 who was primarily responsible
for developing the basic concept, and Nos. 009, 131, and 504 who are in
the role of trainees with regard to this particular methodology. The
procedure designed to minimize overlays coupled with use of a specifically
designed acoustic-tiled featureless room with homogeneous coloring to
minimize environmental overlay, and adoption of a uniform, limited monitor
behavior role to minimize monitor overlay, constitute the basic methodology
for noise reduction in our newly-developed approach to CR\'.
With regard to mapping the signal characteristics of the CRV
channel, a progressive multistage target acquisition process appears to
be emerging. The stages outlined in Table 1 appear to track an increasing
contact with the target site that takes place during the CRS' process.
An example of these stages of elaboration in a completely successful
remote viewing would be the series:
Land surrounded by water,
(1) Recognition and decoding of
major gestalts an island
Humid sensation, tropical
(2) Achieving sensory contact
with target feeling
Approved For Release 2000/08/07 : CIA-RDP96-00788R0p1300140002-2
 'pproved For Release 2000/08/07: CI 001300140002-2
UNCLASSIFIED
Approved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/08/071: CIA-RDP96-00788R
(3) Experiencing motion and
mobility within target
(4) Recognition and decoding of
minor signals while sustaining
major gestalts
(5) Decoding special characteristics
of target
(6) Analytical recognition and
decoding of significant aspects
of the target
Rising up, a panoramic
view
Mountains on the island, a
small port city on the
water's edge
Large areas devoted to
agriculture
Some tourism, agriculture
devoted primarily to sugar
cane, main island in Fiji islands
CRV STAGES OF TARGET ACQUISITION
im,
1. Recognition and decoding of gestalts
2. Achieving sensory contact with target
3. Experiencing motion and mobility within target
4. Recognition and decoding of minor signals while sustaining major
gestalts
5. Decoding special characteristics of target
6. Analytical recognition and decoding of significant aspects of the
target
Knowledge of the above multistage process of target acquisition
appears also to provide a filtering function, in that apparent data that
does not emerge somewhat in this order tends to be overlay (e.g., immediate
recognized image of the St. Louis Arch [Stage 61 as first response to
coordinate presentation).
For the training procedure, in which feedback plays an essential
role, a pool of several hundred target locations was prepared using
Approved For Release 2000/18107 : CIA-RDP96-007/'88R001300140002-2
 Approved For Release 2000/g8/07 : CIA-RDP96-0078$R001300140002-2
material from a large library of National Geographic magazines. The
coordinates for the chosen sites were obtained from the Second Edition
(Revised) of the Times Atlas of the World--Comprehensive Edition, Houghton
Miflin Co., Boston (1971).
In a typical training session, a half dozen targets are chosen
at random for use. In the early stages of the process the monitor makes
himself aware of the target material he is using so that he can provide
running feedback during the session. (We call this a Class C target
protocol.) During this phase the monitor is allowed only a few stock
phrases ("correct," "near the target," etc.) so as to minimize cueing and
leading. Once some degree of apparent competence has been reached, the
monitor is given targets to which he is blind (a Class B target protocol)
so as to eliminate confounding of the results by potential cueing. In
this case feedback information is accessed by both monitor and remote viewer
only at termination of the CRV session. A training series with a given
remote viewer generally consists of roughly a hundred of these trials
spread out over a two-month period. The output of 'a sample successful
trial is shown in Figure 2.
In detail, the training procedure is as follows:
(1) The remote viewer and monitor seat themselves at opposite
ends of a table, the former with a supply of paper and a
pen, the latter with target folders (contents initially
unknown) and reference atlases.
(2) The remote viewer is instructed that the monitor will begin
the CRV process by selecting a folder and reading aloud
target coordinates printed on the outside. The remote
viewer is to note down on paper any immediate impressions
(which he may also express aloud*) and then, rather than
embellishing on his first impressions, to ask for the
coordinates to be read aloud again so that the original
process may be repeated, etc., until a coherent picture
of the site emerges.
Approved For Release 2000(08/07 : CIA-RDP96-00J88R001300140002-2
 UNCLASSIFIED
9? 36'5
54' ~~' W
?s? 35.15
S,D 2?' al
~11
!Qved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
?S? WN
54? 22, k)
WaL" Joe.
FIGURE 2 CRV RESPONSE TO IGUAZU FALLS, BRAZIL TARGET (U)
UNCLASSIFIED
Approved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/08/p7 : CIA-RDP96-00788f2001300140002-2
(3) Following these instructions, the monitor selects a
folder and begins the process described above.
(4) After one or more repetitions of the coordinates
(each followed by a CRV response) leads to
recognizable target characteristics, the folder
is accessed by the monitor, and the atlas consulted
(if necessary) in order to.give feedback. In the
Class B protocol this is the termination of the
session. In the Class C protocol a line is drawn
on the remote viewer's data sheet to separate the
data thus generated from further data, since up to
this point the data were generated in a double-blind
protocol and can be objectively evaluated later as
a test of target acquisition.
(5) In the Class C protocol, having terminated the target
acquisition "test" phase, feedback can now be given
and/or further data solicited. The feedback given at
this point is non-negative, ranging from "correct,"
through "near the target," to "you are at another
target" (giving the remote viewer the benefit of the
doubt). The monitor then has the option of terminating
the viewing, asking for more detail ("there's something
ten miles to the north that should be visible") or
restarting the process when the viewer's original
description did not correspond to the target site.
(In the latter case the monitor can, of course, guide
or cue the remote viewer into a correct response.
This is acceptable in the non-test part of the sequence,
however, and provides an opportunity to investigate
whether such cueing procedures can be useful in
operationally oriented applications in which one
guides the remote viewer onto the target site with
... "f," and then asks for detail in
cues "a," "b,"
a nearby region, or concerning an unknown, "g.")
To give an example, we present here a summary of results obtained
with a remote viewer who was a relative neophyte with regard to CRV. He
was exposed to this protocol, a few targets per session, over a several-
day period, resulting in a data pool of 26 CRV target viewings. They
were: Salt Lake Desert, Utah; Lake Erie; Chicago; Mono Lake; Aruba Island;
Lake Okeechobee; Yount's Peak, Wyoming; Pitcairn Island; Pike's Peak;
Approved For Release 2000/08/0 ? CIA-RDP96-00788001300140002-2
 Approved For Release 2000/08/07/: CIA-RDP96-00788R401300140002-2
Los Angeles; Atlantic Ocean; Rio de Janeiro; Kansas plains; St. Peter and
Paul Islands; Randall Dam, South Dakota; Lake Titicaca; Cape May; Niagara
Falls; Munich; Amazon River; Midwestern plains; Venezuelan Peninsula;
Sierra Blanca Mountain; Oregon Desert; Panama Canal; Puerto Rico.
Following the first pilot session of five, in which essentially
immediate feedback was given (Class C protocol), the remaining twenty-one
were carried out with delayed feedback (Class B protocol) and thus provided
material that could be assessed objectively. As a first cut the targets
can be categorized roughly into five groups (mountains, flats, water,
cities, islands/peninsulas). The target/response matrix obtained in the
series is as shown in Table 2. The probability of such an alignment
DISTRIBUTION OF CRV TARGET/RESPONSE MATCHINGS
Transcripts
Targets
Mountains
Flats
Water
Cities
Islands/Peninsulas
Islands,'
Cities Peninsulas
occurring by chance alone can be calculated by a direct-count-of-
permutations method (see Appendix), and leads to p = 1/5: = 0.0083. The
distribution of responses is therefore statistically significant. Further-
more, beyond simple statistics, certain individual responses were excep-
tionally accurate during the acquisition "test" phase. In the final trial
14
Approved For Release 2000/01/07, : CIA-RDP96-007818001300140002-2
 go
Approved For Release 2000/08/07 : Cl 6-00T88,MOg1300140002-2
in this series, for example, when the target coordinates were for Guayama
in Puerto Rico, the viewer described a "fishing village on the southeast
coast of a boat-shaped island," which is an entirely correct description
of the locale at the target coordinates. He then drew an island, resembling
Puerto Rico in both shape and orientation.
A secondary application of the target pool/training mode procedure
is as an auxiliary calibration tools
prior, during, or after aA
task (which we designate
Class A protocols), a National Geographic CRV can be used to determine
whether a remote viewer is "on."
This procedure was used immediately following the third and
final scan of one of the operational CRVs described in the next section
of this report. Under Class B protocol (monitor blind to target, no
feedback during session) coordinates for the Sault Ste. Marie Locks in
the Soo Canal were given. The CRV response, shown in Figure 3, centered
on a channel with islands in it, leading to a large lake, and traversed
by a large white bridge, a result of high quality. Eventual feedback on
the 1target of interest revealed matching quality.
B. CRV.Applications
In this program SRI is charged with investigating
RV in order to provide data
Specifically, SRI has
been tasked' with examining a series of geographic coordinates using RV
techniques with the goals of:
(1) Establishing the authenticity and reliability
of the RV phenomenon.
r--
Approved For Release 2000/08/0
-RDP96-00788R9D1300140002-2
L
 144# 30t l.1
SY? 2(' W
UNCLASSIFIED
FIGURE 3 CRV RESPONSE TO SAULT STE. MARIE LOCKS IN SOO CANAL (Calibration Test) (U)
UNCLASSIFIED
Approved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/08/07/: CIA-RDP96-0078-qR001300140002-2
(2) Developing and refining experimental techniques and
understanding of the RV phenomenon.
(3) Establishing the best potential kinds of targets and
best potential use of the RV phenomenon.
In response to these requirements SRI has pursued application tasks
of interest' These tasks (Class A protocols) have
been pursued during the time frame in which the reliability-improvement
program of the previous section has been in effect. Therefore, the quality
of response to these tasks provides an,.,indirect measure of the efficiency
of the reliability-improvement procedures.
The CRV tasks described below were carried out in response to quick
reaction requirements-
progress of the work. During these scans all SRI personnel were kept
blind to the target. The tasks and associated response data are outlined
here in summary form. Complete documentation (transcripts, messages,
evaluation, etc.) can be made available4
? RV Session Dates: 22 June 1979 (Session 1);
5 July 1979 (Session 2).
? Remote Viewer: #009
? Interviewer:
Two scans were carried
(during a site
(Session 1);
out on a9 ',site designatedi
visit) to be.a target of interest.
Scan 1., Viewer #009 was closeted alone with ~--no other
personnel were present. The target site was designated by coordinates
only (latitude and longitude to seconds).4
Approved For Release 2000/0 IA-RD -00 R001300140002-2
 Approved For Release 2000/08 : CIA-RDP96- 888001300140002-2
An evaluation was provided
SRI
A second scan was then carried out with the same viewer
on 5 July 1979, targeting not by coordinate, but on the basis of familiarity
with the site established in Scan 1. Details can be made available through
separate channels.
? RV Session Dates: 12 July 1979 (Session 1);
17 July 1979 (Session 2).
It was arranged in advance that an SRI remote viewer would
attempt to target on a site designated only on the basis of the following:
On the day of viewing a( representative, known to the viewer, would be
carrying an envelope, inside of which were coordinates of a target of
s?
interest, location and function. unknown even to him. This was to con-
stitute an exercise in abstract targeting
Two scans were carried out on this basis on different dates.
The viewer's response centered on the description of a building of specific
design, together with information on internal layout and activities,
certain quite unique elements of which have been verified as being correct.
C1,
Approved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/0807: CIA-RDP96-0 R001300140002-2
? RV Session Dates: 14 December 1979 (Session 1);
17 December 1979 (Session 2); 18 December 1979
(Session 3); 21 December 1979 (Session 4); 28 December
1979 (Session 5).
Remote Viewers: #698 (Session 1); #002 (Sessions 2-4);
#009 (Session 5).
A total of five remote viewing sessions, involving three remote
a Targeting was on the basis of coordinates
viewers individually supplied pertinent, relevant data with regard to the
target site, and their data taken together resulted in a target/transcript
correspondence rating of 7 (given by user) on a 0-7 point evaluation scale
shown in Table 3.
The results generated in these 1 tasks to date, all
obtained with remote viewers incorporating the procedures developed in
the reliability-improvement program, appear to provide our first (and
encouraging) evidence with regard to a possible upgraded level of per-
formance. Further data needs to be generated, however, before a definitive
assessment can be provided, and this requirement will be pursued during the
remainder of the program.
"-D
Approved For Release 2000/08/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/0 07 CIA-RDP96- 0788R001300140002-2
0-7 POINT EVALUATION SCALE FOR TARGET/TRANSCRIPT CORRESPONDENCE
7 = Excellent correspondence, including good analytical detail (e.g.,
naming the site by name), and with essentially no incorrect
information.
6 = Good correspondence with good analytical information (e.g., naming
the function) and relatively little incorrect information.
5 = Good correspondence with unambiguous unique matchable elements,
but some incorrect information.
4 = Good correspondence with several matchable elements intermixed
with incorrect information.
3 = Mixture of correct and incorrect elements, but enough of the
former to indicate viewer has made contact with the site.
2 = Some correct elements, but not sufficient to suggest results beyond
chance expectation.
1 = Little correspondence.
Approved For Release 20/07 : CIA-RDP96-00788R001300140002-2
 Approved For Release 2000/08/0 -:-CIA-RDP96-00788RO01300140002-2
STATISTICAL PROCEDURE FOR FIRST-CUT ANALYSIS
OF CRV RESULTS
A precise measure of the statistical significance of a matrix of
target/transcript correspondences is given by a direct-count-of-permutations
method of great generality. It is an exact calculation method requiring
no approximations such as normality assumptions. Furthermore, the judging
process that went into generating the matrix is not required to be inde-
pendent transcript-to-transcript nor target-to-target. The only require-
ment is that no artifactual information is provided as to the order of
targets and transcripts. In particular, it can be shown that if targets
are used with replacement or are nonorthogonal (the case here), then the
method applies even in the case in which there is trial-by-trial feedback
and the target pool is known a priori to both remote viewer and interviewer.
Thus the possibility of interviewer cueing or subject guessing based on
a priori knowledge of the target pool is handled at a fundamental level
by a statistical procedure that assumes the worst. The argument is as
follows.
In the absence of knowledge as to which transcript was generated in
response to which target, one observes that in setting up the target-
transcript matrix there are n: possible ways to label the columns (tran-
scripts), given any particular order of the rows (targets), and vice versa.
Thus, there are n: possible matrices that could be constructed from the
C. Scott, "On the Evaluation of Verbal Material in Parapsychology: A
Discussion of Dr. Pratt's monograph," Jour. Soc. Psych. Res., Vol. 46,
No. 752, pp. 79-90 (June 1972).
27
Approved For Release 2000/08/r eIA-RDP96-0O78T001300140002-2
 Approved For Release 2000/08/07 :1CIA-RDP96-00788 R001'00140002-2
raw data, all of them equally likely under the null hypothesis that the
viewer's remote viewing attempts produce nothing but vague and general
descriptions and/or occasional chance correspondences with various target
sites. Each matrix has its associated sum on the matrix diagonal corre-
sponding to a possible alignment of targets.
The significance level for the experiment is then determined by
counting the number of possible matrices that would yield a result
(diagonal sum) equal to or better than that obtained for the matrix
corresponding to the key, and dividing by n; This ratio gives the
probability of obtaining by chance a result equal to or better than that
obtained in the actual judging process. For the results shown in Table
2 in the body of the report, for example, we find, by direct computer
count of the 5: matrices obtained by interchanging columns, that the
probability of obtaining equal or better matching by chance is
p = 1/5: = 0.0083.
Approved For Release 2000/Q8/01 : CIA-RDP96-00788) 001300140002-2