CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
67.
Final Report -- Objective E, Task 1
December 1987
218
A REMOTE ACTION INVESTIGATION WITH
MARINE ANIMALS
By:
EDWIN C. MAY
C. M. PLEASS
College of Marine Studies
University of Delaware
Prepared for:
PETER J. McNELIS, DSW
CONTRACTING OFFICER'S TECHNICAL REPRESENTATIVE
333 Ravenswood Avenue
Menlo Park, California
94025 U.S.A.
SRI
(415) 326-6200
Cable: SRI INTL MPK
TWX: 910-373-2046
Interpatenard For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
®

(
10
.
7
-
SRI International
C
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Final Report -- Objective E, Task 1
Covering the Period 1 October 1986 to 30 September 1987
December 1987
A REMOTE ACTION INVESTIGATION WITH
MARINE ANIMALS
By:
EDWIN C. MAY
C. M. PLEASS
College of Marine Studies
University of Delaware
Prepared for:
PETER J. McNELIS, DSW
CONTRACTING OFFICER'S TECHNICAL REPRESENTATIVE
SRI Project 1291
Approved by:
MURRAY J. BARON, Director
Geoscience and Engineering Center
SRI
International
333 Ravenswood Avenue . Menlo Park, California 94025 . U.S.A.
(415) 326-6200 . Cable: SRI INTL MPK . TWX: 910-373-2046
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
ABSTRACT
In FY 1986, SRI International awarded a subcontract to the College of Marine Studies of
the University of Delaware to conduct remote action (RA) experiments using marine algae as
target elements. Protocols were developed during that year that would enable SRI to test, with a
living system, the Intuitive Data Sorting model. During FY 1987, significant improvement was
made to stabilize the data so that standard analysis techniques (e.g., ANOVA) might be used.
While much progress was made toward that end, significant autocorrelations persist.
Nonetheless, an attempt was made to generate successful RA. SRI analyzed data for four
participants and found no statistical evidence of RA.
ii
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
TABLE OF CONTENTS
ABSTRACT
ii
LIST OF FIGURES
iv
I
INTRODUCTION
1
II
REVIEW OF THE UNIVERSITY OF DELAWARE REMOTE
ACTION PROJECT
2
A.
Toward Increased Data Stability
2
B.
Protocol for Data Collection and Analysis
3
C.
Results
5
D.
Discussion
5
REFERENCES
8
APPENDIX
A-1
iii
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
LIST OF FIGURES
1.
Results of the IDS and Monte Carlo Calculations
4
2.
Post Session Control Velocity Spectrum -- Operator 48
6
3.
Post Session Control Velocity Data -- Operator 48
6
iv
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
I INTRODUCTION
For a number of years, Dr. C. M. Pleass of the College of Marine Studies at the University
of Delaware, has been constructing a remote action (RA) experiment using various species of
marine algae as targets. In these types of experiments, RA participants attempt to change the
swimming velocities of single algae cells during effort periods as compared to control periods.
Encouraged by a paper presented at an annual meeting of the Parapsychological
Association, 1* SRI International awarded a two-year contract to the University of Delaware to
satisfy Objective E, Task 1 of the FY 1986 statement of work. For the first year, the task was to
stabilize the equipment and protocol, while the task for the second year was to collect RA data.
This report describes SRI's role in that effort and summarizes the work done by Dr. Pleass. t
(The University of Delaware's draft final report is included, verbatim, in the Appendix.)
* References may be found at the end of this report.
t This report constitutes the deliverable for Ojective E, Task 1.
1
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
II REVIEW OF THE UNIVERSITY OF DELAWARE
REMOTE ACTION PROJECT
In the past, the putative RA data from the College of Marine Studies of the University of
Delaware, have been criticized for the statistical methods that were employed. The cell velocity,
as a function of time, exhibited an unknown mixture of biological and environmental correlated
responses (i.e., the data points were not statistically independent). In a report included in SRI's
FY 1986 interim report on this experiment,2 a significant autocorrelation was observed, yet there
was no modification of the statistical analysis to account for the correlation. (The complete FY
1986 results can be found in Reference 2.)
During the first few months of FY 1987, the University of Delaware moved the project to
new and better facilities. The move delayed our joint FY 1987 project; however, as is shown
below, a major improvement of data stability was realized.
Initially, the experimental setup in the new location was identical to the old, but with the
following improvements:
. The floor of the new laboratory is a 3-foot thick concrete slab. This improved
the physical isolation of the apparatus.
. The apparatus was installed in an electrically isolated and sound attenuated
room.
. It was now possible to collect data for 24-hour periods.
SRI personnel inspected the new facilities and found them to be significantly improved. What
follows is a description of the FY 1987 activity.
A.
Toward Increased Data Stability
Baseline velocity data were collected over 24 hours for fifteen 24-hour periods. It was
clear (see Figure 2 in the Appendix) that the cells exhibited erratic behavior during the "day"
(0800 to 1700) but were quiescent between 0000 and 0800. This quiescent period suggested that
the organisms might be sensitive to the increased electromagnetic activity during the day and
might be less stimulated during the night. (SRI analyzed a quiescent data set and found the data
to be normally distributed with a slight tail favoring downward swimming.)
2
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
A second possible contribution to the data instability was noticed. The algae cells tended
to clump, especially if they had been in the experimental chamber for some time. If this
tendency was consistent, it would produce strong autocorrelations in the data.
The solution to the first problem was to attempt to provide better shielding, but it was
realized that ELF shielding was not possible. Instead, a completely new experimental chamber
was designed.
In the old system, cells, which were drawn once per day from a biostat, constituted a fresh
experimental sample daily. While it was good to have fresh cells, the immediate change in
environment led to initial instabilities and, in the absence of light and nutrients, the cells lost
energy over time. This trend was obvious from the data in that cells showed an increasing
tendency to swim with gravity rather than against it.
The new system uses a continuous flow design. Fresh cells flow through an experimental
region at an orthogonal (to the measurement axis) velocity of ~150 mm/sec. This change greatly
improved the stability, and significantly reduced the clumping.
B.
Protocol for Data Collection and Analysis
With the increased stability, it was decided to use two types of analysis:
(1) Standard Analysis of Variance (ANOVA).
(2) Data Analysis sensitive to the Intuitive Data Sorting (IDS) model.
The standard ANOVA 2 x n consists of a design where n is the number of consecutive temporal
data samples collected in each of the effort and control conditions. The IDS analysis was
outlined earlier,2 and a complete description can be found in SRI's final report on Objective E,
Tasks 3 and 4.3
Briefly, the IDS formalism uses a simple application of the central limit theorem to show
that for mean chance expectation (MCE), the expected value of the log of the absolute value of
the detrended velocities is linearly related to the log of an averaging length, n. Under RA
conditions, the relationship is nonlinear in a dramatic way (Figure 1). Under IDS conditions
(i.e., psychoenergetically biasing a sampling distribution) the formalism predicts a linear
relationship once again, but with increased intercepts over MCE. The averaging length, n, is
defined as the number of samples of the parent velocity distribution that are averaged to form
one data point.
3
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
5
Monte Carlo
Theory
Note: Only a few points
are shown for clarity.
1
-
0.5
RA (a =. 0.30)
0.1
1.
0.05
IDS (I=0.38)
MCE
0.01
1
1
5
10
100
500
Averaging Length (n)
FIGURE 1 RESULTS OF THE IDS AND MONTE CARLO CALCULATIONS
SRI conducted extensive Monte Carlo calculations to simulate the experimental IDS
design. The results are shown in Figure 1. The RA curve was calculated assuming a parent
distribution shift of 0.30 -- the only free parameter. The IDS curve was calculated assuming that
the sampling distribution had an increased variance2 of 0.38. Only 4 of the 12 generated data
points (3 for each model) are shown. All points fell within 1o (the error bars) of the theoretical
curves -- an expected result considering that each model has only one specified free parameter.
SRI proposed that 50 trials for each of three averaging lengths (5, 15, and 40) be collected
by one RA operator (blind to the averaging length).
Delaware proposed that the actual data collection proceed as much like their "wave"
protocol as possible. In a single trial, an operator marks a continuous, but undisplayed velocity
data stream when he/she feels that RA has just occurred. This effort period is followed
immediately by a control period of length n points. This is followed by a brief, but random,
interval before a second identical period to the above couplet is collected as a pseudo-effort
period.
In addition to the psychoenergetic trials, SRI requested that a "real" influence be exerted
upon the algae under the same collection format described above. The data would be used to
4
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Change in Velocity (Normalized)

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
demonstrate that if RA behaved like a "real" influence, the IDS formalism would be capable of
describing it:
C.
Results
Four operators produced five trials each at each of the three averaging lengths. This raw
data and one set of "real" influence (spark stimulus) data simulated for five trials was sent to SRI
for analysis.
The "wave" protocol used by the the Delaware group suggested that the ANOVA be
changed to a 2 x 2n design. A separate analysis was then performed for each of the averaging
lengths. The first row of the ANOVA data matrix consisted of the five-trial averages of the
velocity (detrended) for each of n points during the effort period (i.e., just before the operator
marks the data stream) and a second set of n control points. The second row was identical to the
first row, except the data were taken from the pseudo-effort period.
This particular design was sensitive to the "release-of-effort" wave protocol and to
differences between the effort and pseudo-effort periods. The ANOVA was conducted for all
three averaging lengths for the four operators and for the spark stimulus data. There was no
evidence even suggestive of an extra-chance anomaly.
Delaware made two attempts at simulating RA. The first used spark induced E&M as a
perturbing force, however, the intensity was insufficient to produce any effect. Following an SRI
suggestion, Delaware used visible light to induce a change in swimming velocity. SRI was unable
to confirm a stable velocity shift in the data from this attempt as well.
D.
Discussion
Two conditions must be met before an ANOVA is valid. The data must be approximately
normal and they must be statistically independent. ANOVA is particularly forgiving for the first
condition, but if a positive serial correlation exists, a significant underestimate of the residual
variance can result.4 This can result in an artifactually significant F ratio.
Figure 2 shows a typical velocity distribution with a Gaussian fit and a Gaussian fit with a
matched exponential tail. It is clear by inspection and the value of chi-square that the first
criterion (approximate normality) is easily met.
The second criterion is much more important from an ANOVA perspective. While the
autocorrelation has improved greatly since the earlier report,2 it nonetheless, still represents a
5
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
7

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
high degree of significance -- especially for the early lags. This can be seen in Figure 3 as low
frequency oscillations of the velocity envelope. Even after the significant improvement of the
system, some non-random component to the velocity data remains. Because the overall
ANOVA result was not significant, and the autocorrelations are all positive, we are not required
to calculate the appropriate error residuals for our correlated data.
1000
· Data Points
- Gaussian Fit
- - Gaussian + Tail
Velocity Density (arbitrary)
C
002
9
-
-40
-30
-20
-10
0
10
40
Velocity (micrometers/sec)
FIGURE 2 POST SESSION CONTROL VELOCITY SPECTRUM -- OPERATOR 48
120 100 160
1
Pelocity (mm/sec)
=
.
-
H
0
1000
8000
4000
5000
Time (-sec)
FIGURE 3 POST SESSION CONTROL VELOCITY DATA -- OPERATOR 48
6
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
The Delaware efforts were quite disappointing. SRI designed the protocol and analysis
and made numerous suggestions about improving the physical set-up. Virtually all the RA data
were collected in the last month of the contract, so that only 5 trials instead of the requested 50
were collected at each averaging length. Second, the operators were not blind to the averaging
length; thus, even if we had observed significant evidence of an anomaly, we would be unable to
interpret it with regard to the IDS model.
Why the RA simulations (spark data) were not significant remains a mystery. Either the
wrong stimuli were chosen by Delaware, or the stimulus intensities were incorrect.
7
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
REFERENCES
1.
Pleass, C. M., and Day, N. D., "Using the Doppler Effect to Study Behavioral Responses
of Motile Algae to Psi Stimulus," Proceedings of the 28th Parapsychological Association
Convention, pp. 373-406 (August 1985).
2.
May, E. C., Humphrey, B. S., and Pleass, C. M., "Measuring Remote Action Influence
on the Vertical Component of Dunaliella Velocity," Interim Report, Objective E, Task 9,
Project 1291, SRI International, Menlo Park, California (December 1986).
3.
May, E. C., "Intuitive Data Sorting: An Informational Model of Psychoenergetic
Functioning," Final Report, Objective E, Tasks 3 and 4, Project 1291, SRI International,
Menlo Park, California (December 1986).
4.
Neter, J., Wasserman, W., and Kutner, M. H., "Applied Linear Statistical Models,"
Second Edition, Homewood, Illinois, Richard D. Irwin, Inc., p. 445 (1985).
8
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
APPENDIX
A-1
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
1
Final Report on the Stanford Research International Project.
"Evaluation of the Theory of Intuitive Data Sorting, and Its
Application to Bio Laser Doppler Data Generated by Marine
Microorganisms"
Duration:
Two Years from 10/1/85
Program Leader:
Dr. C. M. Pleass
College of Marine Studies
University of Delaware
Newark, Delaware
19716
The following description is excerpted from the original proposal,
so that this final report may be placed in proper perspective:
Year One - 10/1/85 - 9/30/86.
Using the existing Bio Laser Doppler facilities at the Bayside
Laboratory, develop a protocol and statistical analytical procedure
which can effectively address the question of the validity of the
SRI theory known as Intuitive Data Sorting (IDS).
The deliverable, due by 9/30/86, would be a report describing
a mutually acceptable experimental and analytical technique. This
implies frequent contact with the SRI project monitor throughout
the year, as the various options are examined.
1
Year Two - 10/1/86 - 9/30/87.
*
Using the protocol and procedure developed in Year 1, to
carry out Bio Laser Doppler experiments that address the validity
of the IDS theory. This will require the involvement of volunteer
participants from the Lewes community, whose task will be to try to
intuit or anticipate changes in the statistical parameters
describing the motion of marine microorganisms, or in some similar
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

...
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
2
way, to provide evidence bearing on the validity of the SRI theory.
The individuals involved will not be identified in the report by
name, and the appropriate informed consent will be obtained in
each case, to meet the University of Delaware's human subject
requirements.
The deliverables will be a report containing a description
of the detail of each trial, including statistical ensembles that
best represent the data, and copies of the floppy discs carrying
the raw data.
*
In practice, the project progressed in two cycles, both of
which involved program development and experimental psi work. Thus
in Year 1, two appropriate psi protocols were developed (pk 85 and
WAVE), and volunteers from the Lewes community used them to create
data bases which SRI and UD examined together1. The experimental
and analytical techniques appeared acceptable, a priori, but
instability in the baseline time series velocity data threatened
the credibility of the results.
1
.
Very few segments of our time series records of unstimulated
algal movement taken prior to the fall of 1986 were free from some
abrupt changes in average velocity and vector.
However for
practical reasons most of the data had been acquired during the
1 Algal swimming velocity, resolved along a single axis,
constituted the primary experimental variable.
The variance of
this data and the direction of movement, up or down, were secondary
variables.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
3
working day.
.
This suggested 24-hour time series studies of
unstimulated cultures, to examine the possibility that the observed
changes were part of a natural circadian rhythm. After modifying
the system to allow 24 hour studies, fifteen 24-hour data sets
were collected in November - December 1986. On January 13, 1987,
Ed May and Peter MoNelis of SRI met with Mic Pleass and Dean Dey
at Lewes to discuss this data. A typical time series is shown in
the fold-out (Fig. 1).
A "window" was clearly evident in all 15 data sets between
circa midnight and eight AM. (Fig. 2)
The data in this region
were quiet and would have formed an acceptable basis for psi
experiments. However, it seemed unlikely that participants and
experimenters would show much enthusiasm for working through this
period.
.
As a secondary benefit, the data gave us perspective on the
hours 0900 - 1500 which include most of our previous psi runs. It
seemed clear that this was a period when the algae were generally
swimming non-randomly, with quite frequent abrupt changes in average
1
velocity. As we were all aware, this places a heavy stress on the
analytical technique and the credibility associated with the data
from psi experiments. It is impossible to choose between exogenous
or endogenous (circadian) influences when looking for a cause.
However, the group were suspicious of sonic and electromagnetic
disturbances associated with the working day.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

24-HOUR DATA
TEST850
24-HOUR DATA
TEST850A
24-HOUR DATA
TEST850B
24-HOUR DATA
TEST850C
0
-50 -
-100 -
-150 -
-200-
-250-
16:33:42 17:20:30 18:08:07 18:58:01 19:49:42 20:39:41 21:28:38 22: 16:53 23:05:31 23:54: 15 00:42:58 01:32:00 02:24:06 03: 16:41 04: 10:29 05:04:42 05:57: 14
TIME
TIME
TIME
TIME
24-HOUR DATA
TEST850D
24-HOUR DATA
TEST850E
24-HOUR DATA
TEST850F
24-HOUR DATA
TEST850G
-50 -
-100
-150
-200 -
-250 -
05:57:14 06:48:00 07:36:35 08:24:31 09:12:17 09:59:51 10:48:08 11:35:43 12:23:54 13:11:41 13:59:18 14:47:06 15:35:05 16:22:49
TIME
TIME
TIME
TIME
.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
int.
Fig. 2
TIME SERIES VELOCITY DATA
Vertical lines are at
midnight and 0800 hours
Approved For Release 2000/08/10 CIA-RDP96.00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
6
The environmental chamber housing the BLD apparatus was
converted into a modestly good RF and sonic quiet room. The results
were unchanged. The next step was to actively look for evidence
of group dynamic change within the algae in the tissue culture
vessel used to hold the sample. By visually observing a particu-
larly hardy clone of Dunaliella which lived for three months on
Mic Pleass' .office window sill (unnurtured through the period, and
knocked over several times!) we were able to identify "micro
avalanches" of cells, superimposed on upward and downward movement
of the green suspension. Literature search revealed one relevant
reference2 which provided examples of microavalanching in cultures.
of Dunaliella. This seems to be a natural method of redistributing
cells which have physical reasons for normally swimming upward.
To work around this problem a flow system was constructed,
so that a well mixed culture from a biostat external to the
environmental chamber could move in slow (100-200 um sec-1) laminar
flow past the measuring volume.
The virtual fringes of the BLD
system were aligned parallel to the flow, ie, the axis along which
1
velocity measurements were made was chosen to be at right angles
to the artificially induced flow. Experiments confirmed that these
velocity data were unaffected by the slow flow. Twenty-four hour
runs appeared much quieter.
Ed May visited Lewes again on May
5th, just at the right time to see a very good looking time series
2 Kessler, J. O., Cooperative and concentrative phenomena of
swimming microorganisms. Contemp. Phys. 1985, Vol. 26, #2, 147-166.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Fig. 3
24 hour velocity data from Dunaliella
24-HOUR DATA
FLOW TEST 004
24-HOUR DATA
FLOW TEST 004
runa.
0
i'm
-
-40
-60
-80
-
-100
-
VELOCITY
-120
-
-140
-160
-100
-200
-
-220
-240
-2GO
0
2
+
6
8
10
12
14
16
18
20
22
2
4
6
8
10
12
14
16
TIME OF DAY
TIME OF DAY
rize
for best looking data,
but see
Summary stats from
the first six hows +
2
page 7
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Mays
digest above
back
Another duf from
Den runs : peristaltice
adenser, Pumping
resevoir, e head.
canoter overflas
istant secondary

1
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Neaver
namal
-
Variable:
VELO
VEL1
VEL2
Sample size
1658
1757
1890
Average
-162.039
-162.306
-162.547
Median
-162.312
-162.312
-162.312
Mode
-161.547
-163.078
-163.078
Geometric mean
Variance
39.0124
37.563
36.1619
Standard deviation
6.24599
6.12886
6.01348
Standard error
0.153394
0.146216
0.138323
Minimum
-185.281
-184.515
-182.219
Maxi mum
-127.859
-137.813
-141.641
Range
57.4219
46.7029
40.5785
Lower quartile
-166.14
-166.14
-166.907
Upper quartile
-157.719
-158.484
-159.484
Interquartile range
8.4217
7.6563
8.4226
Skewness
0.190437
0.197301
-0.0216271
Standardized skewness
3.16568
3.37628
-0.383844
Kurtosis
0.553595
0.325558
-0.0216545
Standardized kurtosis
4.60128
2.78553
-0.192164
Variable:
VEL3
VEL4
VEL5
Sample size
1857
1926
2096
Average
-163.34
-163.527
-163.456
Medi an
-163.078
-163.844
-163.844
Mode
-162.312
-161.547
-163.844
Geometric mean
Vari ance
38.1946
41.748
37.9794
Standard deviation
6.18018
6.46127
6.16274
Standard error
0.143415
0.147228
0.13461
Minimum
-196
-189.109
-181.453
Maximum
-139.343
-140.875
-139.343
Range
56.6564
48.2338
42.1094
Lower quartile
-167.672
-167.672
-167.672
Upper quartile
-159.25
-159.25
1
-159.25
Interquartile range
8.4217
8.4217
8.4217
Skewness
0.112429
0.0842093
0.167328
Standardized skewness
1.97791
1.50873
3.12743
Kurtosis
0.561965
0.316777
0.0616489
Standardized kurtosis
4.94322
2.83777
0.576123
8
Summary
Test 004
00:00 - 06.00
06:59
Repared from
raw data on
5/14 .
Show awesome
variations in
atd . skewness and
Kartais +
Means and os
are indeed & otable
but this ain't
normally dick.
data!
Why ?
0
Could be the
arbitrary treatment
of pm volks v
gain, explored
on ev. of 5/13.
Therefore switch
To a study of
data Jam that
file.
Fig. 3A
Summary statistics of one hour periods
from data set 004
· Could be the
"windows" we
have been uten
Coutoff at 500m
but it's seems
unlikely , even
though the skew
Is generally toward
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

lest 004 cant
9
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Variable:
VELE
Sample size
2202
Average
-163.21
Med: an
-163.078
Mode
-163.078
Geometric mean
Variance
38.9447
Standard deviation
6.24057
Standard error
0.132989
Minimum
-183.75
Maximum
-139.343*
Range
44.4066
Lower quartile
-167.672
Urper quartile
-159.25
Interquartile range
8.4217
Skewness
0.0855942
Standardized skewness
1.63975
Kurtosis
0.171419
Standardized kurtosis
1.64196
THERE
SOME
INSTRUMENTAL ARTIFAC
OR SAMPLING ARTIFACT
WHICH WE ARE STILL
MISSING ?
Fig. 3B
Summary statistics from 004 continued
.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
10
of 24 hour velocity data (Fig. 3).
The stability
suggested
stochastically stationary data and a chance to use ANOVA'S
with
repeated measurement, an analytical technique which Ed May had
recommended and described in detail. Unfortunately a more careful
examination of the data showed that while the mean velocities were
constant to better than 1% throughout, the higher statistical
moments varied all over.
Figs. 3A and 3B give the details:
the
handwritten marginal notes reflect our thinking at that time.
The
same general pattern prevailed in repeat runs, and it was evident
that these data would not be suitable for a pk vs IDS analysis.
At this point we acquired a new SD382 0-40KHz signal analyzer'
for the BLD spectroscopy project. Using it, we began to search for
artifacts and instrumental problems.
Studying the oscillators in
the Dantec tracker, we found that the shift frequency which Dantec
had quoted as 5.500 KHz drifted from 5.56X to 5.58X KHz. The time
scale of the drift was much longer than that of a set of psi trials.
However, we modified our procedures to calibrate, prior to each
set, largely because this insured the correct sign on the data (up
1
At the same time, we upgraded the precision of
or down motion).
our raw data by changing the data processing route from the 8-bit
Dantec tracker to a 12 bit A/D board. The overall result was to
increase the precision of the velocity measurements by a factor of
16. Also, percent positive (or percent swimming downward) became
a useful, unambiguous, binomial statistic which could be examined
for psi-induced changes independent of the value of the velocity.
.
#
SIRPROJ
CMP/vs
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
11
Fig. 4 shows overnight data from the system.
This general pattern
was replicable, and little different from that obtained earlier
with our less accurate apparatus. On inspection, some portions of
the time series data appeared relatively normally distributed.
Most showed widely varying skew and kurtosis.
We hope that this
kind of biologically active data will help SRI with their pk vs
IDS studies, but we have serious doubts.
We are confident that
the data reflect what is really going on in the culture.
Fig.
4 shows a layout of the environmental chamber and
laboratory.
Two sets of experiments were conducted with the
intention of providing the best possible data for SRI to use for'
hypothesis-testing:
Set 1
psi runs using a modification of the WAVE protocol3,
with sets of five trials at each of three sequence
lengths.
5, 15 and 40 averages of 100 raw data were
.
chosen after discussions with Ed May4
Set 2
parallel runs in which a 5000 VDC spark source within
the environmental chamber provided a low level "real"
3 Pleass, C. M. and N. D. Day, "Behavioral response of marine
microorganisms to psi stimulus:
statistical analysis of data from
Dunaliella," Proc. 29th Annual Conv. Parapsych. Soc., Sonoma State
Univ., CA (5-9 Aug. 1986).
4 SRI originally asked that the duration of the trial be
hidden from the participant, i.e. all would appear to have the
duration of the longest. CMS missed this point and apologize. If
the data are useful in other respects, we would be pleased to repeat
the sets at no cost to SRI.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
24-HOUR DATA
wave441
24-HOUR DATA
Wave441
100
80
80
-
70
60 -
VELOCITY
-
56 -
40
30
20
10
0 -
0 2
4 6
8
10
12
14
16
18
20
22 24
2
4
6
8
10
12
14
TIME OF DAY
TIME OF DAY
Fig. 4
Overnight data from the flow
system after upgrading to
12 bit data transfer
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
12

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Fig. 5
Lay Layout for SRI PSI Experiments
DOOR WITH GLASS PANEL
-
OPERATORS
CHAIR
MONITOR AND KEYBOARD
Experimenters
Laboratory/ Office
Space
LASER
DOPPLER
SYSTEM
IN
VIBRATION
ISOLATION
ON
OPTICAL
BENCH
BIOSTAT
ENVIRONMENTAL
CHAMBER
EXTERNAL
BENCH-TOP
-
EXTERNAL DOORS
CLOSED
L
1
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
13

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
14
(sonic, optical and E.M.) stimulus for comparative
statistical analysis.
Format for the Experiments
The "WAVE" protocol is a technique for examining the release
of psi pressure. Its general characteristics are:
Continuous time series data beginning several hours
prior to the psi run.
Participants "warm up" at their own pace, prior to a
trial.
The participant makes one button push, to mark the end'
of the psi period.
A space between trials at least 4X the data content of
the preceding psi period.
Detail of the modifications used to carry out the SRI work
follow:
1)
Electronics are turned on the evening before a run. If
overnight data is to be collected, three hours will
1
subsequently be deleted from the time series to cover
the warm-up and stabilization of the oscillators.
2)
Data routing: Photomultiplier + tracker, for removal
of 40 MHZ shift + 12 bit A + D conversion + PC +
identification of status (predata, psi, control, etc.)
by a file marker.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
15
3)
Data processing:
PC + raw velocities and up/down
information (± sign on the raw data) + means of 100
raw velocities, and percent moving down. Store this.
·
One line of the file will then read:
Record #, time,
av. velocity, std. deviation, % positive, Marker
4)
Definition of file markers:
0 - unassigned
666
Button-push at end of psi period
888
Historic, pseudopsi and pseudo control
777
markers of no utility for SRI work
5)
Approximately one hour before the participant arrives
the experimenter will break the predata collection,
check the actual shift frequency using the SD380, and
restart the control program with the new value added.
(Normally changes are very slight after the warm up
period, and the time for detectable change is long
compared to the duration of a set of psi trials). The
participants' ID number, data, and the initial number
of points per trial will be added at this time. The
screen will be left ready for the participant with the
instruction
Press F1 to mark end of trial
Trial X coming up!
6)
The participant will be seated by the keyboard and
monitor outside the environmental chamber. They can
be observed by the experimenter through a glass panel
in a connecting door (Fig. 4).
They have been
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
1
16
encouraged to take whatever time they personally feel
that they need to get into a psi-conducive state prior
to each button push. They are motivated to keep this
period short, because we ask that they perform at least
five trials at each of three sequence lengths.
When
they become aware of a period of peak performance they
are instructed to press F1 (which marks the data file)
and abruptly break out of the psi mode'. The monitor
screen turns from grey to blue at the button press.
During the blue screen interval they are asked to read
a dictionary. When the screen returns to grey they may
try again.
7)
After five or more trials at the first sequence length
(500 raw data points), the participant uses a menu to
change to the next sequence length. A 5 x 500, 5 x
1500 and 5 x 4000 experiment normally takes on the
order of 2-2 1/2 hours.
Discussion of the Experimental Format
. There may be an inevitable restriction on producing BLD data
for a comparison of psi and IDS hypotheses. Our observation is
that the psychological state of the participant tends to change as
the set progresses. They may become more psi-competent or less,
and we can see no way to quantify this or eliminate its effect,
5 We feel that the act of becoming aware of deep involvement
is in itself a sign that the period has ended.
SIRPROJ
CMP/vs
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
17
unless a large data base were developed from sets where sequence
length was chosen at random.
We can do this; however, if the
identity of the trial is to be hidden from the operator by
constantly using the longest spacer, it may be desirable to reduce
the length of the longest sequence. Our experience to date suggests
that 100 trials at a sequence length of 40 would take circa 35
participant hours.
. No immediate feedback was provided during the trials whose
results are reported below. Our participants do not like watching
the monitor during the run, and we have not yet found an appropriate
way of reflecting their performance.
After the experiment is
complete the participant is debriefed.
The experimenter notes
their comments and adds his own.
As soon as we can settle on a
"real time" analytical procedure that we can use to present the
results of the experiment we will do so, but at present the
analytical procedure requires transferring the time-series data to
Lotus for processing.
Results are mailed to the participants,
usually within one week.
1
. Although the participant is aware of the progression to longer
sequence lengths this does not directly affect the length of time
they spend working into a psi-conducive state.
The reason that
everyone gives for preferring the WAVE protocol is that they need
not "work by the clock". The time series data which reflect their
effort generally contains several times the number of data points
required for the analysis.
SIRPROJ
CMP /VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
18
.
Results
Four participants completed sets in the format previous
described. One set of data was developed in the same format using
a distant spark as a low level "real" stimulus.6 The experimenter
switched on the spark when the grey screen appeared ( the equivalent
of beginning to get into a psi-conducive state) and switched it
off again one minute later, pressing F1 at the same time.
The
various files are identified in the cover letter which was shipped
to SRI with the data on 9/14/87, and a copy of this is attached for
completeness as Appendix 1.
We routinely check the summary statistics from random samples
from pre-data strings once the data have been transferred to Lotus.
Since the raw data have been shipped to SRI we will only report our
general conclusion:
that very few sample periods contain data
which even approximate a normal distribution. The 8000 point data
set sent to Ed May prior to the January meeting is, ironically,
one of the best approximations to a normal distribution.
We
solemnly swear that this was by chance and not by intent (!).
Because the baseline distributions were not generally normal
we chose to examine the psi and control periods using the non-
parametric two sample Kolmogorov and Smirnov test. This has the
additional practical advantage of presenting a graph of cumulative
6 Two other "spark runs" were carried out, but not sent to
SRI since power failures took out the predata. We would appreciate
feedback before repeating these experiments, since it seems as
though a more intense "real" stimulus would be better.
SIRPROJ
CMP/VS
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
21
C. D. F. '5
+ FAKT. 07.5
CONTROL .5
0.8
0.6
C
0.4
0.2
.
0
0
20
40
60
80
observation
Kolmogorov-Smirnov Two-Sample Test
Sample 1: P:OP7. pos5 SELECT P: OP7.mark5 EQ 666
Sample 2: P:OP7. pos5 SELECT P:OP7. mark5 EQ 999
Estimated overall statistic DN = 0.24
Approximate significance level = 0.467558
Fig. 6
Participant 7, %+ve data.
Sequence length 500 raw data points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
22
C. D.F. 's
PAKT. 07.15
1
0.
CONTROL 15
0.8
.
0.6
C
à
0.4
0.2
0
0
20
40
60
80
observation
Kolmogorov-Smirnov Two-Sample Test
Sample 1: P:OP7. pos15 SELECT P: OP7. mark15 EQ 666 1
Sample 2: P:OP7. pos15 SELECT P:OP7. mark15 EQ 999
Estimated overall statistic DN = 0.32
Approximate significance level = 9.2395E-4
Fig. 7
Participant 7, %+ve data.
Sequence length 1500 raw data points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

23
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
+ PART.07. 40
1
C. D.F. 's
CONTROL 40
0.8
0.6
C
å
i
.
0.4
0.2
.
.
0
0
20
40
60
80
100
observation
..
Kolmogorov-Smirnov Two-Sample Test
Sample 1: P:OP7. pos40 SELECT P:OP7. mark40 EQ 666
Sample 2: P:OP7. pos40 SELECT P: OP7. mark40 EQ 999
Estimated overall statistic DN = 0.195
Approximate significance level = 9.95911E-4
Fig. 8
Participant 7, %+ve data.
Sequence length 4000 raw data points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
24
Kolmogorov-Smirnov Two-Sample Test
Sample 1: T: GLOBO7. pospre5 SELECT T:GLOBO7. mark5 EQ 666
Sample 2: T:GLOBO7. pospre5 SELECT T: GLOBO7. mark5 EQ 999
Estimated overall statistic DN = 0.16
Approximate significance level = 0.999999
C. D. F.'s
- GLOBAL 07.5
.O. CONTROL
.
0.8
0.6
å
0.4
0.2
0
14
24
34
44
54
observation
Fig. 9
Global control from predata file
for Participant 7, %+ve, 500.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Kolmogorov-Smirnov Two-Sample Test
Sample 1: T:GLOB07. pospre15 SELECT T: GLOB07. mark15 EQ 666
Sample 2: T: GLOBO7. pospre15 SELECT T: GLOB07. mark15 EQ 999
Estimated overall statistic DN = 0.146667
Approximate significance level = 0.395295
C. D. F. '5
GLOBAL 07.15
O. CONTROL
1
0.8
0.6
C
å
0.4
0.2
G
0
0
20
40
60
90
observation
Fig. 10
Global control from predata file,
for Participant 7, %+ve, 1500.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
25

.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
26
Kolmogorov-Smirnov Two-Sample Test
Sample 1: T:GLOB07. pospre40 SELECT T:GLOB07. mark40 EQ 666
Sample 2: TyGLOB07. pospre40 SELECT T:GLOB07.mark40 EQ 999
Estimated overall statistic DN = 0.12
Approximate significance level = 0.11225
GLOBAL 07.40
C. D.F. 's
CONTROL
1
0.8
0.6
C
å
0.4
0.2
.
0
0
20
40
60
80
observation
Fig. 11
Global control from predata file,
for Participant 7, %+ve, 4000.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
27
Matrix Analysis
ALL PARTICIPANTS - 15 POINTS
100
90
80
70
60
-
.
50
40
-
30 -
20
10
0 -
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
PSI <15 : CONTROL >15
+
"positivo
0
Button push
0
velocity
Fig. 12
Averages by colum.
All files for 40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

28
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Kolmogorov-Smirnov Two-Sample Test
ample 1: V:MATRIX15. velocity SELECT V:MATRIX15. marker EQ 666
ample 2: VIMATRIX15. velocity SELECT VIMATRIX15, marker EQ 999
Estimated overall statistic DN = 0.131532
pproximate significance level = 1.35214E-4
C. D.F. 'S
MATRIX-15 ALL PARTICIPANTS
+ VEL ALL.15
CONTROL
0.8
0.6
.
0.4
0.2
.
1
0
0
30
60
90
120
150
observation
Fig. 13
All participants.
Average of 40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
29
Kolmogorov-Smirnov Two-Sample Test
Sample 1: VIMATRIX15. positive SELECT V:MATRIX15. marker EQ 666
Sample 2: V:MATRIX15. positive SELECT V:MATRIX15. marker EQ 999
Estimated overall statistic DN = 0.0792793
Approximate significance level = 0.0611045
C. D. F. 's
MATRIX 15 ALL PARTICIPANTS
+ % POSITIVE
CONTROL
1
0.8
0.6
0.0 -4-
C
.
d
0.4
0.2
0
0
20
40
60
80
100
observation
Fig. 14
All participants.
Average of 40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

-
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
.
30
Frequency Histogram
MATRIX-15 - ALL PARTICIPANTS
0.15
0.1
0.05
0
0.05
0.1
.
0.15
1
1
0
30
6,0
90
120
150
VELOCITY ( MI CRONS/SEC)
Fig. 15
Average of 40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

31
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
Frequency Histogram
MATRIX-15 ALL PARTICIPANTS
0.1
0.06
{
0.02
0.02
.
0.06
0.1
0
20
40
60
80
100
% POSITIVE
Fig. 16
'Average of 40 and 15 data point trials.
Datum + 15 points.
·
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
kolmogorov-Smirnov Two-Sample Test
Sample 1: VIGMATRIX. gvel15 SELECT VIGMATRIX. gmarki5 EQ 666
Sample 2: VIGMATRIX. gvel15 SELECT V: GMATRIX. gmark15 EQ 999
Estimated overall statistic DN = 0.0747967
Approximate significance level = 0.0640856
C. D. F. 's
GLOBAL MATRIX PRE-DATA
+ VEL ALL-15
CONTROL
1
0.8
0.6
c
0.4
0.2
-
00
40
80
120
160
observation
Fig. 17
Global control. All participants.
40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
kolmogorov-Smirnov Two-Sample Test
Sample 1: VIGMATRIX. gros15 SELECT V: GMATRIX. gmark15 EQ 666
Sample 2: VIGMATRIX. gpos15 SELECT V: GMATRIX. gmark15 EQ 999
.
Estimated overall statistic DN = 0.0666667
Approximate significance level = 0.129969
+ V:GMATRIX. gpost
C. D. F. 's
.0.
V: GMATRIX. gposť
1
0.8
0.6
0.4
0.2
-
-
1
1
1
1
1
0
20
40
60
80
100
observation
.
Fig. 18
Global control. All participants.
40 and 15 data point trials.
Datum + 15 points.
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
1
1
ـاسمها٥

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
34
APPENDIX 1
Description of Data on Discs Sent to SRI on 15 September 1987
September 14, 1987
SUMMARY OF WAVE 1987 PROTOCOL
The Wave protocol allows a participant to become involved in
a trial at their own speed. When the participant recognizes that
they are fully involved in the experiment (i.e., they emerge from
the state of complete involvement
momentarily) , they
end the
trial by pressing the F1 key.
This places a 666 in the marker
field, identifying the end of the psi period and the beginning of
the period which we call control. Please note that the involve-
ment of the participant
is
normally substantially longer than
that required by
the
analytical
protocol.
At the end of the
control period a random
spacer
is
placed
in
the data files,
followed by a Pseudo-psi and Pseudo-control trial.
An additional
longer spacer period
is
then
added before
another
trial
is
allowed
to begin.
Note
that
although the pseudo-trials are
historic remnants (we could have deleted the markers for you), we
have experimental
evidence that
if the algae are stimulated by
some "real"
source such
as a spark, returning to the baseline
condition requires a period at least 2 - 3 times the duration of
the stimulus.
The
total
interval
therefore defines
a trial
period appropriate
for experiments which address a comparison of
psi and spark data.
The data
files
consists of
ASCII
records
written
in a
Sequential file, recorded as the data are collected. Each record
consists of 6 fields, each field is separated by a comma.
Each data record consists of the following fields:
Record #, Time,
Average Velocity,
Standard
Deviation, %
Positive, Marker
All data used are the absolute value (i.e., sign is ignored
except in the calculation of % positive) .
Each recorded velocity
field
(#3) of
each record
is
in reality the mean of 100 in-
dividual raw absolute velocities.
The
standard deviation field
(#4) is calculated using the raw absolute velocities.
The % Positive field (#5) records the number of raw velocit-
ies
which were
positive during
the collection
of the
100 raw
data points.
The Marker field (#6) has the following codes:
Approved For Release 2000/08/10 - CIA-RDP96-00787R000300070001-3
Unassigned Period

Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
35
666 - Button Push at end of psi Period
999 - Control Period
888 - Pseudo-psi Period
777 - Pseudo Control Period
All non-zero Marker periods are of the same length.
The 666
marker indicates the END of the psi period, as this point is of
unknown prior to the decision of the participant.
At the beginning of the set, the number (N) of records to be
collected in each trial
is selected. To analyze the data, this
number used to delineate the psi period.
This is not yet
programmed in:
one must go back to the data files and add (N-1)
666 marker records prior to each existing 666 marker.
The data are on IBM AT 1.2 MB floppies. The filename format
is:
WAVExxxx.dat
where xxxx represents the filenumber shown below.
Test #
#/Trial
PreData File
PostData File
Participant 5
358
5
353
15
357
40
Participant 7
419
5
420
15
417g
---
421
40
Participant 48
408
5
409
15
406
411
410
40
353
359
Participant_17
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
-

-
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3
36
440
5
439
15
437
441
438
40
Spark
412
5
413
15
411e
417
414
40
415
5
The directory structure is:
DISK 1:
ROOT
PART
OP05
OP07
OP48
SPARK
OP17
PRE
POST
DISK 2:
ROOT
PART
OPOS
PRE
POST
OP07
PRE
OP48
PRE
POST
SPARK
PRE
POST
Approved For Release 2000/08/10 : CIA-RDP96-00787R000300070001-3