Cosmophysical Influences, Shnoll, et al.

Simon Shnoll and colleagues have been gathering evidence for the past 40 years on apparent "cosmophysical influences" that produce concentrations of identifiable similarity of regions in time series that should have no similarity beyond chance occurence. They create histograms to represent segments of time series, and examine all the histograms in pairs which have randomly assigned identifiers. Blind judges select pairs which appear similar (algorithmic methods are still being developed), and the frequency is computed for pairs separated by differing time intervals. The finding is that synchronous or adjacent pairs, and pairs at intervals of 24 hours, 27 days, or 365 days show similarity more frequently than they should by chance. Data from a variety of different physical systems have been assessed, for example, counts of radioactive decay for plutonium and cesium sources separated by 200 Kilometers. A recent paper which discusses the procedures and responds to various questions and criticisms is available here, as a pdf document. A selection of comments is available, drawn from the extensive email discussion among researchers interested in automating the comparison process.

Most recently Shnoll and his team have examined data from the GCP -- and they have found the same pattern of similarity between synchronous or adjacent segments of data displayed as histograms, implying that the source of the structure is informational. The following email correspondence and the figures tell the story.

Date: Sun, 25 Jun 2000 16:31:07 +0400
From: Tatyana Zenchenko zenchenko@pbc.iteb.serpukhov.su
To: rdnelson@Princeton.EDU

Dear Dr. Nelson,

We were very glad to know about the existence of global physical
generator net realized within Global Consciousness Project.
We deal with physical noises for many years. These investigations have
resulted in a conclusion that processes of any nature are affected by
some global cosmophysical factor. It follows from the fact that
distributions of time lags between similar histograms have significant
maxima, corresponding to basic Earth-Sun periods. Another fact is a very
high probability of form coincidence for synchronous (by local time)
histograms.
These results have been [published over] many years. The latest are:
http://ufn.ioc.ac.ru/abstracts/abst98/abst9810.html#d

http://www.ufn.ru/ufn2000/ufn00_2/ufn002h.pdf
and
http://www.copernicus.org/EGS/pce/part_a_vol_24_no8b.html
(but we are not sure that the last paper text is available in internet).

But this fact was out of attention up to today. Our point of view allows
to see the features of time series [in a different way from] analyzing 
by more traditional methods (Fourier transformation, correlation, random
walk and so on). The replacement of original time series by the sequence
of histograms gives us the possibility to see the regularities that are
invisible for others. Therefore it was rare pleasure for us to see
multiple time series, created by your global net (picked up from
Internet) and containing (according to our primary treatment) the same
effects.
We would be glad to know your opinion on the using these data in
official publications with all necessary references. And we would be
very grateful for information about precise position (coordinates) of
each EGG. It is very important for our investigation.
During The International Congress of Biometeorology (Austria, 1990) we
suggested to create such a global net to study the mutual histograms
features for these time series sets.
So, we are glad, indeed.

With best regards from Prof. Simon Shnoll

Sincerely
Tatyana Zenchenko

Subject: Re: our result on GCP data
Date: Tue, 27 Jun 2000 01:09:01 +0400
From: Tatyana Zenchenko  zenchenko@pbc.iteb.serpukhov.su
To: "Roger D. Nelson" 

Dear Roger,

The result attached is not for publishing anywhere, because it does not
contain any statistical estimation yet - unfortunately, there is no time
for this now, I have to prepare materials for big report planned in a
week.

But this distribution could be interesting for you as received from data
of GCP (time series of EGGs #37 and #1000, from 00.00h up to 10.00h UT
June 9 2000). The choice of EGG and time was spontaneous, in that time
we had no any information about EGG's position. Fragment of original
data, summed up for 6 sec, was used for construction of histograms (60
points per one histogram, so one histogram covers 6 min). So we obtained
two sequences, each 100 histograms long. 1585 histogram pairs from 10000
possible pair combinations were found to be similar. The distribution of
time intervals between similar histograms is shown in figure, black line
corresponds to case of completely random distribution. Data for this
plot are also attached.

It is interesting, that time of measurement beginning was treated
according to UT, so the maximum number of similar pairs in this plot
(t=6 min) corresponds to difference of local times between Switzerland
and Netherlands (about 8 minutes. It is not [optimal], that these two
EGGs are turn to be not far from each other, but I repeat, it was
spontaneous choice.

I hope that I will be back from the Congress on July 5 and will have a
chance to give this result an appropriate statistical estimation and
devote much more time to treatment of GCP data, it is extremely
interesting for me. But if your have any question now, I will be glad to
answer as soon as possible.

With best regards,
Tatyana

Subject: Re: our result on GCP data
Date: Wed, 28 Jun 2000 13:40:06 +0400
From: Tatyana Zenchenko zenchenko@pbc.iteb.serpukhov.su
To: rdnelson rdnelson@Princeton.EDU
References: 1

Dear Roger,

My colleague, Maxim Fedorov, fulfilled necessary statistical estimation 
for this distribution, so now you can place it on GCP site. I can add, 
that the red line indicates 99% confidence interval, numbers mean the
probabilities of random realization for two central spikes.
The description of the estimation procedure used here is now in press, but
I doubt, that you use Microsoft Word, so I attached it as a gif-file.

rdnelson wrote:

> What is most interesting, as I am sure you agree, is that
> the random sequences of the GCP data aren't vulnerable to
> ordinary EM energies, so the implication is that the effect
> is in the informational domain.  Of course it will require
> much more experience with these materials before any
> conclusions should be drawn, but the results will be very
> instructive in any case.

I absolutely agree.

> Also, I think it will be worthwhile to work more on finding
> an effective algorithmic matching procedure.  Even though the
> human judge method can be sound, it would be valuable to
> apply the method to much more data.

I agree again! Two members of our group, Alexander Konradov
and Konstantin Zenchenko, deal with this problem, and can give 
more competent answers than I, for your possible questions in 
this realm. But they also participate in the Congress in 
S-Peterburg next week, so they will be available for links
after July 6.

My best wishes,
Tatyana
 

A selection of comments is available, drawn from the extensive email discussion generated by recent web-based publications, including the analysis of GCP data. Among them are some analyses using algorithmic methods. Other independent efforts to find algorithmic procedures are under way. One is by Dick Bierman, using wavelet analysis, as described here.

Date: Mon, 17 Jul 2000 21:47:55 +0200
From: Dick J Bierman 
To: Dean Radin 
Cc: kelvin@fourmilab.ch, rdnelson@Princeton.EDU
Subject: wavelet results (preliminary)

Hi Dean,
I have played a bit with the wavelet transform and I might be onto
something:

Procedure:
1. daubechies-16 wavelet transform
2. keep coefficients 4-16 (the lower coeffcients mostly contain the
   gaussian characteristics)
3. correlate the 12 remaining coefficients.

This gives the following correlation table for the historgrams
1,2,3,4,7 and 8
Note that I have the data only in already transformed format (i.e
stretched, mirrored and shifted)
Nevertheless this looks good especially because 2 and 3 also
correlate a little bit which I think should be the case although the
russians don't report that.



Note also that the correlation goes down rather systematically  with
increasing temporal distance!

What I need now are the RAW histograms in order to see how sensitive
the wavelet transforms are for the transformations (they shouldn't 
but one never knows). Could you mail them or do I need to ask
Tatiana?

Best,
Dick