IN THE WINTER OF 1991 we circulated a paper in manuscript claiming to have discovered 13 unusual radioactive isotopes in soil samples from an English crop circle. We argued that the isotopes were products of a type of radioactive bombardment called "deuteron activation."
We are satisfied with our logic, but,
unfortunately, the basic data turned out to be less solid than we had believed. For that reason, we pulled the paper from publication, and are withdrawing some of the claims we made in it.
IN THIS ARTICLE we will discuss which claims we have withdrawn, and which we are retaining. We will also discuss our data problems in some detail, so that our learning experience may be shared with the cereological community.
THE SOIL SAMPLES were taken from a "fish" formation formed on July 31/August 1, 1991 at SU 0865 6810 (near Beckhampton.) (1) On August 5th, we took two samples (named "1A" and "IB") from inside, and took a control several dozen feet away ("1C"). These samples were airmailed to Tennelec/Nucleus Instruments (now renamed Oxford Instruments) of Oak Ridge, Tennessee. There, they were placed in a Tennelec/Nucleus LB40008 gas flow low background counter to measure their emissions of alpha and beta particles. (Alpha and beta particles make up two types of radiation.) This testing, performed on August 18th, yielded the following data:
(The three tables presented in this article remain valid.) Sample LA yielded alpha emissions 198% above the control and beta emissions 48% above the control. Sample IB yielded alpha emissions 103% above the control, and beta emissions 57% above the control. While we could not submit these figures to statistical analysis for significance (that requires more data), they seemed to us to be strikingly elevated.
The soil in the samples was two and three times as radioactive as that of the control.
We wished we had taken a second control to compare to the first, but nevertheless, the disparity seemed striking. We had taken two controls from another formation, made August 9/10, 1991, at SU 076 679.
The controls' emissions were quite close to each other, whereas the two samples from inside registered considerably higher. The two controls yielded alpha and beta counts within 2 % and 4% of each other.
By contrast, the two samples from within the formation yielded alpha and beta counts 22% to 45% higher than the averaged controls. But perhaps the soil was more variable than we knew. Perhaps it was just a fluke that the controls were lower.
In two other formations we had tested, the controls were actually somewhat higher than the samples. (2) Thus, to test soil homogeneity, one of our colleagues, Kevin Folta (a graduate student in molecular biology at the University of Illinois at Chicago), tested 20 soil samples collected in a walk around DeKalb, Illinois. The tests were made with a liquid scintillation counter, which works differently from a gas flow counter, but it is relative counts that matter here. All 20 samples fell within a range of 50 to 78 counts per minute, a 28-count spread.
Does that 28-count spread show that the soil is homogeneous or heterogeneous? Consider how it stacks up against soil samples which Folta took from inside and outside a crop circle near Argonne, Illinois:
The variation between samples and controls in the Argonne circle was much larger than the 28-count range of variation exhibited by the 20 samples taken around DeKalb. Folta's measurements gave us more confidence that the radiological anomalies seen in the English circles was not a fluke. (Of course, we will have to put our hypothesis to the test next summer by analyzing many more samples and controls, and that is precisely what we are planning to do.)
Before going on with our discussion, we want to reassure readers that the radiological anomalies did not appear to present any health threat. Even though the samples emitted higher percentages of radiation than the control, their total emissions were far below the danger threshold. We are dealing with very slight effects which are detectable only by extremely sensitive instruments. It is likely that a handheld Geiger counter would not be sensitive enough to detect them. Our hypothesis was, and still is, that there were genuine radiological anomalies in the soil from these three crop circles. But what was causing those anomalies? If the high emissions were "smoke," what was the "fire"? To answer that question, we sent samples 1A, IB and 1C to a government laboratory for testing with a gamma spectroscope. Unlike a gas flow counter or a liquid scintillation counter, a gamma spectroscope does not measure levels of radioactivity
Rather, it finds out what is causing radioactivity. It identifies the specific radioactive isotopes within a given substance. The gamma spectroscopy was performed on August 26th, 1991. The results were supplied to us in the form of a computer-processed table of "peaks." Each radioactive isotope has a characteristic signature composed of several peaks. When we first saw the data we felt sure there was something significant in it. The control had 90 peaks, and sample LA had 200 peaks! This looked like very strong evidence that there was something in the sample which was not in the control.
Due to the government lab's security restrictions, we were unable to view the raw data which the computer had processed to make the table of peaks.
However, we assumed that the lab had followed consistent and statistically valid procedures in processing the three samples' data, so we felt safe in working with the table of peaks. We later found that this assumption was incorrect. Not knowing this at the time, we proceeded with our analysis. Using the peak tables, we identified 13 highly unusual radioactive isotopes in sample 1A, and one (possibly two) in IB. These isotopes were not known to be produced in nature, nor were they known to be emissions from atomic tests, nuclear power plants, or Chernobyl.
We carefully considered a variety of other mundane causes: natural radionuclides, cosmogenic radionuclides, sample jar contamination, airport X-ray detectors, thermal neutron activators and contamination with hospital waste by hoaxers. None of them held up as valid sources for the isotopes.
Many of these possible sources were ruled out by the fact that many of the isotopes had halflives of about two weeks, which indicated that they had not been in existence for very long! It seemed reasonable to guess that they had been formed when the crop circle itself was. Since the 13 isotopes had one, and only one, common denominator — the property of being producible by bombardment with heavy hydrogen nuclei — we proposed that that was indeed what had happened. We were very excited by this theory, because if it was true, it meant that something extraordinarily exotic had happened when the crop circle was formed. Hoaxers could not possibly have done it. We wrote up our findings and circulated the manuscript quite widely.
The paper was slated to be published in February 1992, but just prior to publication, we were afforded the opportunity to view the lab's raw data.
We were dismayed to find that many of the peaks were so close to the noise level that we could not be confident they really existed. The lab's computer had been programmed to hunt for peaks quite aggressively, identifying many of them on statistically inadequate grounds. Furthermore, the lab had made appalling errors in handling the samples. Sample IB had been counted on a different detector, and with a different system, than samples 1A and 1C. We therefore had no choice but to throw IB's data out, since there was no similarly analyzed control to which it could be compared. This was dismaying but not fatal, since our analysis had focused almost entirely on sample 1A. Also, we found that the lab's computer had used different sensitivity levels to analyze the raw data for sample 1A and the control, making the analysis (the table of peaks) supplied to us useless.
We pulled the paper — but we kept working on the raw data, to see what could be salvaged from it. We reanalyzed the raw data for sample 1A and the control with proper sensitivity levels, to pick out the peaks which were statistically valid. Once this was done, 53 peaks were left in sample 1A, and 40 were left in 1C, the control. All of these peaks were three standard deviations removed from the noise level, which meant that there was a 99.9% chance that they were not due to noise or statistical variation. The simple numerical disparity of peaks — 53 versus 40 — suggested that something was in LA which was not in 1C.
Based on these peaks, we now tentatively argue that four unusual radioactive isotopes (three of which are from the original 13 isotopes) were in fact truly present in LA: vanadium-48 (half-life: 16.1 days), europium-146 (4.6 days), ytterbium-169 (32 days), and gold-192 (4.9 hours) (3). The gold-192 was present in the control as well, so we think its presence was due to contamination from the government lab. (We have several other reasons for thinking this as well, which are a little too technical to get into here.)
We are still not totally satisfied with the identification of these four isotopes, since in each case we had to use one or more peaks which were less than three standard deviations from the noise. (It takes a combination of peaks to identify any given isotope.) It's rather like identifying an automobile in a snowstorm. Parts of it may be blurred, but enough of it is visible to rule out its being anything else. Parts of the isotopes were "blurred" — i.e., some of their peaks were at 95% confidence rather than 99.9% — but the total combination of peaks spoke strongly for positive identification. In addition, the alpha/beta counts suggested that there was something unusual in sample LA, and these peaks represented by far the most likely candidates. But rather than make definitive claims, we prefer to wait for next summer and do more testing.
We cross-checked our conclusion in another way. The four isotopes have halflives measured in weeks to hours, so the peaks identifying them should be absent from a recount made months later. Thus we had sample 1A recounted between February 20th and 25th, 1992 (for five days, or about 6,800 minutes) on a gamma spectroscope provided by Oxford Instruments.
This gave us good statistics, and we were allowed to work on the equipment ourselves, ensuring that scientific methods were employed, the correct calibration curves used, and the raw data recorded for re-examination at will. The natural radioactive isotopes we originally detected in the soil were still there, such as uranium-238 and radium-226. There was also cesium-137 from Chernobyl. These served as checks on the reliability of the original analysis. But the very peaks that identified the four unusual isotopes were gone: precisely those peaks, and none other. This suggests that those four isotopes had decayed off, as predicted. Although we suspect these isotopes were generated through activation by whatever energy source created the circle, it is possible they were caused by contamination at some point. One of the primary foci of the work in 1992 will be to analyze samples quickly after the formation is made, to avoid losing information on isotopes with short halflives. The fact that we will have a gas flow counter and a gamma spectroscope under our own control in Oxford, England, will cut the time between sampling and testing to mere hours. In addition, we plan to analyze the plant material as well, which will help us determine whether any unusual isotopes discovered are from activation or contamination.
What lessons should we (and other cereologists) draw from our troubles?
First, the recognition that scientists and scientific labs are fallible, too; their claims should never be uncritically accepted, but rather probed, questioned, and replicated by others. Second, that cereologists should make their raw data freely available to others for review. This enables the community to detect problematic data or invalid claims. This in fact happened to us, since it was certain skeptical responses to our manuscript that made us insist on getting to the lab's raw data.
The system worked: our paper never was published.
Third, the lesson that the discovery of an error constitutes an advance in knowledge. We have learned, albeit the hard way, a great deal about what assumptions not to make.
Finally, we have learned not to put all our eggs in one basket. Even in our darkest moments, we were heartened by discoveries made by colleagues in completely different fields. In addition to his scintillation counts, Kevin Folta examined plant DNA in samples and controls from the Argonne circle, and found that DNA from inside the circle was considerably more degraded than DNA from outside it. This suggested that the plants had been exposed to some form of radiation. Dr. W. C. Levengood of Pinelandia Biophysical Laboratories found consistent anomalies in plants from crop circles around the world, including node swelling, cell wall pit enhancement, polyembryony, increased seed germination rates, and variations of oxidation and reduction characteristics. And Cassandra McDonough of Texas A&M examined seeds under a scanning electron microscope, and found effects consistent with Levengood's hypothesis that the plants had been rapidly heated and cooled. (These findings are now being written up for submission to refereed journals.) In short, we learned that in cereology, diversification is critical. In the summer of 1992, we will put these lessons to good use.
Those who would criticize our mistakes would do well to turn their energies in the same direction.
Notes
(1) Date and location data supplied by John Langrish. Langrish's figures differ slightly from the ones given in Michael Chorost's report, The Summer 1991 Crop Circles (Fund for UFO Research, P. O. Box 277, Mt. Rainier, MD, 1992.) They are more authoritative, so we use them here.
(2) The six cases we tested are discussed at length in The Summer 1991 Crop Circles (see note 1.) A condensed version of the report was printed in the Mufon UFO Journal, October 1991, pp. 315.
(3) The gold-192 could also be interpreted as gold-194, which has many of the same peaks. If this is the case, then four, not three, of the original 13 isotopes would be present.
______________________
Francesco Grassi
