Chapter 6
Extensions and contractions

In the last chapter I attempted to interpret synchronous strains by invoking a physical model of a moving 'surface of action' at which strains are experienced. This might be regarded as some sort of field of dynamic strain or of stress, but it is a field with unusual properties; we imagine the strain to be zero or very small at points in space which do not lie on the surface; but at points which do lie on the surface, possibly at all these points, bending strains or stresses are experienced. The surface moves slowly in space, possibly under weak control of the subject.
However, everybody knows that to bend a strip of metal it takes the action not of one force but of at least three, arranged thus: /\ \/ /\
This action is known as a 'three-point load'. It might be supplemented by other forces acting at neighbouring points, or even by a continuous array of forces; yet always there must be two opposed torques, centring on different points within the specimen (a shear). How can such torques be produced at a 'surface of action'?
In the Nicholas Williams data there are strain gauge signals which indicate a permanent deformation of a metal specimen, without any actual bending being visible. Examples (not illustrated) are signals A 4(2), B 2(2), C 1(2). I did not enter such events on the sensitivity graph (Figure 4.3) since, being drawn fully logarithmically, it cannot display zero bend angles.
The signals indicate either a permanent extension or a permanent bend (since the strain gauge is not on the neutral axis). But since no bend is visible, a permanent extension must be indicated. Possibly many of the elastic deformation signals are also extensions rather than bends. For the bending of a strip of metal there is extension on the convex side and contraction on the concave side, so that a single strain gauge would be inadequate to distinguish a bend from an extension.
There is another group of experiments which supports the idea that the action consists of extension or contraction pulses, and not necessarily bending pulses; the fracture of epoxy-resin bonds between thin strips of aluminium (Table 3.1). The most likely interpretation of these fractures is that one strip is expanded without the expansion of the other; a shear force produces fracture, and no bend is observed.
However, the issue of whether there are only single extension pulses, or more complicated action, can be decided only by studies with more than one strain gauge. The first data with two strain gauges mounted inside a single specimen were obtained using as a subject a physicist and transcendental meditator, Dr Rob Basto, who has on several occasions proved his ability to produce paranormal signals under observation. The two strain gauges were mounted within a thick metal specimen, parallel to each other and to the neutral plane; all the principal signals extended both strain gauges, and no permanent deformation was observed. But this represented the results of only a single experiment.
When thirteen-year-old Stephen North became known to me as a metal-bender, I set up the strain gauge equipment in his home; within half an hour I found that he was producing an abundance of strain pulses, under good conditions of observation. His sister Sarah, his mother and his father sat round in the most natural and informal way possible while these phenomena developed. I determined to give priority to the exposure of several strain gauges mounted on a single metal specimen.
A session was arranged with a 250 X 9 X 0.75 mm aluminium strip mounted horizontally, opposite the subject, and with the surface of the strip vertical. It was suspended from a wooden stool by the electrical connections to three strain gauges; these were mounted on one side of the metal, and were evenly spaced along it. It was about half an hour before Stephen settled down to produce a series of synchronous signals. What was significant about them was that the signals on the left-hand and centre strain gauges were contractions, while the right-hand strain gauge signals were extensions.
I was forced to the conclusion that there can be simultaneous extensions and contractions on a single piece of metal; I must therefore design experiments to measure the distributions of sign and magnitude throughout the thickness of a metal strip. This requires the use of an array of strain gauges through the metal; but at first I had to be content with one strain gauge mounted on the front and one on the back. If the observed nominal strains were equal, and of opposite sign, then there would be pure bending about a neutral plane passing down the centre of the strip. On the other hand, if the signals were equal and of the same sign, the simplest inference would be that there was no bending, but only pure extension or contraction. During sessions of about 100 minutes, I usually observed rather fewer than 50 pairs of signals, which were analysed as follows:
Suppose that the signal I1 at strain gauge 1 on the convex surface of a bent metal strip consists of a contribution st from a stretching pulse and a contribution b from a bending pulse, so that I1 = st+ b. The signal I2 at strain gauge 2 on the concave surface is I2 = st+b. We define a 'proportion of stretching' R = St/(St + b) for each pair of signals, and for a session we calculate the arithmetic mean R of values of R. This quantity defines the 'proportion of stretching' in the action of the subject in a particular session. The distribution of individual signals about these means is defined by the standard deviations O(S)/St and sigma(b)/b, where
sigma(St) = {Sum(St-St)/(n-1)}^0.5
where n is the number of pairs of signals in the session.
Metal specimens of different thicknesses t were offered on different occasions to Stephen North, Mark Henry and Rob Basto; the data are summarized in Table 6.1. The subject was normally situated within one metre of the specimen, and was observed throughout the session, not being allowed to touch the specimen.
It is apparent that Rbar the proportion of stretching, shows variation over about one order of magnitude, and that this correlates well with the thickness t of the metal specimen. The quantity Rbar/t is seen from Table 6.1 to be reasonably constant in this range and a plot of R against t is shown in Figure 6.1. Presumably this correlation relates to the psychological approach of the subjects to the metal specimens, which were of course seen by them. A subject has sufficient confidence with a thin specimen to 'produce' pure bending forces, but when presented with a thick specimen he has not this confidence and 'produces' a large proportion of stretching.
The distribution of individual signal pairs about the means is always fairly wide, since the proportional standard deviations sigma(St)/Sbar.t and sigma(b)/bbar do not differ greatly from unity (mean value 1.05). An exception to this rule is the short burst of signals recorded within 1 minute from Rob Basto; these were remarkably self-consistent. Apparently it is much more difficult to maintain this consistency over an entire experimental session lasting about 100 minutes; this would be expected in any human phenomenon.
RB, Rob Basto; SN, Stephen North; MH, Mark Henry.
Table 6.1 Analysis of bending with stretching sessions
Subject and session Specimen dimensions l (cm)w(mm) MaterialNo. of visible deformationsNo of signal pairsI mean St (mV)mean b (mV)mean Rt (mm)mean R/t sigma(st)/stsigma(b)/b
RB 11 7.5 Eutectic 0 11 0 1.23 1.09 0.53 6.5 0.08 0.094 0.086
SN E 10.2 7.5 Aluminium 1 52 0.14 0.20 2.74 0.068 0.75 0.09 1.34 O.85
SN F 10.2 14 Brass 0 32 0.39 2.30 0.87 0.725 5.0 0.15 2.71 0.73
MH 1 18 12.5 Aluminium 0 14 0.10 0.41 1.84 0.182 1.25 0.15 0.610.62
MH 2 10.2 7.5 Aluminium 0 37 0.43 0.51 5.06 0.092 0.75 0.1221.27 0.72
MH 3 10.2 7.5 Aluminium 0 56 0.21 0.85 5.08 0.143 0.75 0.190.95 0.70

No distinction has been made in this analysis between signals of different polarity; i.e. contraction as opposed to extension, or bending in one direction as opposed to bending in the opposite direction. Nevertheless there is considerable alternation in these polarities, and l have chosen to characterize it in the following way. Each closely-spaced group of signals, or each isolated signal, is called an 'event'; the 'in-decision parameter', I, is defined as the ratio;
I= number of changes of polarity during session
number of events during session
Figure 6.1 Variation of mean value R. of bending-stretching ratio during sessions, with thickness t of metal strip, on each side of which two resistive strain gauges were mounted.
Closed circles, Mark Henry; closed triangles, Julie Knowles; open triangles, Stephen North; crosses, Rob Basto; open circles, Jean-Pierre Girard.
For thin metal strips there is some justification for a linear R-t dependence (i.e. the thinner the strip the purer the bending). For the thick metal bars there is a tendency for R/t to approach 0.5 (broken line) (i.e. signal on one strain gauge only; failure to penetrate the thick bar). Only in one session with Jean-Pierre Girard was a good 'bending purity' obtained with a thick bar.
Values of I have been recorded in Table 6.1 and elsewhere. Since the unweighted mean of tabulated values of I is 0.18, there is on the average a change of sign after every five events.
Since conducting these experiments I have been able to work with the adult French metal-bender Jean-Pierre Girard, using a very thick bar of aluminium. In one session he produced pure bending signals, but in the second session nearly all the signals were on the top strain gauge only (his hand is always above the metal). The two values of R are shown in Figure 6.1, and it is clear from this representation that a value of R = 0.5 (i.e. signals on one strain gauge only) is approached for large thickness t. There may well be signals of opposite sign within the metal, but they do not penetrate as far as the other side. Further sessions with thick metal specimens acted on by other metal-benders showed similar results, and are included in Figure 6.1.
Figure 6.2a,b Dimensions of thick metal strips containing six resistive strain gauges for profile studies: (a) laminar, (b) slotted.
l could not abandon the possibility that the profile of stretching and contraction across the thickness of a metal strip might be more complicated than the simple bending and stretching envisaged so far. I therefore designed thick metal strips with six strain gauges distributed across the thickness, and successful exposures of them were made with Stephen North. The dimensions of these specimens are given in Figure 6.2 a and b and typical profiles from amongst the hundreds of signal events are shown in Figure 6.3. It is seen that the action is in fact more complicated than a simple bending or stretching. There is an important fraction of events in which the sign of the signal changes more than once as we proceed across the thickness of the specimen. They are neither stretching, contractions nor bending events--they are distortions. The metal is not being bent; an attempt is being made to churn it up!
Figure 6.3 Profiles of a typical run of signals, in Stephen North's session S. from six resistive strain gauges mounted across thickness of a metal bar. Signals to the right are extensions, signals to the left are contractions. Thus in the schematic representation of a pure downwards bend shown on the left the arrows represent the expected signals. The recorded signals are three Ws and a \/\.
A simple characterization is by the number of times the gradient changes across the profile. In a pure bend, as can be seen from Figure 6.1, there is no change of gradient. In the remainder of the events in this Figure there are changes of gradient, and we characterise events 44, 46, 47 as W events. Also possible are \/, \/\ and /\/\/ events. During the three Stephen North sessions Q. S and T. the distributions of these events were as indicated. Thus it appears that there is a distribution of complexities of profile; the simplest and most complex are perhaps less common than the mean.
I have recently conducted experiments with strain gauges mounted within solid specimens: cubes and spheres. The strain tensors are complicated. In the first experiment with a sphere of 1 inch diameter, by far the most powerful signals were recorded on the strain gauge pointing radially to the subject. Almost no signals were recorded on the one pointing radially away from him. Integrated signal strengths were in the proportions 501, 78, 46 and 1 at respective orientations 0°, 60°, 120° and 180°. It is possible that this sphere experiment represents some measure of the attenuation produced by screening within a really thick piece of metal. Incidentally, the experiment could hardly be described as an experiment on metal-bending, since bending of the spherical form is not possible without a previous major distortion. Compression of the sphere between the fingers gave no observable signals, and the observed extensions and contractions in various directions are impressive as validation. The indecision parameter C had the unusually high value of 0.3 for the session.
The concept of 'surface of action' has therefore to be modified, in the sense that it is now only a macroscopic model, applicable over distances of several centimetres or metres, and not necessarily valid on a microscopic scale. It still might be regarded as a sort of extension of the subject's arm, but it is more of a slab-like region than a surface.

Back to books list.
Back to main index.
Back to Uri Geller's home page.