What the instruments actually measure

PrefaceThe Apparatus Is the Subject

Every other page in this family is about a practice. This one is about the apparatus, which is a different subject and a harder one, because a practice can be described and an instrument has to work.

I have built instruments for this field for twenty-three years, and the single most useful thing I have learned is that almost every device sold into it is sold on a confusion between three separate claims. The first is that a machine helps the operator. The second is that a machine records what happened. The third is that a machine detects something that would be there whether anyone was watching or not. These are not the same claim, they do not need the same engineering, and the trade runs them together because the confusion is profitable.

A pendulum is the first kind. It is an amplifier for the operator's own hand, and it is a superb one. A voice recorder is the second kind, and an honest one is simply a good recorder. The third kind, the detector, is the one almost nothing on the market actually is. A meter that responds to a mains cable in the wall is not detecting a presence; it is detecting a mains cable, correctly, and the person reading it has supplied the rest.

What follows is the whole line of apparatus, from a polished stone to a semiconductor junction, with what each one measures written plainly next to what it is usually sold as. I make and sell instruments at the end of it. I would rather you knew which of the three kinds you were buying.

IAmplifier, Recorder, Detector

Three different claims, routinely sold as one.

Start with the distinction, because everything else hangs on it.

An amplifier takes a signal already present in the operator and makes it large enough to read. The pendulum, the rod, the planchette and the talking board are all amplifiers. The signal they amplify is the ideomotor response: small, unconscious, genuinely involuntary muscle movement produced by expectation. This is not a debunking. It is the mechanism, it was established in the 1830s, and it is the reason these instruments work at all. An amplifier with no signal in it does nothing, which is why a pendulum held by someone with no opinion about the question hangs still.

A recorder preserves what happened so it can be examined later by someone who was not in the room. Its entire value is that it is dumber than you are. A recorder that cleans up its input, that gates out quiet passages, that sharpens consonants, is worth less than a cheap one that does none of those things, because every one of those features manufactures the thing you are listening for.

A detector claims that a quantity exists in the world and that it can be read. This is the only one of the three that makes a claim about reality rather than about the operator, and it is the only one that can be straightforwardly wrong. Nearly every device marketed as a ghost detector is a detector of something real and mundane, worn as a costume.

Hold those three words while you read the rest. Most of the arguments in this field dissolve once you ask which kind of machine is being discussed.

IIThe Surface That Gives the Eye Nothing

The oldest instrument is a surface that gives the eye nothing to do.

Polished obsidian in Mesoamerica, an ink pool in the palm in Egypt and across the Arab world, water in a bowl, a rock-crystal sphere in medieval Europe, and from the sixteenth century the black mirror. John Dee's own scrying stone, an Aztec obsidian mirror that reached England through the Spanish conquest, sits in the British Museum. He did not look into it himself; Edward Kelley did, and what Kelley reported became the Enochian material.

The optics are simple and worth stating, because they explain the appeal without explaining it away. A black mirror is a sheet of glass with a dark backing, so it reflects twice: once weakly from the front surface and once from the back, a few millimetres deeper. Neither image is bright and the two do not coincide. The eye is handed a field with no fixable detail and no plane to settle on, and within a minute or two accommodation drifts, the periphery greys out, and the visual system starts generating structure of its own.

That greying has a name. In the 1930s Wolfgang Metzger showed that a completely uniform visual field, a Ganzfeld, produces exactly this collapse, and that after a short time subjects begin to see forms that are not there. The scryers found the effect four hundred years before the psychologists named it, and the laboratory rediscovered it and built an experiment on it, which you will find further down this page.

So a good scrying mirror is a piece of real optics. It wants a genuinely black backing rather than a dark grey one, a surface flat enough not to hand the eye a distortion to lock onto, no bevel in the visual field, and low ambient light behind the viewer rather than in front. A cheap mirror fails on the backing and the flatness, which is why it shows you the room instead of nothing.

IIIThe Hanging Weight, and Chevreul's Glass Plate

In 1833 Michel-Eugène Chevreul, a chemist of the first rank, did the experiment that settled the pendulum and was then ignored for nearly two centuries.

He held a pendulum over mercury and it swung. He interposed a glass plate and it slowed and stopped. Glass does not block a force that passes through air. What changed was his own expectation, so he went further: he rested his arm on a support, and blindfolded himself, and the motion died. His conclusion was that the swing came from his hand, that the hand moved without his knowing, and that the belief came first and the motion second.

The scale involved is the part people find hard. A tremor of a few tenths of a millimetre at the fingers, sustained at the pendulum's natural period, is enough to build an arc of several centimetres, because a pendulum is a resonator and resonators accumulate. You are not being tricked by a large movement you failed to notice. You are being shown a small one you could not possibly have noticed, made visible by a device designed to accumulate it.

I build pendulums and boards and I have no interest in pretending Chevreul was wrong. He was right, and his being right is precisely what makes the instrument worth owning. A device that reliably reveals what your own nervous system already knows, or believes, or fears, is a genuinely useful object. The uses are on Pendulum Boards. The engineering that matters is mass low and concentrated, a point that is actually a point, a chain with no clasp and no play, and a chart whose divisions were laid out with a compass rather than by a printer's default.

The same argument covers dowsing rods exactly, with one addition: the classic Y-rod is held in unstable equilibrium, so the tiniest wrist rotation produces a violent, startling flip. The drama of the movement is engineered into the grip. It is not evidence of the size of the cause.

IVTaking the Hands Off the Mechanism

The middle of the nineteenth century is when someone first tried to put a number on it.

Robert Hare, professor of chemistry at Pennsylvania, inventor of the oxy-hydrogen blowpipe, sat down in 1853 intending to expose the table-rappers and published in 1855 a book saying he had been convinced. What makes Hare worth reading is not the conclusion but the apparatus. He understood that a séance is uncontrolled because the medium's hands are on the thing that spells, so he built machines that took the hands off it: a lettered disk driven by a cord, a six-foot table on an axle so that any horizontal push would roll a plate off its balls before it moved anything, a spring balance that turned a claim of force into grams.

The whole apparatus, drawn from his own plates and descriptions, is on Spiritoscopes, and the instruments are made there. What belongs here is the principle, which is the most valuable thing the nineteenth century contributed to this subject: separate the person who wants a result from the mechanism that produces it. Every honest instrument since is a variation on that sentence. Crookes weighing an accordion in a cage, the cabinet with the medium tied inside it, the modern sealed random generator in a different room, all of them are Hare's move repeated.

It is also why the trade dislikes it. An instrument built on that principle usually returns nothing, and nothing is difficult to sell.

VThe Instruments That Measured Their Own Operators

The cautionary half of this page is the long list of instruments that measured their own operators.

Baron von Reichenbach in the 1840s had sensitives report a luminous od flowing from magnets in a darkened room. The reports were consistent, detailed, and produced entirely by dark adaptation and suggestion. In 1903 René Blondlot, a respected physicist at Nancy, announced N-rays, and dozens of French papers confirmed them, until Robert Wood visited the laboratory, removed the prism from the apparatus in the dark without telling anyone, and Blondlot's assistant went on reading the same results off an instrument that was no longer whole.

Spirit photography is the same story told with silver. Every effect in the famous plates of the 1860s onward, from William Mumler onward, is a darkroom effect: double exposure, a reused plate insufficiently cleaned, a hand-painted mask. Kirlian photography, which still sells, is a corona discharge in a high-voltage field, and the corona's size and colour track moisture, pressure and humidity with excellent fidelity. Photograph a leaf, tear a piece off, and the phantom shape that is supposed to prove a subtle body is residual moisture on the plate.

The pattern in all four is identical and it is the reason I am tedious about specifications. An instrument with an ambiguous output and an invested operator will produce whatever the operator expects, reliably, repeatably, and with complete sincerity. The remedy is never more sensitivity. It is a fixed protocol, a written record made before the result is known, and somebody in the room who does not care how it comes out.

VIThe Meter, and What It Is Honestly For

The meter in every investigator's hand is a magnetometer, and it is honest right up until someone says what it is for.

Three coils at right angles, one per axis, each generating a current when the magnetic field through it changes. That is the whole device. It reads alternating fields well, which in a house means the mains: 60 Hz in North America, 50 Hz across most of the rest of the world, plus whatever the wiring, the fridge compressor, the dimmer switch and the phone in your pocket are doing. A single-axis meter also changes its reading when you rotate your wrist, which is the source of an enormous amount of excitement.

Two facts make the readings less mysterious than they look. A near-field source falls off as roughly the inverse cube of distance, so a step of twenty centimetres toward an unseen cable multiplies the reading several times over, which feels exactly like something approaching. And old houses, the ones that get investigated, have the worst wiring, so the buildings with the most reported activity are also the buildings with the highest and most erratic baselines.

Used properly the meter is genuinely valuable, in the one role the trade never advertises: it finds the mundane cause. Sweep the room before anyone sits down, write the baseline in a notebook, mark the hot spots on a plan. A field strong enough to be interesting is also, at the top of the range, strong enough to be a candidate explanation for the feelings people report in the room. That is a real finding. It is just not the one anybody wants.

VIIThe Voice, and the Equipment That Invents It

Electronic voice phenomena are the clearest case in the field of an instrument being improved until it manufactures its own evidence.

Friedrich Jürgenson in 1959 heard voices on bird-song recordings; Konstantin Raudive collected tens of thousands of examples and published Breakthrough in 1971. The recordings are real. The question is only ever what is on them and who decided.

The mechanism is well understood and does not require anything unusual. Speech perception is aggressively top-down: the auditory system is built to pull words out of poor signals, and it will do so from noise, especially when primed. Play a fragment of hiss to a room cold and you get nothing; announce what it says and play it again and nearly everyone hears it, permanently. That priming is not reversible, which is why the order of operations in a session matters more than the equipment.

Then there is the hardware. A spirit box sweeps the radio band a few times a second, so it is emitting real fragments of real broadcast speech, chopped to syllable length. Getting words out of it is not surprising; getting anything else would be. Noise-gated recorders, automatic gain control and de-noising all make the same error in the other direction, inventing clean edges where the signal had none.

The honest arrangement is unglamorous. Record flat, at the highest bit depth available, with no gate, no gain-riding and no processing. Run two recorders at once in different parts of the room; a voice present on one and absent on the other came from the room, not the tape. Log every sound anyone makes. Never tell a listener what to expect before they have written down what they heard. Most of what survives that protocol is nothing, and the very little that does is worth more than a thousand hours of the alternative.

VIIIRandomness Used as a Sense Organ

The most interesting machine in this field is also the most boring to watch: a box that produces random numbers and is asked to stop being random.

A hardware random event generator takes a genuinely physical noise source, usually the shot noise across a reverse-biased semiconductor junction or a radioactive decay, and turns it into a stream of bits. The design goal is that nothing should bias it. The experiment is then simple to state: a person tries, by intention alone, to push the mean of the stream away from chance, and the deviation is measured over many trials.

Princeton's PEAR laboratory ran this from 1979 to 2007. Their claim was never a large effect; it was a very small one, of the order of a few parts in ten thousand, sustained across millions of trials. The Global Consciousness Project extended the idea to a network of generators running continuously, looking for departures around events of wide attention.

The criticisms are substantial and you should know them. Effects that small are exactly the size that selective stopping, optional analysis choices and file-drawer effects produce. A multi-laboratory replication published in 2000 did not reproduce the original result at the original size. And an unbiased physical source is hard: temperature drift and supply ripple can bias a junction, so any serious instrument has to log its own calibration alongside the data.

I build these because the design constraints are honest ones and because the machine is the rare case where the apparatus is the experiment. Mine log raw counts with a timestamp, run a continuous calibration channel, and write a file you can hand to somebody hostile. What the numbers mean is not mine to tell you. Whether they are trustworthy numbers is, and that part I will put in writing.

IXThe Uniform Field, Four Centuries Later

The scrying mirror's descendant lives in a laboratory and has a statistics literature.

The ganzfeld procedure puts a receiver in mild sensory homogeneity: halved ping-pong balls over the eyes under red light, white noise in headphones, a reclining chair. A sender in another room, shielded, views one of four targets. Afterwards the receiver is shown all four and ranks them. Chance is twenty-five per cent.

Honest summary of a long argument: the early studies were criticised on randomisation and on sensory leakage between rooms, Charles Honorton and the sceptic Ray Hyman jointly published a set of methodological standards in 1986, the autoganzfeld studies run to those standards reported hit rates around a third, later meta-analyses disagree with each other, and the disagreement is mostly about which studies belong in the pool. That is where it stands. Anyone who tells you the matter is settled, in either direction, has stopped reading early.

What is not in dispute is the first half: a uniform field plus uniform sound reliably produces vivid imagery in most people within fifteen minutes. That part is ordinary perceptual psychology, it is repeatable at home, and it costs almost nothing to arrange. Whether the imagery ever corresponds to anything outside the room is the open question, and it is open.

XWhat Goes on the Label

Here is what I will and will not put on a label.

Every instrument leaves here with its specification written out: what it senses, the range, the sampling rate where it has one, the calibration method, and the failure modes. If a device logs, it logs raw, and the raw file is the deliverable. If a device has a gain control, the position it was in is recorded with the data. Nothing I build smooths, gates, de-noises or enhances, because those are all ways of adding your expectations to the evidence before you see it.

I do not sell detectors of the dead, because I cannot build one and neither can anyone else. I sell amplifiers that are honest about amplifying you, recorders that are honest about being dumb, and measuring instruments that are honest about the quantity they measure. Where a classic instrument's whole value is the operator's own hand, the description says so in the first sentence.

No instrument here is medical, and none of it diagnoses, treats or measures anything about a body. Nothing is sold to anyone under eighteen. If a piece is a reproduction of a historical apparatus, the original and the departures are named. If a claim is contested, the page says it is contested and tells you who contests it.

That is a thinner set of promises than the trade makes. It is the set I can keep.

QuestionsAsked Often