The SRI Geller Experiments 1972/73 – What Was Tested and What Came Out

Published on 2026-05-22 · Reading time approx. 13 minutes

Between November 1972 and August 1973, Uri Geller was the central test subject of a controlled parapsychological research programme at the Stanford Research Institute (SRI) in Menlo Park, California. It was led by the laser physicists Harold Puthoff and Russell Targ. The central results appeared on 18 October 1974 in the journal Nature; a technically more extensive treatment followed in March 1976 in the Proceedings of the IEEE. This article reconstructs the three main experimental series session by session, documents the hits and the misses, goes through the most important points of criticism (Randi, Gardner, Hyman) and the replies, and shows what has become independently checkable after the 2017 CIA declassification. The article complements the Geller portrait and the Puthoff/Targ portrait by the part those two only touch on briefly: the concrete experiments.

The setting: the shielded room at SRI

The experiments took place in a doubly shielded room: an electromagnetically tight steel chamber (classic Faraday cage configuration) inside a further acoustically and visually isolated area at SRI. During the actual sessions, Geller was enclosed in this chamber, alone, with no line of sight outwards and no electromagnetic coupling to the rest of the building. The door of the chamber was heavy from the outside and secured against any simple opening. The experimenters were in another room of the same building, often separated by additional walls and partly by considerable distance.

This arrangement excluded the most obvious conventional channels: no direct visual contact, no audio contact, no easy radio contact. It did not exclude any possible unknown sensory channel – this point will become important in the criticism below – but it was, for parapsychology research of the early 1970s, methodologically markedly stricter than what had previously been usual in most academic laboratories.

Series 1 – Picture reproduction under sensory shielding

The central and most extensively documented series was the 13 picture-reproduction sessions. The procedure was as follows:

1. An experimenter who was not otherwise in test contact with Geller or with Puthoff/Targ generated the target picture – either by random selection from a predefined picture pool, or by spontaneous own drawing in a remote room.
2. The picture was placed in opaque, sealed envelopes and kept in a third room, away from the chamber.
3. Geller sat in the shielded room, was asked to concentrate, and drew what he "saw".
4. Only after the session was complete was the original picture opened and compared with Geller's sketch.
5. In the subsequent evaluation, five independent judges received the ten evaluable Geller sketches and the ten original pictures, in randomised order, and were asked to match them to one another.

The best-known documented hits (original picture → Geller sketch):

The statistical evaluation of the ten evaluable sessions yielded a hit rate whose chance probability, by the independent judges' estimate, was clearly below one in a million. Three further sessions were "passed" by Geller himself – he indicated he was not receiving a clear picture and declined to answer. These three were not entered into the statistics. This is methodologically a double-edged point (more on this in the critique below), but it is transparently noted in the Nature publication.

Series 2 – The die in the steel box

Methodologically particularly clean was the die-in-a-steel-box series. The procedure:

1. An ordinary six-sided die was placed in a closed aluminium or steel box.
2. The box was shaken vigorously by an experimenter on running record – so that neither he nor anyone else could know which face was uppermost.
3. The box was set down at a location where Geller could neither see nor touch it.
4. Geller was asked which number was uppermost on the die in the box – or whether he wanted to "pass".
5. Only after his answer was the box opened.

The result over ten trials: eight of ten answers were correct. Twice, Geller declined to answer. At a chance probability of 1/6 per trial, the probability of eight correct hits in ten trials is around a million to one against chance.

This series matters methodologically because it excludes conventional sensory channels almost completely: no visual contact with the die, no auditory information about its position, no advance knowledge by the experimenter, no spoken or body-language cue from which Geller could have read. This is precisely why even the sharpest critics of the picture-reproduction series usually deal with this series only briefly – it is hard to explain with classical stage-trick hypotheses.

Series 3 – The magnetometer

Outside the test chamber, SRI had a sensitive magnetometer which was used in the institute's regular physics research for measuring very small magnetic-field changes. In a series of filmed sessions, Geller was placed beside it and asked to concentrate on the device without touching it. The magnetometer then registered signal deflections clearly different from the baseline.

These sessions are documented in the original SRI films, which today are partially publicly available (in the documentary Third Eye Spies, Lance Mungia 2017, the footage can be seen in full length). The question whether the observed effects were actually caused by Geller's presence alone or whether conventional explanations (vibration, electrostatic charging, motion artefacts) could be at play belongs to the points discussed in the critique – but has not been fully resolved.

Series 4 – Classical ESP cards

Targ and Puthoff also tested Geller with the classical Zener ESP cards in a Rhine-style protocol (five symbols, long runs). The results in this series were markedly weaker than in the free picture-reproduction sessions. Geller himself told the experimenters that the free, open format suited him better; the fixed card format "made it difficult for him". The experimenters insisted on the classical protocol in a number of sessions nonetheless and documented transparently that this was not the format in which Geller delivered his best results.

This too is part of the honest balance: the Geller data are not equally strong across all formats, and Puthoff/Targ described it this way in the Nature publication and more extensively in Mind-Reach (1977).

The statistical evaluation

The evaluation of the picture-reproduction series was not carried out by the experimenters themselves but delegated to five independent judges. These received the ten Geller sketches and the ten originals in randomised order and were asked to match each sketch to one of the ten originals (so-called forced-choice ranking). The judges' hit rates were statistically compared with chance expectation.

The decisive point: the assessment was thereby decoupled from the scientifically often problematic question whether an individual observer (e.g. Targ or Puthoff themselves) regarded a hit as "close enough". The judges did not know which session was which, and their judgement was based on a purely comparative matching. This methodology was new in parapsychology at the time and became, in the following years, the standard for free-response tests (free-response paradigm).

Criticism and reply

James Randi (1982)

The most extensive and influential critique is James Randi's book The Truth About Uri Geller (Prometheus 1982, originally 1975 as The Magic of Uri Geller). Randi's main points against the SRI picture-reproduction series:

• Possible visual leakage: Randi argued that the wall of the shielded room had had an opening (for communication or ventilation) through which Geller could have seen outwards.
• Target picture storage: in some sessions the target picture was supposedly kept not in a remote third room but in a not sufficiently secured drawer near the chamber.
• Accomplice hypothesis: Andrija Puharich (who had brought Geller to the USA) had been intermittently present in the building and could have served as a possible intermediary.
• Stage-magic explanation: for Geller's famous TV effects (spoon bending, watch restart) classical stage tricks are well documented. Randi transfers this reading to the SRI experiments as well.

Randi's critique is plausible for a portion of Geller's appearances (in particular the public stage performances). For the picture-reproduction series it depends on whether the alleged leaks actually existed – a point that Puthoff and Targ disputed in their reply.

Targ & Puthoff reply (Nature 252, 1974; Proceedings of the IEEE 1976)

A few weeks after the Nature publication, several critical letters and an extensive reply by the authors appeared in Nature 252. The fullest methodological response then appeared in the IEEE paper of 1976. The main points:

• The alleged visual leakages had not been physically possible; the chamber had been checked before the start of every session.
• The target-picture storage had, in the ten evaluated sessions, been in a sufficiently distant room.
• Andrija Puharich had not been in the same building as the target picture in any of the evaluated sessions.
• The die-in-a-steel-box series and the magnetometer effects were not explainable with stage-magic hypotheses in any case – Randi's critique factually applied almost only to the picture-reproduction series.

Martin Gardner and the Skeptical Inquirer milieu

Martin Gardner and several authors of the Skeptical Inquirer have over the years made points similar to Randi's, often with the emphasis on the weak randomisation of target-picture generation in some sessions and on the after-the-fact selection of "evaluable" sessions (those ten of thirteen in which Geller did not pass). This is a methodologically serious point, because a selective inclusion can distort the statistics – but Targ and Puthoff transparently documented that the three not-evaluated sessions had been marked as "pass" before the start of the respective evaluation, not afterwards.

Ray Hyman in the 1995 AIR Stargate review

Twenty years after the SRI experiments, Ray Hyman (psychologist at the University of Oregon, self-described sceptic) was called in as part of the CIA-commissioned review of the Stargate programme to evaluate the entire SRI data. Hyman concluded – as did the statistician Jessica Utts – that the statistical effects are real and not explainable by chance. He attributed them, however, to suspected methodological flaws that he was unable to specify concretely. That makes Hyman one of the few prominent sceptics who acknowledge the raw statistical findings but consider them inconclusive.

The film record and the 2017 CIA declassification

A portion of the SRI Geller sessions was filmed. The original recordings are today partly publicly available – both on YouTube and at length in the documentary Third Eye Spies (Lance Mungia, 2017), in which Russell Targ personally guides the viewer through the material. To be seen, among other things, are the picture-reproduction sessions in the shielded room, one of the magnetometer sessions, and Geller's reaction in the moments in which he "passes".

In 2017, under pressure from FOIA litigation, the CIA released around twelve million pages of Stargate files, in which a portion of the SRI Geller original protocols is included. The material is today publicly available in the CIA Reading Room. Anyone who wants to see the session protocols, the original sketches and the evaluation calculations directly no longer has to take anyone's word – the material is accessible at the source.

What stands up after fifty years

An honest balance of the SRI Geller experiments today, more than half a century later, can be summarised in three points:

1. The statistical effects are real. Even the sharpest sceptics (Hyman, in part Gardner) concede this. A chance explanation for the evaluated sessions is ruled out.
2. The methodological quality was high for the time, but not perfect. The picture-reproduction series in particular has assailable points (possible visual leakage, after-the-fact definition of evaluability). The die series and the magnetometer sessions are methodologically cleaner and are rarely attacked frontally even in the sceptical literature.
3. The binary question "real or trick?" is not the right format. Some of Geller's TV effects were, with high probability, classical stage magic. Some SRI sessions cannot be explained this way. A scientifically careful position has to distinguish between these two levels – and that is precisely what the original Nature publication and the IEEE paper have always done.

Anyone wanting to form their own view today has an unusually good source base: the Nature original publication (Nature 251, 602–607), the methodologically extensive Proceedings of the IEEE paper (64(3), 329–354), the book Mind-Reach (Delacorte 1977) and the declassified CIA files. Plus the original film footage. That is markedly more material than is publicly available for most research programmes of the early 1970s.

Sources: Russell Targ & Harold Puthoff, Information transmission under conditions of sensory shielding, Nature 251, 602–607, 18 October 1974. Follow-up correspondence in Nature 252 (1974). Harold E. Puthoff & Russell Targ, A perceptual channel for information transfer over kilometer distances: Historical perspective and recent research, Proceedings of the IEEE 64(3), 329–354, March 1976. Targ & Puthoff, Mind-Reach: Scientists Look at Psychic Ability, Delacorte Press 1977. James Randi, The Truth About Uri Geller, Prometheus Books 1982. Martin Gardner, columns in the Skeptical Inquirer, several volumes. Ray Hyman & Jessica Utts, AIR review of the Stargate programme, 1995 (public). Lance Mungia, Third Eye Spies, documentary 2017. Declassified CIA Stargate files, CIA Reading Room, online since 2017. Russell Targ, The Reality of ESP, Quest Books 2012 (retrospective on the SRI phase).