"Pathological Science" is a short review by Sheldon Stone of the Physics Department at Syracuse University, posted as arXiv paper hep-ph/0010295 (the second version is dated 31 October 2000). It is not a history of N-rays. It is a working physicist's warning to other physicists, and N-rays occupy about a page of it. What the paper adds to this dossier is a frame: it places René Blondlot's rays beside other results that were believed and then vanished, and it offers an explanation of how the detection method itself could have produced them.
Where the paper's framework comes from
Stone borrows his subject from Irving Langmuir, whom he identifies as a Nobel prize winner in chemistry. Langmuir gave a talk in 1953, later transcribed, on what he called "the science of things that aren't so": results reported with impressive statistical weight that later turned out to be wrong. Stone repeats some of Langmuir's examples and adds others from high energy physics.
One condition governs the whole paper. Stone states that none of his cases is fraud. The people involved believed what they presented, and they were mistaken. That matches the account of the Nancy observers given in Robert Wood's 1904 Visit to Nancy: The Removed Prism and the Collapse of N-Rays, which also does not allege dishonesty.
What the paper reports about N-rays
Stone's summary sets the discovery in 1903, a period of heavy experimentation with X-rays, and describes Blondlot as a respected member of the French Academy of Sciences. The source he describes is specific: a wire heated inside an iron tube with a window cut in it. Rays were said to leave through the window and to pass through aluminum.
The properties Stone lists are these:
- The rays could pass through 2" or 3" of aluminum but not through iron.
- They could be stored. A brick wrapped in black paper and left in sunlight was said to take them up and give them out again, and the effect did not depend on how many bricks were used.
- Many other things were said to emit them, people included.
- An aluminum prism was said to split them into components, and Blondlot measured an index of refraction for each.
Detection, in Stone's telling, meant looking at an object lit by a faint light. With N-rays present, the object was said to be much easier to see. No instrument other than the eye is described.
For the claims themselves Stone cites Blondlot's book The N-rays (Longmans, Green and Co., London, 1905). He does not quote it or give any of the refractive index values.
The prism, briefly
The paper tells the Nancy episode in three sentences. Robert W. Wood, an American physicist, visited the laboratory and was shown the experiments; while Blondlot demonstrated the refractive index measurement, Wood palmed the prism; the measurements were unaffected. Stone cites Wood's reports as Nature 72, 195 (1904) and Physik. Z. 5, 789 (1904). He calls the publication cruel and says it "was the end of Blondlot."
That phrase is a compression. The American Physical Society account summarised elsewhere in this dossier says Blondlot never conceded and went on working on the rays for years. Stone's paper gives no detail of the visit beyond the prism and does not say how quickly support fell away.
Pringsheim and the hand behind the paper
The part of Stone's account that goes beyond the prism story is an explanation of the error, attributed to a physicist he names only as Pringsheim, who tried to repeat the experiments and concentrated on the method of detection.
The setup, as Stone describes it, is a paper screen under very faint light. To check that the screen really is visible, the observer moves a hand back and forth near it; if the moving hand can be seen, the screen is taken to be illuminated. One of Blondlot's observations was that this became easier when N-rays fell on the paper.
Pringsheim's finding, as reported, was that the test worked just as well when the observer did not know whether the paper was in front of the hand or behind it. A hand held behind the paper was "seen" as readily as one held in front.
Stone's reading is that this is what a threshold phenomenon looks like. At the limit of vision, the observer cannot tell whether anything is being seen. What the observer does know is where their own hand is, and that knowledge is enough to tip an uncertain impression into a confident one. On this account the brightening reported in Nancy needed no rays, only an observer who knew what to expect and where to look. The paper gives no citation, date or numerical result for Pringsheim's work, so the test stands here as a described procedure and outcome, not a documented data set.
Langmuir's six symptoms
Stone then sets out Langmuir's list of the marks of pathological science. In paraphrase:
- The largest effect comes from a cause that is itself barely detectable, and the size of the effect hardly depends on the strength of the cause.
- The effect stays near the limit of detection, or needs many measurements because each has low statistical significance.
- Great accuracy is claimed.
- The supporting theories are fantastic and run against experience.
- Criticism is answered with excuses invented on the spot.
- The proportion of supporters to critics climbs towards roughly 50% and then declines into oblivion.
Stone does not go through this list point by point for N-rays; he applies it explicitly only to a later particle physics case. A reader can still see why the N-ray description sits next to it. A stored effect that is the same for one brick or many is the first symptom as Stone words it. Detection by judging a faint light is the second. Separate refractive indices for components of a ray that could only just be seen is the third. These matches are an inference from Stone's own description, not a scoring he performs.
The company N-rays keep
The other cases show what Stone thinks N-rays are an example of.
The Davis-Barnes effect, from Columbia University circa 1930, involved counting by eye, through a microscope in a dark room, the flashes that α particles made on zinc sulfide screens. Electrons were reported to bind to the α particles at 590 V and at other sharply defined voltages, in peaks 0.01 V wide. Langmuir visited, had the assistant alter the voltage without the observer knowing, and found that the counts no longer bore any relation to it. The authors later published a retraction granting that visual counts of a threshold phenomenon may be swayed by suggestion, and reporting that a Geiger counter had not confirmed the effect.
The parallel with Nancy is close: a faint visual signal, a dark room, a visitor who changes a condition unknown to the observer, and a result that carries on regardless.
The remaining examples come from particle physics and involve detectors, not eyes: an apparent split in the A2 meson peak around 1300 MeV, reported by two CERN experiments and absent from discussion by 1972; a group of bosons labelled R, S, T and U that were never confirmed; and the F meson, accepted at 2020±10 MeV on weak signals until the CLEO experiment in 1983 placed it at 1970±5±5 MeV. Stone's point is that removing the human eye does not remove the problem.
What the paper does and does not establish
Stone's conclusions are practical: doubt new results until they make sense, hide the final answer from analysts until systematic checks are finished (a practice he attributes to the KTeV, BaBar and Belle experiments), and have separate groups reach results independently.
On N-rays specifically, the paper is a brief secondary account. It contains no new evidence, no measurement and nothing from Blondlot's side beyond the book reference. Some details differ from other retellings: the American Physical Society piece lists metals generally among the materials the rays were said to pass through, while Stone has them stopped by iron. The paper does not address how many papers were published or why so many observers agreed with each other. Its contribution is narrower: a proposed mechanism, the unreliability of judgments made at the threshold of perception, and a checklist against which the N-ray claims can be read.
Further reading
- The N-Ray Affair in Overview: Scale, Failed Replications, Priority Claims and the Historical Literature takes up two points left open here: it gives counts for the papers and the scientists involved, and it cites Wood's Nature report under a different volume and page from the one Stone gives. Its section on finding the cited studies traces those counts to Lagemann's 1977 paper and gives the identifiers for Wood's report.
Source notes & attribution
- https://arxiv.org/pdf/hep-ph/0010295