The Bélmez Faces

Summary

In August 1971 an image resembling a face formed in the cement kitchen floor of a house in Bélmez de la Moraleda, a village in Jaén province. The face was removed and the hole refilled with cement; another face appeared. Through the first year roughly one new face appeared each month.

Four documented material analyses between 1976 and 2014 found no paint or artists’ pigments, and where silver was looked for, none was found. Two of the three conventional mechanisms proposed were not supported by the tests made. No mechanism has been demonstrated for the original faces, and the analyses that exist carry limitations their own reviewers state plainly.

What is documented

The sequence. The first face appeared on the cement kitchen floor in August 1971, in a house occupied by María Gómez Cámara, then fifty-two. In early September her son removed it and refilled the hole with cement; a new face appeared on 8 September. On 2 November 1971 a face was cut out of the floor and mounted behind glass on a wall. On 2 December 1971 an excavation beneath the house found human bones at 2.80 metres, without a skull, and the grave was refilled. Through the first year roughly one new face appeared per month. A police unit from Madrid examined the faces for eight days; its results have never been released.

The earliest claim of an analysis. In April 1972 the parapsychologist Germán de Argumosa said he had commissioned chemical analyses that found no trace of paint, only the pure cement base material. No report of them exists, and Mayer and Ramiro de Pano suggest he may have meant the police tests.

The 1976 analysis. J. J. Alonso and colleagues at the Universidad Politécnica de Valencia examined six samples: cement from an eye, earth from 1.5 and 3 metres, and three splinters from a cement block. They used element analysis, fire reduction and solvent extraction. They found calcium, iron, zinc, potassium, silicon, strontium and chlorine in the cement. They found no silver. Zinc was low, and on that basis they excluded a painting medium. They did find a dark, coal-like organic component in the cement and in one earth sample, and their first conclusion was that the substance forming the faces was solely that organic material. Mayer and Ramiro de Pano note that the report does not say who collected the soil samples, and that it is unclear whether the carbon-like substance is in fact organic.

The 1991 analysis. The Instituto de Cerámica y Vidrio of the CSIC in Madrid examined two samples using granulometry, X-ray diffraction and mass spectrometry. The main component was calcite. The samples were too small for diffraction to resolve anything further. Zinc, lead and chromium were present at fractions of a per cent. The analysts’ own assessment was that although those elements do occur in paints, the quantities were far too low to be paint constituents, and that lead in particular would have been easily visible at the surface and would not have resisted abrasion.

The 1994 analysis. A second diffraction run, made specifically to remedy the sample-size problem, found calcite, vaterite, alpha-quartz, portlandite and dolomite: ordinary cement minerals. No silver salts. No artists’ pigment components.

The 2014 analysis. Sampling was witnessed by a notary. José J. Gracenea at the Universitat Jaume I used infrared spectroscopy to look for organic paint components such as resins, and electron microscopy to look for inorganic pigments. He reported no silver salts, no organic residues, and substances naturally present in the cement, attributing the marks to heterogeneity in the material and finding no sign of external manipulation. It covered a single face, "La Pelona", and Mayer and Ramiro de Pano call the talk of the faces in the plural an unjustified generalisation.

The 2014 reproduction trials. In the same programme of work, the three conventional explanations were tested on cement slabs. Cement solvent produced bubbles and white streaks that sealed the pores and did not match the faces’ surface. Hydrochloric acid at 24 per cent produced traces that did resemble the faces, but they faded after thirty-two hours. Silver nitrate was invisible on application and darkened under ultraviolet light only after about thirty-six hours. On the original face, no traces of brushes, tools or applied substances were found.

The sceptical reading. Brian Dunning wrote in 2010 that the faces "were shown to have been painted on the concrete floor, the first with paint and later with acid". Tort and Ruíz-Noguez published on the case in the Journal of the Society for Psychical Research in 1993, and Mayer and Ramiro de Pano note that Ruíz-Noguez still favours a light-sensitive, silver-containing substance, although no analysis found silver. They cite Joe Nickell as an example of sceptics dismissing the faces as man-made without closer examination.

What the investigator best known for the case actually said. Hans Bender of the Institut für Grenzgebiete der Psychologie und Psychohygiene never published a formal report on Bélmez. He wrote that technical obstacles had prevented reaching the intended highest level of documentary evidence.

The 2004 episode is separate, and is a dispute. María Gómez Cámara died in February 2004. New faces appeared afterwards in the house next door. In November 2004 a newspaper alleged that faces found that September had been faked, naming individuals and a method. The accused party published counter-analyses reporting that samples taken before a justice of the peace were examined by the University of Jaén, which found no organic matter, no oil and no silver nitrate. The Registry read the allegation only through the accused party’s own rebuttal and could not open the original article. It records a documented dispute, not a documented forgery.

Leading explanations

Painting or deliberate fraud (the sceptical position; not established by analysis). That the faces were painted or otherwise made by a member of the household. Brian Dunning states it directly, and Mayer and Ramiro de Pano cite Joe Nickell as an example of sceptics dismissing the faces as man-made. For the faces of 2004, a newspaper made a specific fraud allegation, which is contested. None of the four documented material analyses found paint, pigment or binder.

An acid or other agent applied to the cement (a conventional explanation tested in 2014; not supported by that test). That a chemical was applied and the images developed after a latent period. The only test of the idea found that hydrochloric acid traces resembled the faces but faded within thirty-two hours, and that cement solvent did not match the surface at all.

Silver salts developed by light (José Luis Jordán, in the newspaper Pueblo, 1972; rejected). A silver-salt formula published in Pueblo in February 1972. No silver was found in 1976, none by diffraction in 1994, and none by spectroscopy or microscopy in 2014. The 2014 reproduction also showed that silver nitrate is invisible on application and takes about thirty-six hours to darken.

Heterogeneity in the floor material (Alonso, 1976; Gracenea, 2014; hypothesis, untested). That the marks arise from material naturally present in the cement, a dark substance on Alonso’s account and the heterogeneity of the material on Gracenea’s, making the images a property of the floor rather than something applied to it. It is consistent with the published analyses. It does not by itself account for the recognisable facial configurations or for their recurrence, and the origin of the dark substance is unresolved.

Paranormal production (Argumosa and Bender, 1970s; hypothesis, unsupported). That the images were produced by the occupant’s own psychic action. On the account of the historians who reconstructed the case, the planned high-grade documentation failed and Bender never published.

Current status

Partially explained. The published analyses do not support a classification as debunked. Four documented material analyses, from 1976 to 2014, found no paint or artists’ pigments, and the three that looked for silver found none. Mayer and Ramiro de Pano conclude that they support neither silver salts, nor paint, nor external manipulation of the surface. Two of the three conventional mechanisms were not supported by the tests made, and the acid variant produced marks that faded within thirty-two hours. Against that, every analysis has a real weakness. The 1972 claim of an analysis has no surviving report. The 1976 report does not say who collected the soil samples. The 1991 samples were too small to resolve beyond calcite. The 2014 work examined a single face while speaking of the faces. And the 1991, 1994 and 2014 analyses were all commissioned by people who favoured a paranormal explanation, a limitation the reviewers themselves state. The sceptical case rests on the pictorial character of the images, on journalistic investigation and on a contested allegation about a later set of faces, rather than on any laboratory identification of a medium. An independently commissioned analysis of several of the original faces by a conservation-science laboratory with no stake in the outcome, using cross-sectional microscopy and binder analysis on adequate sample volumes, would be decisive either way. So would release of the results of the Madrid police examination of 1971 and 1972.

Last reviewed: November 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.