When faces began surfacing in the concrete floor of a peasant kitchen in Andalusia, the Spanish state did not laugh it off. A commission tied to the Ministry of the Interior cut a chunk of the floor out and sent it for analysis, and decades later a national research institute ran the concrete through granulometric, mineralogical, and chemical tests. That is the genuinely odd core of the Bélmez affair: not the faces themselves, but the fact that government-linked chemists were dispatched to study a haunted floor, and the lab results are still being fought over half a century later.

The story began on 23 August 1971, when María Gómez Cámara said a human face had formed on her own in the concrete floor of her home at Calle Real 5 in Bélmez de la Moraleda. The family destroyed it with a pickaxe and laid fresh concrete; a new face appeared. The local mayor forbade further destruction and ordered a section of the floor removed for study. Over the following thirty years, the Pereira-Gómez household reported a steady procession of faces, male and female, large and small, in varying expressions, drawing crowds of sightseers and a parade of parapsychologists, most prominently Germán de Argumosa and the German researcher Hans Bender.

The physical-evidence trail is real but tangled. A commission connected to the Interior Ministry examined the case early on, and its handling explicitly weighed fraud scenarios: pigmentation with a dark brownish substance, a mixture of soot and vinegar, or the action of a chemical compound on the concrete. Decades later, samples recovered in September 1990 by a team including Father José María Pilón were analyzed by the Instituto de Cerámica y Vidrio (ICV), a center within Spain's national research council CSIC. Ramos Pereira of the Spanish Society of Parapsychology reported that infrared photography of "La Pava," the first and most famous face, revealed added pigmentation, and that even brush bristles could be made out, evidence of paint.

That is the strongest case for fraud, and it is a serious one: a national ceramics-and-glass institute, infrared imaging, and an apparent brushstroke. Investigations over the years repeatedly converged on the conclusion that the faces were applied to the concrete, first with paint and later, plausibly, with an acid or oxidizing agent that would etch the image into the surface and let it "develop" over days, mimicking spontaneous appearance.

Here is why the debunk never fully closed. The testing history is a chain-of-custody nightmare. Early press reports described traces consistent with photographic silver salts; later ICV work reportedly found no paint in certain samples. The samples were tiny, thirty and sixty milligrams, recovered years after the fact from a floor that had been crowded with visitors, cut, recoated, and handled for decades. Different labs, different samples, different eras, and conflicting findings. When the evidence is that contaminated, "no paint found" and "brushstrokes visible" can both be true of different fragments without telling you who made the original faces or how.

The fair reading still tilts hard toward the mundane. Faces are the single image the human visual system is most ruthlessly tuned to find; a few stains in rough concrete will read as eyes and a mouth to anyone primed to see them. The phenomenon was conveniently confined to one family's home, persisted for the family's lifetime, and faded into outright debunked theatrics in later years, including discredited claims by the psychic Pedro Amorós after María Gómez died in 2004. The most economical explanation is human hands, paint, and pareidolia.

But the file does not close cleanly, and that is the durable strangeness. A national government took a haunted floor seriously enough to commission chemistry on it; the chemistry, run too late and on compromised samples, produced a contradictory record that neither side can use as a clean knockout. The faces are almost certainly made by people. The lingering question is smaller and more honest: with the evidence this degraded, can anyone actually prove it, or has the floor simply outlasted everyone who tried?