In 1922 the United States Geological Survey sent geologist George Rogers Mansfield to a ridge in the North Carolina foothills to kill a ghost story. He spent two weeks doing it properly. He set up surveying stations, sighted on every light that appeared with an alidade, took azimuth readings, and plotted each one against a map showing the rail lines, the roads, and the homesteads in the valleys beyond Brown Mountain. His verdict, later reissued by the USGS as Circular 646, was blunt: the lights were 'clearly not of unusual nature or origin.' He attributed roughly 47 percent to automobile headlights, about a third to locomotive headlights, and the remainder to stationary lights and brush fires. Case closed.

It was good science, and it should be respected as such. Mansfield even built in the right caveat for a 1922 investigator: a 1916 flood had knocked out the local rail line and roads for a stretch, and people swore they saw the lights during the outage. The headlight theory has to explain reported sightings from before cars dominated those valleys, and that is exactly where it strains. Decades later the explanation got reinforcement: in 1977 a team from Oak Ridge National Laboratory aimed a distant arc lamp toward observers, and the white beam arrived as an orange-red orb that appeared to hover above the crest, a clean demonstration that atmospheric refraction can manufacture the classic effect. Most reported Brown Mountain Lights are, almost certainly, exactly what Mansfield said they were.

Most is not all, and that gap is where the case refuses to die. Daniel Caton, a physics and astronomy professor and director of observatories at Appalachian State University, has been chasing these lights since the mid-1980s. He is not a believer running on vibes; he is an astronomer who got tired of the hand-waving on both sides and decided to put instruments on the problem. With observatory engineer Lee Hawkins he installed automated cameras aimed at the ridge, running night after night, precisely so that any light could be checked against the known map of roads and rails that Mansfield used to dismiss them.

Then the cameras caught something. On the night of July 16 into July 17, in footage running from roughly 9:40 p.m. to 5:37 a.m., a stationary light appeared in the same spot four separate times across about twenty minutes, with no lateral motion between exposures. A car headlight sweeps. A train moves. This sat, blinked out, and returned to the same coordinates. Caton's public position has been carefully unspectacular: he does not claim it is paranormal, only that it does not match anything he can readily identify, and that it is worth more data, not less.

That is the whole tension of the case in one sentence. The skeptical explanation is correct about the overwhelming majority of sightings and was arrived at with real rigor, while a credentialed scientist using better tools than Mansfield ever had keeps recording a residue the explanation does not cover. The lazy move is to pick a side: either the government solved it and everyone since is fooling themselves, or the government covered up a wonder. Neither is honest.

The honest version is harder. Mansfield's report is one of the best pieces of skeptical fieldwork ever aimed at a 'haunted' phenomenon, and it deserves its standing. And a hundred years on, an Appalachian State astronomer with calibrated cameras pointed at the same ridge is still pulling out frames that the headlight-and-refraction model does not obviously explain. The mountain was supposedly debunked in 1922. So the unresolved question is the only one that matters: if it was really solved a century ago, why are serious scientists still aiming cameras at it, and why do some of the frames still come back wrong?