Marfa Lights
Partially explainedSummary
Stand at the viewing area on Highway 90 outside Marfa, look southeast across the desert, and on many nights you will see lights. They hover. They split in two. They merge. They drift, brighten, and go out.
In 2004 a group of physics students from Texas State University did the obvious thing, which nobody had done in a hundred and twenty years. They put cameras at the viewing area, and at the same time they watched the traffic on Highway 67, which runs across the flat about thirty kilometres away, on the far side of the desert.
Every time a car appeared on the highway, a Marfa light appeared. Every time a car went behind a rise, a Marfa light went out.
What is documented
The lights. Seen from the official viewing area on Highway 90, east of Marfa, looking southeast across Mitchell Flat toward the Chinati Mountains. They appear as glowing points, usually white or yellow, sometimes red. They appear to hover, to move laterally, to divide, and to combine. They are seen on many nights of the year and they are seen by a great many people.
The tradition. The lights are conventionally said to have been first reported in 1883 by a young cowhand named Robert Reed Ellison. The registry notes that this attribution is not well documented, that the account is a later one, and that origin stories of this kind are frequently retrofitted to establish that a phenomenon predates the obvious modern explanation for it.
The 2004 study. The Society of Physics Students at Texas State University, working with Karl Stephan, conducted a systematic observation.
They placed instruments at the viewing area. They simultaneously observed United States Highway 67, which runs from Marfa southwest toward Presidio, crossing the field of view at a distance of roughly thirty kilometres, on the far side of Mitchell Flat.
The correlation was direct and repeatable. A vehicle appears on Highway 67, and a light appears in the position where a Marfa light would appear. The vehicle passes behind terrain, and the light goes out.
They demonstrated it, on multiple occasions, and they published it.
Why the lights behave so strangely. The line of sight from the viewing area to Highway 67 passes low over a large flat expanse of desert.
Desert air on such a plain forms strong thermal layers, particularly at night when the ground cools rapidly and the layers invert. Light passing horizontally through those layers is refracted, and the refraction is not steady: it shifts, it magnifies, it splits an image into two, it makes a point of light appear to hover well above where it is, and it makes it wobble and swim.
A pair of headlights thirty kilometres away, seen through a shimmering wedge of desert air, does not look like a pair of headlights. It looks like a light that hovers, splits and merges.
Leading explanations
Vehicle headlights on Highway 67, refracted over Mitchell Flat. Established for the bulk of the sightings.
The 2004 observation is not an inference. It is a simultaneous measurement of the cause and the effect, and it is repeatable by anybody willing to sit in the dark with a camera and a view of the highway.
The residue. James Bunnell, an engineer who has monitored the flat for years with instrumented stations, has argued that a small number of the lights he has recorded do not correlate with traffic and have properties that headlights do not.
The registry records this claim and does not endorse it. It has not been independently confirmed, and it is the sort of claim that requires a great deal of confirming, because the null hypothesis, that everything is headlights, has been demonstrated for the great majority of events.
Anything predating cars. If lights were genuinely seen on Mitchell Flat before there were headlights, that would be significant. The evidence for it is a nineteenth-century attribution that first appears in the twentieth century, and the registry treats it accordingly.
What the popular version gets wrong
"They cannot be car headlights, because the lights were seen before cars existed." The claim that the lights were seen before cars existed rests on a single origin story of poor provenance. The claim that they are car headlights rests on somebody watching the cars and the lights at the same time.
"They move in ways headlights cannot." They move in ways headlights appear to move when their light travels thirty kilometres horizontally through layered desert air at night. The hovering, the splitting and the merging are not properties of the light source. They are properties of the atmosphere between you and it.
"Scientists have never been able to explain them." A group of undergraduates explained them in 2004, using cameras and a road map, and the reason it took a hundred and twenty years is that nobody had thought to look at the highway.
Current status
Partially explained. The mechanism for the great majority of sightings is established by direct simultaneous observation. A residue is claimed and is not established.
The registry keeps this record beside AR-0087 and AR-0032, because they are the same story in different clothes. The Devil's Kettle lasted eighty years because nobody put a flow meter in the river. The sailing stones lasted sixty because nobody put a camera on the playa.
The Marfa lights lasted more than a century because nobody turned around and looked at the road.
Sources
- Society of Physics Students, Texas State University, with Stephan, K. (2004). Simultaneous observation of the Marfa lights and of traffic on US Highway 67.
- Bunnell, J. Instrumented monitoring of Mitchell Flat and published claims of non-traffic events.
- Literature on atmospheric refraction, thermal inversion layers, and superior mirage over flat terrain.
Last reviewed: July 2026. Records are provisional. Where the evidence changes, the entry changes. Found an error? Tell us.