The Hessdalen lights

The modern series began at the end of 1981, when lights started to show up in the valley; most were reported near the ground or above the hills, some high in the sky, and they could stand still, move slowly, stop, or move fast⁠[1]. The rate was highest from December 1981 to the end of 1984, then fell⁠[12]⁠[2]. The automatic station's statistics from 1998 to 2002 show the lights recorded more often in winter and between 22:00 and 01:00⁠[3]. Witness counts gathered from the project's reports decline from double figures a year in the late 1990s to low single figures in the 2010s and early 2020s⁠[8]. Reports continue: in 2025 and 2026 the project posted several unexplained camera events and, on 31 August 2026, a witness report matched by camera footage⁠[11].

The early accounts come from residents, collected by Project Hessdalen; the 1984 campaign logged 53 visual observations with times and quality indexes⁠[1]⁠[3]. Since then reports have come from visitors and from the project's field trips and science camps, and are posted on the project's observation pages by year⁠[13]. An amateur analysis in 2017 geocoded the reports with location certainty to map where lights are seen⁠[14].

The most recent documented case is typical in its detail. A witness stopped at Vårhuskjølen at 00:08 local time on 1 September 2026 saw a small white orb under the cloud to the west-south-west that "skipped" twice, growing larger and more orange each time, and faded after two to three seconds. A camera near the community house recorded a small light at the same time and direction; the witness confirmed it was what he saw, while noting that on video it looked smaller and briefer⁠[11].

The 1984 campaign deployed radar, a spectrum analyser, a seismograph, a magnetometer, a laser, a Geiger counter, an infrared viewer and cameras with gratings⁠[1]. Of 23 grating photographs, only three of the light were usable; Teodorani reports that they showed a continuum with no lines⁠[12]⁠[3]. The radar and magnetometer gave the most reliable results: radar tracks were often recorded with no visual sighting, and one light visible to observers appeared on radar only on every second sweep⁠[3].

The EMBLA expeditions of 2000–2002 added optical photometry, VLF-ELF receivers and a low-power radar. They reported radiant power up to 19 kW, clusters of light balls that appear and disappear, ejection of smaller balls, an unidentified frequency shift in the VLF band and a possible deposit of metallic particles⁠[3]. The 2002 report gave up to 100 kW, a light profile unlike a plasma, and spectra with a three-peaked feature that could be artificial lighting or blended lines, possibly of silicon⁠[15]. Leone reanalysed the same data and identified the recurrent blinking light as car headlights on a private road about 11.5 km away; he found the photometry based on a wrong distance and the temperature calculation in error by a factor of about 10⁷⁠[4].

The automatic station, the "Blue Box", has run since 1998 with cameras, a video alarm system written by the Italian Institute for Radio Astronomy, alarm systems from the University of Würzburg, and an aircraft transponder receiver that logs flights⁠[13]⁠[3]. In 2024 it was re-equipped with 8-megapixel night cameras; images are stacked and posted daily to a public folder and the cameras stream live⁠[10]. Geophysical campaigns from 2010 surveyed the ground: about 100 km of VLF traces over 100 km² found conductive zones, mainly sulphide deposits, outlining a 6 × 12 km ellipse that follows the gabbro intrusion⁠[9]⁠[16].

Misidentification. Some recorded lights are certainly ordinary: car headlights, aircraft and stars are present in the valley's sky, and the best-studied light of 2002 has a strong case for being headlights⁠[4]. The station's aircraft logger and video alarms exist partly to remove such sources⁠[13]. Misidentification does not account for the residents' close-range reports of 1981–1984 or for the 1984 radar tracks, as Leone himself accepts for some reports⁠[4]⁠[3].

Plasma and combustion models. Paiva and Taft proposed that the lights are dusty plasmas, and extended the model to a recorded spectrum⁠[5]⁠[17]; Paiva later proposed an electrically active temperature inversion over the valley⁠[7]. Hauge proposed combustion of dust clouds from the valley floor containing scandium⁠[18]⁠[6]. The EMBLA team compared the light profile to the silicon-nanoparticle model of ball lightning⁠[15]. None of these models has been tested quantitatively against the station's records.

Geology. The valley's sulphide and copper deposits have been proposed as a natural battery, with groundwater as electrolyte, supplying charge to the air⁠[8]. The 2024 VLF survey found the conductive bodies the idea needs and argues that near-surface structures could supply the lights' generation and explain why they occur in this valley⁠[9]. Against it, a geobattery delivers at most about a volt, roughly six orders of magnitude below what is needed to ionise air at any geological scale; a static source does not predict the decline in reports; and the stress-driven piezoelectric alternative was judged inapplicable because the valley's seismicity is too weak⁠[8].

Exotic proposals, such as micrometric wormholes, have been published as working hypotheses without physical tests⁠[19].

The station now produces the data a decisive study needs. A catalogue of every camera detection since 2024, classified automatically against aircraft logs, road traffic, astronomical ephemerides and weather, would show what fraction of detections is unexplained and whether that fraction is real. Triangulation from two or more cameras would give distance, size and true brightness, which no single camera can. Simultaneous spectra at useful resolution would separate plasma, combustion and nanoparticle models. And mapping the geocoded reports and unexplained detections against the 2024 conductive zones would test the geological reading directly.