Reports of unidentified aircraft, lights and engine sounds spread through northern Sweden in the winter of 1933–1934 and returned around the turn of 1936–1937. They became known as spökflygare, or ghost fliers. The name described elusive supposed aircraft, not reports of the spirits of dead pilots. Military patrols, civilian inquiries and parliamentary debate followed.
The surviving public record is more complicated than either a confirmed invasion or a completely explained collection of illusions. A military review accepted a small subset of reports as credible, while searches failed to locate bases or establish landings. Critics questioned whether lights and sounds had been assembled into aircraft journeys through assumptions. The disagreement concerns the interpretation of reports, not evidence of an identified extraordinary vehicle.
A contemporary official account
The most useful starting point is the Swedish Second Chamber’s proceedings of 18 May 1937, protocol no. 32. Defence Minister Nilsson answered Elof Lindberg’s questions about alleged unauthorised flights over Upper Norrland. His opening answer, on printed pages 5–7, summarises the earlier investigations and the renewed reports of the preceding winter. It is a contemporary ministerial account of the files, not the complete underlying dossier.
Nilsson said inquiries had begun in late December 1933, with the defence and social ministries, air-force leadership and northern county governors in contact. Aerial reconnaissance followed in the first half of January 1934, initially in Västerbotten’s inland districts and then over lower Västerbotten and part of the Gulf of Bothnia. He recalled the prime minister’s February 1934 statement that no foreign bases or depots had been found in the searched areas and no landing of an unknown aircraft on Swedish territory had been established.
Other measures included air observation. According to the minister, those observation units also failed to see unauthorised flights. As reports diminished, the special arrangements were wound down, although some reporting continued. This is a negative result for particular searches and observation measures. It does not mean that every civilian report had been individually identified, nor does failure to intercept an aircraft establish exceptional performance.
What “12 credible reports” meant
Nilsson described a later General Staff review, submitted to the defence department on 8 July 1935. The material included personal reports of engine noise, lights and sometimes passing aircraft, together with interviews conducted by local officials. Reports from Norway and Finland were also considered. The review excluded material regarded as definitely wrong, less reliable or insufficiently substantiated.
Of 96 Swedish reports, twelve were judged fully credible. With selected reports from neighbouring countries, the accepted group totalled 46. Those figures belong to a particular historical selection process. They are not counts of recovered machines or independent instrument detections, and 46 should not be presented as the Swedish total alone. Equally, the rejected material should not automatically be described as deliberate fabrication.
The General Staff’s conclusion, as reported by the minister, was that aircraft of undetermined nationality had flown over northern Scandinavia, including Swedish and Norwegian territory, during the earlier winter. It proposed that much of the flying had occurred high above sparsely populated areas and that some journeys were circuits of varying extent. The minister acknowledged that such heterogeneous material was difficult to assess and could lead different reviewers to different conclusions.
The return of reports in 1936–1937
The minister said no reports of unauthorised flying in Norrland had arrived during 1935 and 1936 before the renewed concern around the year’s end. Military authorities then cited observations of light and sound phenomena and, in some accounts, aircraft around Norrbotten. The number was substantially smaller than in 1934. Regional military leadership and the General Staff again considered that foreign flying had probably occurred.
Special searchlights, listening apparatus and personnel for training were provided. Nilsson said that these measures produced no observations of foreign aircraft, while reports soon decreased and ceased. There had not been a new systematic scrutiny of the incoming material comparable to the earlier review. Listening equipment is not radar evidence; the protocol does not turn these reports into radar-tracked incursions.
Nilsson declined to reopen the older material merely because several months or years had elapsed. He nevertheless promised a broader inquiry if new reports began arriving. That administrative decision is worth keeping separate from an explanation. A refusal to repeat an investigation is not a demonstrated identification of the reported phenomena, and his promise of future scrutiny did not confirm their extraordinary nature.
Lindberg’s challenge to the reconstruction
Lindberg’s reply disputed the strength of the official conclusion. In the passages on printed pages 9–12, he returned to early reports from the Sorsele area. He contrasted an initial newspaper story of repeated aircraft landings on a mountain plateau with the witness’s later qualification that, in darkness and poor visibility, he had seen a moving light and could not be sure it was an aircraft.
Lindberg also discussed a gathering at Vilhelmina in which people initially took a conspicuous light for an aircraft before recognising a planet after moving upstairs and extinguishing indoor lights. This is a reported example advanced in a parliamentary argument, not an independent re-analysis performed for this article. It illustrates why a light’s apparent behaviour and the observer’s interpretation have to be distinguished.
In the closing passages on printed pages 19–20, Lindberg challenged the way flight routes were inferred. High altitude could be invoked to explain missing observations between towns, circuits to bridge timing gaps and extra aircraft to reconcile incompatible reports. His criticism was that a reconstruction could become too flexible to test. That concern does not itself prove every report illusory, but it directly contests the inference of a coherent pattern of foreign flights.
The press and later historical research
Anders Berglund’s 2020 History C study, Spökflygarnas dagordning, examines 44 editorials published in 1934–1938. Its method and conclusions show how the reports became arguments about defence spending, military authority and security policy. Different political newspapers interpreted the same uncertainty differently. The study is an undergraduate historical analysis of editorial positions, not an aircraft-identification investigation or a controlled psychological study of the witnesses.
Clas Svahn’s archival article, Spökflygarens vinter, likewise distinguishes misidentifications from possible actual aircraft. It includes his 1989 interview with former observer Sven Karlsson, who recalled mistaken interpretations of stars as well as hearing engines without seeing aircraft. Svahn also proposes German reconnaissance as a possible source for some flights. The reviewed article does not supply authenticated sortie logs matching every accepted report, so that proposal is retained as an interpretation.
Assessment, scope and map
The case is classified as Disputed. A contemporary military assessment supported unidentified aircraft activity, while contemporary critics challenged the evidence and reconstruction. Observational mistakes and possible ordinary aircraft need not be mutually exclusive within a collection of reports. No single demonstrated cause for the whole episode is established here, and disagreement over conventional aircraft does not provide positive evidence of non-human technology.
The numerical year list covers 1933, 1934, 1936 and 1937. The 1935 assessment and later publications are source dates, not additional sighting years. This case is distinct from the 1946 ghost-rocket reports; the separate source trail should not be merged simply because both names begin with “ghost”.
The map uses Boden as a representative locality for the official dispute, rounded from GeoNames entry 606531 to 65.83, 21.69. It does not identify a flight track, searchlight, bunker, home or exact sighting position. The modern station photograph is locality context only. The cited ministerial answer and selected portions of Lindberg’s reply were read directly, alongside the complete Svahn article and Berglund’s method, research-context and conclusion sections. The underlying military files and all 44 original editorials were not independently audited.