Barents Sea, 18 June 1928: the last flight of Roald Amundsen

At 18:45 local time on 18 June 1928, the telegraph operator at the Geophysical Institute in Tromsø logged a French flying boat trying to raise the radio station at Kings Bay, on the far side of the Barents Sea. The station did not reply, because the station did not hear it. There is no record of any further transmission from the aircraft. Two and a half hours earlier it had taken off from Tromsø harbour with six men aboard, bound for Spitsbergen to join the rescue of another expedition that had also gone missing. The last thing the Latham 47.02 ever did, in the documentary record, was attempt a call that no one answered.

This is the loss the radio logs preserve as an absence. No distress signal was ever positively confirmed. No eyewitness saw a flying boat in trouble. A long Norwegian summer of searching turned up nothing, then a damaged float washed up off Torsvåg, then a fuel tank on Haltenbanken, then a second tank in Lofoten the following winter. Ninety-seven years later, an autonomous underwater vehicle towed across the seabed north of Bear Island found nothing more. The aircraft and the six men are still in the Barents Sea. We keep three things separate, as always: what is documented, what the evidence shows, and what is still only a hypothesis.

The documented account

The flight was a rescue mission, and an unlikely one. Roald Amundsen, then 55, had spent two decades as the most famous polar explorer alive: the first ship transit of the Northwest Passage aboard Gjøa from 1903 to 1906, the first expedition to reach the South Pole on 14 December 1911, and the 1926 transpolar flight of the semi-rigid airship Norge from Spitsbergen to Alaska with the Italian airship designer Umberto Nobile and the American financier Lincoln Ellsworth. By 1928 he had retired from polar work, his recent projects refused state funding, and his relations with Nobile had collapsed in a public credit dispute over the Norge flight.

In May 1928 Nobile flew north again, without Amundsen, in a new Italian semi-rigid airship called Italia. Italia launched from Ny-Alesund on Spitsbergen on 23 May, reached the North Pole on 24 May, and turned south. At about 10:33 local on 25 May the airship lost altitude and its control car struck the polar ice northeast of Spitsbergen. One crew member died at the impact. Nine men and supplies were thrown clear onto the ice; the lightened envelope, with six men aboard, rose, drifted off, and was never seen again. The men on the ice put up the dyed canvas later known as the Red Tent. On 3 June 1928 the Soviet amateur radio operator Nikolai Schmidt picked up the first SOS to reach the outside world. The international rescue race began.

Amundsen volunteered. The aircraft and the crew came from France: a prototype Latham 47, registration 47.02, the only Latham 47 produced beyond the original prototype, then being prepared at Caudebec-en-Caux in Normandy. The Latham 47 was a twin-engine biplane flying boat for the French Marine Nationale, powered by two Farman 12We engines. It had not finished its flight test programme. Six men were aboard for the run north. Capitaine de corvette Rene Cyprien Guilbaud of the Marine Nationale was the commander and pilot, Capitaine de corvette Albert Cavelier de Cuverville the co-pilot, Gilbert Brazy the mechanic, and Emile Valette the radio operator, all four Marine Nationale. The Norwegian Air Force pilot Leif Dietrichson, a veteran of Amundsen’s 1925 N25 and N24 polar attempt, was the navigator. Amundsen flew as expedition leader and passenger.

The aircraft left Caudebec-en-Caux on 16 June 1928 with the French crew of four, flew to Bergen on 17 June, picked up Amundsen and Dietrichson there, and continued north to Tromsø. At about 16:00 local on 18 June 1928 it lifted from Tromsø harbour and turned for Spitsbergen.

The radio logs from that afternoon, preserved in the Norwegian coastal stations, are sparse and consistent. At 17:40 local the Ingoy radio station on Masoy in Finnmark received a message signed by Guilbaud, asking for any available ice reports. About fifteen minutes later the Ingoy operator overheard the Latham trying to call the station at Longyearbyen on Svalbard; no one at Longyearbyen heard it, and no reply went back. At 18:45 local the Tromsø Geophysical Institute logged the Latham attempting to contact Kings Bay. That is the last reliable record of the aircraft. A frequently repeated fragment to the effect of asking the listening stations not to stop listening appears only in some secondary accounts; we do not present it as verbatim from the Latham.

The search began when the Latham failed to reach Kings Bay. Norwegian and French vessels worked the Barents Sea between Tromso, Bear Island, and Svalbard through the summer and autumn of 1928. The weather on 18 June had been variable, bright sun at takeoff with a fog bank reported on the projected track. No further contact came. On 12 July 1928 the Soviet icebreaker Krasin reached the Red Tent and lifted off the remaining survivors, the rescue Amundsen had set out to join. The Latham was not among the things the searchers found.

What came back, instead, came back in pieces. On 31 August 1928 the crew of the Norwegian fishing vessel Brodd, working off Karlsoy near Torsvag lighthouse north of Tromsø, recovered a damaged float. It was identified as the port float of Latham 47.02 by officers from the research ship Michael Sars and the French transport Durance and by the French consul. It was 2.32 metres long, blue-grey, with a 20-centimetre gash on the front left and a wooden patch on a strut, violently torn from the airframe. It was shipped to Bergen and then to Paris. On 13 October 1928 the fishing boat Leif found a blue-grey aluminium fuel tank afloat on Haltenbanken west of Namsos. It was identified as the forward internal fuel tank of Latham 47.02, the one nearest the hatch behind the cockpit, and a vent pipe on its lid had been sealed with a hand-cut wooden plug bearing knife marks. On 10 January 1929 the fisherman Martin Jorgensen found a second Latham 47.02 fuel tank washed ashore at Borge in Lofoten, now on display at the Polarmuseet in Tromsø.

No bodies were ever recovered. No further confirmed wreckage has surfaced. The float was examined by the French Ministry of the Navy; the secondary literature refers to a Marine Nationale examination, but the precise findings of any French naval inquiry have not been pinned to a published report in the open record. Norway observed a two-minute silence at midday on 14 December 1928, the anniversary of Amundsen’s South Pole arrival, and Fridtjof Nansen delivered the memorial address. On 21 June 1931 a memorial to the Latham 47.02 crew was inaugurated at Caudebec-en-Caux. Amundsen was subsequently declared legally dead.

The evidence

The case has very little evidence. There are radio logs that record the absence of a reply. There are three pieces of an aircraft that drifted ashore in the months after the loss. There is a modern sonar survey that found nothing. None of it shows, by itself, what happened between 18:45 and the moment the aircraft entered the water.

The radio record is the documentary spine. The 17:40 Ingoy log establishes that Guilbaud was on the air and asking the right operational question, namely ice. The 17:55 attempt to call Longyearbyen, overheard at Ingoy but not received by the addressee, establishes that the aircraft was still flying and within signal range of the Finnmark coast. The 18:45 attempt to call Kings Bay establishes that the crew still believed Spitsbergen was a feasible destination. The limit matters: these are three transmissions logged by their senders, not by their intended receivers, and none of them carries a position fix or a distress code. They show that the aircraft was airborne and trying to communicate. They do not show where it was.

The recovered port float is the most direct physical evidence. Its provenance is firm: recovered by a named vessel at a known location, examined by a multinational group of officers, identified by the same Bergen workers who had repaired the aircraft on its stop two months earlier. The float had been violently torn from the airframe and carried a 20-centimetre gash and an in-field wooden patch on a strut. That shows a major separation of the port float from the airframe, consistent with a hard water impact. The limit is sharp. The damage cannot be dated against the moment of loss; the float could have been torn off in the impact, in a forced landing on heavy seas, or in a sequence of events on the water after a landing.

The forward fuel tank, found six weeks after the float, is the case’s most discussed object. Its identification as Latham 47.02 hardware is firm. The feature that has carried decades of speculation is the hand-cut wooden plug sealing the vent pipe on its lid. The 1928 examiners and later commentators have read it as the crew, with one float destroyed, sealing the tank to use it as an improvised replacement float. If correct, at least some of the crew survived the initial event and worked the airframe in the water before a second event sank them. What the artifact shows is that the vent was deliberately stoppered with a hand-cut plug. The inference from object to action is not closed; a stowage or manufacturing detail cannot be excluded. The 2015 peer-reviewed reassessment by Karen May and George Lewis in Polar Record treats the improvised-float reading as one plausible reconstruction, not as established fact.

The modern search work runs out at the same place. In August 2009 the Royal Norwegian Navy minelayer KNM Tyr operated a Kongsberg HUGIN 1000 MR autonomous underwater vehicle with synthetic aperture sonar over about 34 square nautical miles north of Bear Island, a sector chosen on drift modelling supplied by the Norwegian Aviation Museum. The sonar narrowed to between 15 and 25 targets, some near the size of a Latham 47 engine. ROV follow-up cleared each one as a natural seabed feature or fishing-gear debris. The search ended without finding the aircraft. The honest reading of the result is narrow: a defined box, chosen on the best available drift model, did not contain the wreck. The Barents Sea is larger than that.

A Norwegian fisherman is reported in secondary accounts as having watched the Latham climb to clear a fog bank shortly after takeoff and disappear into it, the last visual record of the aircraft. The original press source has not been located in the open record. We report it as a secondary account.

The theories

Six readings of the loss are in circulation, four of them serious. None has been established by the evidence. The honest practice is to lay them out with what they do and do not explain.

Catastrophic icing and a forced landing. The Latham 47 was a flying boat without de-icing equipment, and the Barents Sea in mid-June can produce sea fog and rapid superstructure icing. For: the weather, the fisherman’s account of the aircraft climbing into a fog bank, the violently torn port float. Against: an experienced crew making a controlled forced landing would normally send a distress call, and none was confirmed.

In-flight structural failure of the prototype. The Latham 47.02 had not completed its flight test programme. The May and Lewis 2015 Polar Record analysis treats the decision to deploy an unfinished prototype on an Arctic mission as a major risk factor in itself. For: prototype status; the violent condition of the recovered float. Against: the float damage cannot be dated against impact, and the same signature fits a hard landing as well as an in-flight break-up.

Fuel exhaustion or navigational error in fog. A long over-water leg in poor weather, with no land reference and no clear ice picture, could produce a drift off-track and a fuel exhaustion short of any coast. For: Guilbaud’s 17:40 request for ice information; the absence of any distress call, consistent with a slow degradation rather than a sudden catastrophe. Against: the 18:45 attempt to call Kings Bay suggests the crew still believed they were within range of Spitsbergen.

Engine failure with marginal single-engine performance. Contemporary and later commentary describes the Latham 47.02 as heavily loaded for its role. For: a heavily-loaded prototype, two Farman 12We engines worked at the limit of their rating. Against: a single-engine failure in cruise would normally allow time for a distress call.

The improvised-float reading. The sealed forward fuel tank, with its hand-cut wooden vent plug, has been read as the crew’s attempt to replace the lost port float. If correct, the crew survived the initial event, worked the aircraft in the water, and were lost in a second event later. For: the physical condition of the tank. Against: the inference is built on a single artifact, and a stowage or manufacturing explanation cannot be excluded.

Mid-air fuel-line fire or explosion. Occasionally raised in popular commentary; flagged as fringe. For: would explain the abrupt loss of contact. Against: nothing in the recovered debris supports it, and the serious literature does not advance it.

The conspiratorial readings, that Amundsen faked his death, that the loss was staged, are not supported by the record. The independent radio logs, the multinational identification of the recovered float and tanks, and the unanimous 1928 finding that the artifacts were from Latham 47.02 are inconsistent with any survival scenario. We record the readings only to disclaim them.

One contextual point. The 1926 dispute between Amundsen and Nobile over the Norge flight was public, and the Italian Royal Commission that later examined the Italia rescue returned findings critical of Nobile. That controversy is part of the historical record. The Latham 47.02 was lost on the rescue, not in the dispute, and this article does not re-prosecute the commission.

What remains unknown

The state of the case, ninety-seven years on, is unusually clean and unusually empty.

The location of the main wreckage is unknown. The 2009 HUGIN search was a careful negative over a defined box, not a closure of the larger sea. The cause of loss is unknown. The four serious readings, icing and forced landing, structural failure of the prototype, navigational drift in fog, and engine failure, are all consistent with the surviving evidence, and the evidence does not decide between them. The manner of death of Amundsen, Guilbaud, Cavelier de Cuverville, Brazy, Valette, and Dietrichson is unknown. Whether the hand-cut wooden plug reflects a survival episode at sea, or has an innocent explanation, is unknown. The final sequence between 18:45 and the moment the aircraft entered the water has never been reconstructed from any source that survived.

What remains in the documentary record is a 17:40 request for ice reports, an overheard call to Longyearbyen that no one answered, an 18:45 attempt to call Kings Bay that no one heard, and silence. What remains in the physical record is a torn port float, a sealed fuel tank, and a second tank on a museum stand in Tromsø. The Latham 47.02 and the six men who flew it left those things behind, and the Barents Sea has kept the rest.

Barents Sea, 18 June 1928: the last flight of Roald Amundsen

The documented account

The evidence

The theories

What remains unknown

Sources

Primary

Secondary