Flight 19: the Navy training flight that vanished off Florida

At about ten past two on the afternoon of 5 December 1945, five US Navy TBM Avenger torpedo bombers lifted off the runway at Naval Air Station Fort Lauderdale and turned east over the Atlantic. They were Flight 19, an advanced training detachment, fourteen men in five aircraft, flying a triangular over-water navigation problem that should have brought them home in under three hours. The weather at takeoff was warm and bright. The flight leader had more than two thousand five hundred flying hours, six hundred of them in Avengers. The mission was a routine one the squadron had flown many times. None of the five aircraft and none of the fourteen men came back.

About an hour and a half into the flight, an instructor airborne over the Fort Lauderdale area intercepted a transmission from the flight leader, Lt Charles Carroll Taylor, USNR, reporting that both his compasses had failed and that he believed he was somewhere over the Florida Keys. Through the late afternoon the radio fragments grew more confused as a frontal system pushed south across the training area, raising winds aloft and bringing rough seas. At 19:27 a PBM-5 Mariner flying boat launched from NAS Banana River to search for them, with thirteen men aboard. It sent a routine “out” report at about 19:30 and then it, too, fell silent. Twenty-three minutes later, far out to the east, a merchant tanker watched a fireball collapse into the sea.

Coast Guard and Navy units searched more than 250,000 square miles of ocean over the next five days, with 248 Navy aircraft and eighteen surface vessels including the escort carrier USS Solomons. They found no Avenger wreckage, no Mariner wreckage, no life raft, no debris, no oil slick. In April 1946 the Navy’s board of investigation issued a report of about five hundred pages naming Lt Taylor as the cause of the loss. In 1947 the Navy revised that finding to “cause unknown” and exonerated him. The two acts, the naming and the unnaming, are the spine of the case. Both are institutional. Eighty years on, neither has been overturned by physical evidence, because there is no physical evidence to overturn it with.

The documented account

The exercise that afternoon was Navigation Problem Number One, a standard advanced-training route flown from NAS Fort Lauderdale. The flight was to fly east-southeast to Hen and Chicken Shoals, a practice bombing area north of Bimini, conduct a low-level bombing run, then continue east, turn north over Grand Bahama Island, and finally turn southwest to return to base. The planned route was about 316 miles. The flight was expected to last about two hours and forty minutes.

Takeoff had been scheduled for 13:45 and was delayed by Taylor’s late arrival, going at approximately 14:10. The five aircraft were FT-28, a TBM-3D, with Taylor leading; FT-36, a TBM-1E; FT-81, a TBM-1C; FT-3, a TBM-1C; and FT-117, a TBM-1C. The students were not novices. They were qualified naval aviators completing advanced training, and several were commissioned officers in their own right. Taylor was twenty-eight, USNR, combat-experienced, with 2,509.3 total flight hours and 616 hours on the Avenger. He had transferred to NAS Fort Lauderdale on 21 November 1945, two weeks before the loss, from Miami Naval Air Station.

The bombing run was completed at about 15:00. At roughly 15:40 to 15:50, an instructor flying near Fort Lauderdale, Lt Robert F. Cox, intercepted Taylor’s first call asking for help. Taylor reported that both his compasses were out and that he believed he was over the Florida Keys. He was not. The flight was east of the Bahamas, in the Atlantic, several hundred miles from the position he believed himself to be in. Through the afternoon Cox and shore stations picked up fragmentary UHF transmissions. The flight argued internally about which way was home. As late afternoon turned to evening, weather conditions in the training area deteriorated. By 16:00 winds at 1,000 feet were running at about 40 mph, with substantially higher winds aloft and squalls scattered across the area. Sunset came at about 17:30 local. By the time search aircraft reached the suspected ditching zone, conditions there were overcast and showery, with 25 to 30 knot surface winds and very rough seas. The last clearly attributable transmission from Taylor came at approximately 18:20. The last fragmentary signals heard from anyone in the flight were logged at around 19:04.

The PBM-5 Mariner BuNo 59225, of Patrol Bombing Squadron 49, call sign Training 49, launched from NAS Banana River (now Patrick Space Force Base) at 19:27 to search for Flight 19. It was commanded by Lt (jg) Walter G. Jeffery, USN. It transmitted a routine “out” report at about 19:30 and was never heard from again.

At 0050 GMT on 6 December, which is 19:50 local Eastern Standard Time on the evening of 5 December, the merchant tanker SS Gaines Mills, under Captain Shonna Stanley, logged the observation of an explosion at sea. The position was 28°59′N, 80°25′W, on the Mariner’s outbound track, and Stanley described leaping flames a hundred feet high burning for about ten minutes. He turned to search and reported steaming through a pool of oil and aviation gasoline without finding survivors. The PBM-5 was the only aircraft unaccounted for in that area at that time.

For five days the search continued. The Navy alone put 248 aircraft into the air, supported by 18 surface vessels and merchant shipping in the area. No wreckage from any of the five Avengers or from the Mariner was found. None has been found since.

The crews

The dossier convention of The Cold File is to name the dead. Twenty-seven men were lost on the evening of 5 December 1945 across the two aircraft losses. A twenty-eighth, Cpl Allan Kosnar, was on the original Flight 19 manifest and was not aboard.

FT-28, TBM-3D, BuNo 23307

• Lt Charles Carroll Taylor, USNR, 28, pilot and flight leader.
• George Francis Devlin Jr., AOM3c, USNR, 17, gunner.
• Walter Reed Parpart Jr., ARM3c, USNR, 18, radioman.

FT-36, TBM-1E, BuNo 46094

• Capt Edward Joseph Powers Jr., USMC, 26, pilot.
• Sgt Howell Orrin Thompson, USMCR, 20, gunner.
• Sgt George Richard Paonessa, USMCR, 28, radioman.

FT-81, TBM-1C, BuNo 46325 (two souls aboard)

• 2nd Lt Forrest James Gerber, USMCR, 24, pilot.
• Pfc William Earl Lightfoot, USMCR, 19, crewman.
• Cpl Allan Kosnar, USMCR, had been assigned as FT-81’s third crewman and asked to be excused that day. He was not aboard. He survived and died in 1990.

FT-3, TBM-1C, BuNo 45714

• Ensign Joseph Tipton Bossi, USNR, 20, pilot.
• Herman Arthur Thelander, S1c, USNR, 19, gunner.
• Burt Edward Baluk Jr., S1c, USNR, 19, radioman.

FT-117, TBM-1C, BuNo 73209

• Capt George William Stivers Jr., USMC, 25, pilot.
• Sgt Robert Francis Gallivan, USMCR, 25, gunner.
• Pvt Robert Peter Gruebel, USMCR, 18, radioman.

PBM-5 Mariner, BuNo 59225, NAS Banana River, 13 souls

• Lt (jg) Walter G. Jeffery, USN, pilot and commander.
• Lt (jg) Harrie G. Cone, USN, co-pilot.
• Ensign Roger M. Allen, USN.
• Ensign Lloyd A. Eliason, USN.
• Ensign Charles D. Arceneaux, USN.
• Robert C. Cameron, RM3, USN.
• Wiley D. Cargill Sr., Seaman 1st, USN.
• James F. Jordan, ARM3, USN.
• John T. Menendez, AOM3, USN.
• Philip B. Neeman, Seaman 1st, USN.
• James F. Osterheld, AOM3, USN.
• Donald E. Peterson, AMM1, USN.
• Alfred J. Zywicki, Seaman 1st, USN.

A monument bearing the names of the thirteen men of the Mariner was dedicated at Patrick Air Force Base, the former NAS Banana River, in 2017.

The evidence

The case has, in the strict sense, very little evidence at all. There is no physical evidence: no recovered aircraft, no engine, no instrument, no fragment confirmed as belonging to any of the five Flight 19 Avengers or to the Mariner. The evidence the case does have is documentary. The Navy preserved a record, and the Navy then revised the record. Both versions can be read today.

The radio fragments

The board of investigation reproduced or summarized a series of UHF transmissions from Flight 19 between roughly 15:40 and 19:04 on the afternoon of 5 December. The fragments below are the ones the board recorded; popular accounts add others that have no documentary standing. Three are worth quoting because the wording itself is the evidence.

Taylor’s first call to Lt Cox, at approximately 15:40 to 15:50, asked for help finding base:

Both of my compasses are out, and I am trying to find Fort Lauderdale, Florida. I am over land but it’s broken. I am sure I’m in the Keys but I don’t know how far down and I don’t know how to get back to Fort Lauderdale.

Sometime after 17:00 the board recorded Taylor telling the formation, in summary form, that he believed the flight had not gone far enough east. Soon after 18:00 he transmitted:

I suggest we fly due east until we run out of gas.

The last transmission attributed to him, at approximately 18:20, was a ditching order:

All planes close up tight. We’ll have to ditch unless landfall. When the first plane drops below ten gallons we all go down together.

There is no recorded transmission after roughly 19:04. The board did not preserve a final goodbye from any aircraft.

The Navy Board of Investigation

The board convened on 7 December 1945 at NAS Jacksonville, two days after the loss. Its full report, running about five hundred pages, was issued in April 1946. Its core finding, as the NARA Prologue blog (working directly from the file) summarizes it, was that Taylor had become “uncertain as to in what position lay the Peninsula of Florida”: he was over the Atlantic east of the Bahamas, he believed he was over the Florida Keys, and as a consequence he led the formation northeast, deeper into the Atlantic, rather than west toward the coast. The NAS Fort Lauderdale Museum biographical page for Taylor characterizes the board’s original verdict more starkly, recording that the board found him “guilty of mental aberration.” Where the two phrasings differ, the NARA reading of the original file is the safer ground. The reported “mental aberration” formulation is the museum’s gloss; the conservative paraphrase is the NARA one.

In the press and in derivative summaries the original finding was reported as pilot error attributable to Taylor.

The 1947 revision

In 1947 the Navy revised the finding. The cause of the loss of Flight 19 became “cause unknown.” The revision followed an appeal pursued by Lt Taylor’s mother, Mrs Katherine Taylor, who argued that with neither the bodies nor the wreckage recovered, the Navy had no basis on which to assign blame to her son. The mechanism of the revision was an action of the Board for Correction of Naval Records, which exonerated Taylor with respect to responsibility for the loss of lives and naval aircraft.

The 1947 act is the institutional contradiction the case is built around. The Navy had named a man and then unnamed him, using its own machinery in both directions. No new physical evidence had emerged in the intervening sixteen months. The revision was a re-reading of the same record by a different body inside the same institution.

The Mariner and the fireball

The PBM-5 Mariner BuNo 59225 has a likelier cause than the Avengers. The type had a documented history of fuel vapor accumulation in the bilges, prone enough that Navy crews nicknamed it the “flying gas tank.” The Navy board attributed the Mariner’s loss to a probable in-flight explosion. The SS Gaines Mills observation of a hundred-foot fireball at 28°59′N, 80°25′W at 0050 GMT on 6 December, consistent in time and location with the Mariner’s outbound track from NAS Banana River, supports that inference. Captain Stanley’s report of steaming through aviation gasoline and oil without finding survivors is consistent with it as well. None of that is the same as confirmation. No wreckage of the Mariner has been recovered either.

The two losses are a parallel mystery, not the same mystery. They are usually told as one because they happened the same evening, but the most defensible reconstruction holds the Mariner separately: a search aircraft, launched in worsening weather to find Flight 19, that probably exploded in flight within the first half hour, for reasons grounded in the type itself rather than the crew or the conditions.

Meteorology

The board record, the NHHC summary, and Larry Kusche’s 1975 reconstruction are in broad agreement on the weather. The afternoon began clear and warm at Fort Lauderdale and degraded as a frontal system moved south through the training area. By 16:00 the winds at 1,000 feet were about 40 mph, with substantially higher winds aloft, and squalls were scattered across the area. By the time search aircraft reached the suspected ditching zone, conditions were overcast and showery, with 25 to 30 knot surface winds, turbulence, and very rough seas. Sunset on that latitude in early December was about 17:30. Whatever Taylor was looking at after that hour, he was looking at it in the dark.

Eighty years of search

The absence of wreckage is the case’s most stubborn fact. The most prominent attempt to find Flight 19 was the Scientific Search Project led by the oceanographic engineer Graham Hawkes in 1991. Hawkes’s expedition, prospecting for sunken Spanish galleons off Fort Lauderdale, located five TBM Avengers under approximately 600 feet of water within about a mile and a half of one another, roughly ten miles northeast of the base. The aircraft bore “FT” tail markings indicating NAS Fort Lauderdale origin, and a partial “28” was visible on one of them. The initial announcement, on 8 May 1991, was that the find was the likely Flight 19.

Bureau Number checks, the Navy’s primary aircraft identifier, broke that identification within weeks. By early June 1991 UPI, the Baltimore Sun, and the Deseret News were reporting that the five recovered Avengers were not Flight 19, but five separate training-loss incidents whose aircraft had come to rest in the same area. Hawkes did not formally abandon the possibility in later interviews, but he acknowledged that no conclusive identification had ever been made. The actual Flight 19 has never been located. The find that briefly looked like it remains the case’s most-cited near-miss.

The theories

There is no theory of the loss that the surviving evidence can confirm. The hypotheses are these.

Compass failure and navigational disorientation in deteriorating weather. This is the Navy’s de facto leading explanation, the one the board reached for in 1946 before the 1947 revision and the one the documentary record most directly supports. The radio fragments establish that Taylor reported both compasses out and believed he was over the Florida Keys when he was east of the Bahamas. The meteorology establishes that a front overtook the area with high winds aloft and rough seas. The aircraft would have run out of fuel and ditched somewhere in the western Atlantic east of Florida in darkness and heavy weather. The objection to this reading is straightforward: the other pilots had functioning compasses, and the question of why none of them broke formation and turned west has no clean answer. Wartime training discipline, in which a flight follows its leader, is the standard one. The objection survives the answer. Even taking the navigational hypothesis at its strongest, the board’s eventual position in 1947 was that the cause was unknown, because no wreckage had ever been recovered to test any reconstruction against physical evidence.

An in-flight explosion of the Mariner. This applies to the Mariner only and is the leading explanation for that loss. The supporting evidence is the type’s documented fuel-vapor history, the SS Gaines Mills fireball at a position consistent with the Mariner’s outbound track, and the absence of any other aircraft unaccounted for in that area at that time. It is the most concrete explanation in the case, and it still does not have a recovered wreck behind it.

Rogue wave, waterspout, or severe localized weather. The Atlantic that evening was rough but not extraordinary by frontal-system standards. The board treated weather as the proximate cause of the ditching itself, not as a singular event striking the formation, and the available record supports a frontal-system reading rather than a freak local phenomenon. A waterspout-style explanation has no specific evidentiary support and is best treated as a speculative subset of the weather hypothesis.

The Bermuda Triangle and the paranormal. Flight 19 is the foundational case of the Bermuda Triangle legend, and the framing post-dates the loss by two decades. The phrase “Bermuda Triangle” was coined by the writer Vincent Gaddis in a February 1964 article in Argosy magazine, “The Deadly Bermuda Triangle,” which he expanded into the 1965 book Invisible Horizons. Charles Berlitz’s 1974 mass-market book The Bermuda Triangle propagated the framing globally and is the primary vector for several apocryphal Flight 19 quotations that do not appear anywhere in the Navy board file. The framing is cultural overlay, named to its sources, and not evidence. The contemporaneous record contains none of it.

UAP or extraterrestrial involvement. Absent from the contemporaneous record, absent from the Navy board file, absent from the preserved radio fragments, and an artifact of 1960s and 1970s popular literature rather than the 1945 case. It is reportable as cultural elaboration that the historical record does not support. It is not reportable as evidence.

What remains unknown

The most accurate summary of Flight 19 is that the Navy first attributed the loss to Lt Taylor, then attributed it to no one, and then never resolved the question physically. The first verdict survives in the press of the period and in popular memory. The second verdict survives in the institutional record. Eighty years on, no wreckage from any of the five Avengers has been confirmed, and no wreckage from the Mariner has been confirmed either. Twenty-seven men are still at sea. The lieutenant the Navy named is still the lieutenant the Navy un-named.

What is genuinely unknown is everything that would resolve the case. Where did Flight 19 ditch. Why did the other pilots not break formation. Whether the Mariner exploded for the reason the board suspected. Whether any of the five Avengers Hawkes found in 1991 could yet, with better identification, prove to be one of them. None of these can be answered from the documents alone. They will be answered, if they are answered at all, by wreckage. The framing as a Bermuda Triangle case, which arrived in 1964 with Gaddis and globalized in 1974 with Berlitz, is two decades younger than the loss it claims to explain. It explains nothing, and it does not need to be argued with. The case is older than the legend, and it is the case that is still open.