Off the Farallones, 25 March 1921: the loss of USS Conestoga

On 23 March 2016, at the Naval History and Heritage Command at the Washington Navy Yard, NOAA’s Office of National Marine Sanctuaries and the United States Navy jointly announced that a wreck on the seabed of the Greater Farallones National Marine Sanctuary, three miles off Southeast Farallon Island and thirty miles southwest of the Golden Gate, had been positively identified as USS Conestoga (AT-54). She had sailed from Mare Island Naval Shipyard ninety-five years earlier, almost to the day, with fifty-six men aboard, bound first for Pearl Harbor and then for American Samoa. She had cleared the Golden Gate at 3:25 on the afternoon of 25 March 1921 and vanished. NOAA described her, before the announcement, as the last US Navy ship lost without trace in peacetime.

The announcement was the case’s pivot. After almost a century in which Conestoga’s loss had been a sustained absence in the Navy’s books, she was located. She was not, however, explained. The wreck had been found by NOAA’s Office of Coast Survey in 2009 and identified across a two-year archaeological investigation in 2014 to 2015. The 2016 announcement put her on the chart. It did not assign a cause. NOAA treats heavy-weather foundering as the most probable scenario, but ONMS continued, after the identification, to list Conestoga among the top maritime unsolved mysteries in US naval history. The wreck answered the where. It has not answered the why. We keep three things separate, as always: what is documented, what the evidence shows, and what is still only a hypothesis.

The documented account

Conestoga was a steel-hulled, single-screw, ocean-going steam tug, built in 1904 at Maryland Steel Company in Sparrows Point for the Philadelphia and Reading Railway. (NHHC’s DANFS entry gives the build year as 1903; Wikipedia and NOAA-cited summaries give 1904, and this article uses 1904 with the discrepancy noted.) She was 170 feet long, 29 feet in beam, 420 long tons, driven by a triple-expansion steam engine turning a four-bladed propeller of 12 feet 3 inches. The Navy purchased her on 14 September 1917, commissioned her on 10 November 1917, and reclassified her as AT-54 on 17 July 1920. Forward of the pilothouse she carried a 3-inch 50-caliber gun on a World War I pedestal mount.

In early 1921 she was ordered to American Samoa as station ship at Tutuila, Pago Pago. She departed Mare Island Naval Shipyard on 25 March 1921, cleared the Golden Gate at 3:25 p.m. local time bound for Pearl Harbor to refuel, and was due at Pearl Harbor on or about 5 April 1921.

She was commanded by Lieutenant Ernest Larkin Jones, age forty-one, a former enlisted sailor from Walnut, Kansas. (Jones held the rank of Lieutenant at the time of loss per NHHC; some popular accounts elevate him to Lieutenant Commander.) Her complement of fifty-six comprised Jones, three warrant officers, and fifty-two enlisted men. Several sources describe her sailing with a barge of coal for the long Pacific transit, but the exact tow configuration at the moment of loss is not established in the 1921 record and is treated here as uncertain.

The afternoon she sailed into was deteriorating. NOAA’s account, drawing on the weather log kept by the United States Lighthouse Service lightship San Francisco, records that the choppy morning seas of 25 March 1921 had given way to a gale by mid-afternoon, with winds reading 40 to 48 miles per hour by 4:00 p.m. and continuing past 5:00 p.m. NOAA’s reconstruction concludes that Conestoga steamed directly into building gale-force winds and heavy seas after clearing the Golden Gate, on a course that took her past the Farallones.

On 8 April 1921 a garbled radio transmission was logged from the eastern Pacific, later interpreted as a Navy vessel battling a storm with its barge torn adrift. The message was not at the time conclusively attributed to Conestoga. The Navy did not become aware that she was overdue until mid-May, and then launched what NOAA describes as the largest US Navy sea and air search up to that point, eclipsed only by the 1937 search for Amelia Earhart, roughly 300,000 square miles covered over eleven days, concentrated around Hawaii.

The only contemporary physical trace ever reported was a single derelict lifeboat. On 17 May 1921 the steamship Senator sighted a battered lifeboat with the brass letter “C” on her bow at 18 degrees 15 minutes north, 115 degrees 42 minutes west, about 650 miles west of Manzanillo and thirty miles off Clarion Island. NHHC records the lifeboat’s number, M5535 B, as not matching Navy Department records. The Senator did not recover it. The “C” boat was assumed to be Conestoga’s; the identification was probable rather than certain.

After all leads were exhausted, the Navy declared Conestoga and her fifty-six crew lost on 30 June 1921. She was investigated in the 1921 Navy inquiry that followed. The specific procedural form, the date, and the venue of that inquiry are not stated in the public NOAA and NHHC materials, and it is referred to here generically. NHHC notes that, lacking the ship’s deck log and any surviving witness, no cause of loss could be assigned in 1921. She then dropped out of the active record.

She returned to it in August 2009. A hydrographic survey by NOAA’s Office of Coast Survey, working within what is now Greater Farallones National Marine Sanctuary, logged an uncharted multibeam sonar target three miles off Southeast Farallon Island. (Secondary accounts attribute the survey platform to F/V Pacific Star; NOAA’s contracted vessels of the period also included Fugro Pegasus, and the attribution is given here as soft.) The depth was 189 feet. The target did not match any vessel known to have been lost in the sanctuary.

From September 2014 the NOAA Office of National Marine Sanctuaries Maritime Heritage Program ran a two-year archaeological investigation of the contact, co-directed by James P. Delgado and Robert V. Schwemmer in cooperation with NHHC, using multibeam sonar and ROV inspection. In October 2015 a confirmation mission with an NHHC archaeologist and senior Navy officers aboard verified the identification. The joint NOAA and Navy announcement followed on 23 March 2016. The wreck is protected as a sunken military craft under United States law and is treated as a war grave.

The evidence

Conestoga’s record sits in two parts that did not meet for ninety-five years. The 1921 file is thin: pre-departure records at Mare Island, the lightship weather log, the 8 April garbled transmission, the 17 May Senator sighting, and a Navy inquiry that could assign no cause. The modern file is thick: the 2009 sonar contact, the 2014 to 2015 ROV survey, and the published NOAA archaeological account. Together they establish that this is the ship and where she lies. They do not, between them, establish what failed.

The 1921 weather log. The United States Lighthouse Service lightship San Francisco kept a continuous weather log off the Golden Gate. NOAA reads it directly. The morning of 25 March 1921 was choppy; the afternoon built into a gale, with winds reading 40 to 48 miles per hour by 4:00 p.m. and continuing past 5:00 p.m. What the log establishes is the weather Conestoga sailed into within hours of clearing the Gate. The limit is that the log is a fixed observation at a fixed point. It is not the weather over the wreck site at the moment of loss, and it cannot fix the timing of the sinking.

The 8 April 1921 transmission. What the message establishes is that something was transmitting in distress in the eastern Pacific. The limit is decisive: it was not at the time conclusively attributed to Conestoga, and the attribution has remained inferential. It sits in the file as a piece of evidence consistent with one reading of the loss, not as a positive identification.

The Senator lifeboat report. The 17 May 1921 sighting placed a battered lifeboat with a brass “C” on her bow at 18 degrees 15 minutes north, 115 degrees 42 minutes west, far south of Conestoga’s expected track. What it establishes is a derelict consistent with a Navy ship’s lifeboat. The limits are decisive. The number, M5535 B, did not match Navy Department records, and the boat was never recovered. The “C” lifeboat is suggestive corroboration. It is not a positive piece of Conestoga.

The 1921 Navy inquiry. The Navy investigated the loss in 1921 and exhausted its inquiries that summer. NHHC’s summary is plain: lacking the deck log and any surviving witness, no cause could be assigned. What this establishes is the Navy’s own contemporary conclusion. The limit is that there was nothing to examine. The inquiry’s record is a record of an absence.

The wreck itself. The wreck lies in 189 feet of water in the Greater Farallones sanctuary, three miles off Southeast Farallon Island, oriented north-northwest. NOAA’s survey matched a series of features to Conestoga’s 1904 plans: hull dimensions; the four-bladed, 12-foot-3-inch propeller; the triple-expansion engine and boilers; portholes; mooring bitts and ventilators; a large steam towing winch with twisted wire on the drum; two porcelain marine heads; and the 3-inch 50-caliber gun on a WWI pedestal mount forward of the pilothouse. NOAA and the Navy treated the deck gun as decisive. What the wreck establishes is the identification of the ship. The limit is that the survey did not isolate a failure mechanism. The hull and machinery are read as consistent with heavy-weather foundering, but consistent with is not caused by. The wreck has named the contact. It has not given the loss a cause.

Eyewitness testimony. None to the sinking. No survivor reached land. Pre-departure observations at Mare Island and the 1921 lightship weather log are the closest contemporary observations to the event.

The throughline is the one the 1921 inquiry already named. The deck log went down with the ship, no one survived, no distress signal was conclusively heard, and the only contemporary debris was never brought aboard. The 2009 to 2016 work added the wreck. It did not add the log.

The theories

Everything in this section is a hypothesis. NOAA and the Navy continue to record Conestoga’s cause of loss as undetermined; ONMS lists her among the top maritime unsolved mysteries in US naval history even after the identification. None of what follows is established. They are set out roughly in the order the modern record gives them weight.

Hypothesis A: heavy-weather foundering on 25 or 26 March 1921. This is NOAA’s most probable scenario. A low-freeboard ocean tug, steaming into a gale within hours of clearing the Gate, overrun by seas and overwhelmed before she could make shelter. In favor: the documented weather; the wreck’s location only three miles off Southeast Farallon Island, consistent with very early loss; the absence of any recovered distress call; the low freeboard typical of the design. Against: no single failure feature was observed on the wreck. The reading is consistent with the evidence, not diagnosed from the hull.

Hypothesis B: mechanical, boiler, or steam-plant failure. A casualty in the engine room in heavy seas could produce a loss of power, a broach, and a capsizing. In favor: the vessel’s age and documented period failure modes in similar tugs. Against: NOAA’s survey did not report a specific mechanical-failure signature, and the engine and boilers as observed are read as broadly intact.

Hypothesis C: coal cargo or bunker shift under heavy weather. A shifting coal load in heavy roll could produce a sudden list and rapid capsizing. In favor: she was carrying coal for the long Pacific transit, and a shift of dense cargo is a documented sinking mechanism. Against: NOAA’s findings do not cite physical evidence on the wreck of a cargo shift.

Hypothesis D: capsizing under tow or a sudden barge casualty. NOAA cites the 8 April garbled transmission as describing a Navy vessel battling a storm with its barge torn adrift. In favor: tow configurations are documented in several sources; a barge breaking free in a gale can apply destructive loads to its towing vessel. Against: it is not definitively established that Conestoga was towing a barge at the moment of loss, and the 8 April transmission was not conclusively attributed to her. This hypothesis depends on a tow configuration the surviving record does not pin down.

Hypothesis E: loss of watertight integrity through riveted seams under wave impact. NOAA’s account describes a mechanism in which heavy seas overwhelmed the low-freeboard tug and flooded her through riveted seams. In favor: the design, the behavior of similar tugs, and the consistency with rapid loss and total silence. Against: this is not a discrete observable failure on the wreck. It is inferred from the design and the conditions, not a specific feature on the hull.

Hypothesis F: a combination. The most probable modern reading is a combination of the above: an aging, heavily-laden, low-freeboard tug driven into a sudden gale, possibly with tow complications, possibly with progressive flooding through her riveted seams, and overwhelmed before she could make shelter. In favor: each constituent reading is independently consistent with the documented weather, the wreck’s location, the absence of a distress signal, and the survey findings. Against: the combination is untestable as a sequence. It is the best available reconstruction. It is not a finding.

What remains unknown

The honest residue is this. A 170-foot Navy tug cleared the Golden Gate at 3:25 on the afternoon of 25 March 1921 with fifty-six men aboard, bound for Pearl Harbor and Samoa, and steamed into a building gale. She was never heard from again with certainty. A garbled transmission two weeks later may or may not have been her. A battered lifeboat sighted six weeks after that, with a brass “C” on her bow and a number that did not match Navy records, may or may not have been hers. In 2009 a NOAA survey logged the wreck three miles off Southeast Farallon Island; ONMS and NHHC matched her to Conestoga’s 1904 plans in 2014 to 2015; NOAA and the Navy announced the identification on 23 March 2016.

What is now known is that Conestoga lies in 189 feet of water within the Greater Farallones sanctuary, with her four-bladed propeller, her steam towing winch, and her 3-inch deck gun where she came to rest. What is not known is the exact moment of sinking, the exact failure mechanism, or whether she was towing a barge when she went down. NOAA places the loss on 25 or 26 March 1921, but no log, no distress call, and no eyewitness survives, and NOAA declines to assign a single cause. The manner of death of the fifty-six men aboard, now more than a century deceased and protected on site as a war grave, is not in the record.

The case is no longer undocumented. It is located, with an unfilled cause. ONMS continues to list Conestoga among the top maritime unsolved mysteries in US naval history. The wreck is a war grave. The file is open.

Off the Farallones, 25 March 1921: the loss of USS Conestoga

The documented account

The evidence

The theories

What remains unknown

Sources

Primary / official

Secondary / contextual