Evidence of Endeavour

Presenting compelling new evidence identifying a shipwreck site in Rhode Island as HMB Endeavour.

James Cook’s Endeavour is one of the most significant vessels in global history. Recently, its shipwreck site was identified in the waters of Rhode Island in the United States – far from the Pacific Ocean where both Cook and Endeavour earned their extraordinary reputations.

Combining maritime archaeology, 250-year-old archival records and cutting-edge research, what follows below is the latest evidence used to identify a shipwreck site (known as RI 2394) in Rhode Island’s Newport Harbor as Endeavour.

Delve into the maritime archaeology team’s research and learn what lines of evidence were used to identify RI 2394 as Endeavour after more than 20 years of determined pursuit. Examine the key findings yourself and peel back the centuries to reveal what remains of this iconic shipwreck.

Between 2018 and 2021, the Australian National Maritime Museum’s maritime archaeology team investigated an eighteenth-century shipwreck site in Newport Harbor, Rhode Island with members of the Rhode Island Marine Archaeology Project (RIMAP), a non-profit volunteer organization, and professional maritime archaeologists from the Silentworld Foundation and Spritsail Enterprises. The site, known by its Rhode Island state archaeological site number RI 2394, is the largest of four eighteenth-century shipwreck sites found within the Limited Study Area (LSA), a section of the harbor between the northern end of Goat Island and Newport’s historic North Battery (indicated by oval).

In 2015, the museum’s former Head of Research, Dr Nigel Erskine, identified this area as the location where the British transport Lord Sandwich, formerly HMB Endeavour, was scuttled (intentionally sunk) as a blockship during the Battle of Rhode Island in August 1778. The battle was a defensive action by British and Hessian troops to prevent a combined American and French expeditionary force from capturing Newport during the American War of Independence.

Prior to commencement of investigations within the LSA, the museum’s maritime archaeology team and its RIMAP counterparts developed and agreed upon a list of comparative criteria between the historical and archaeological records that, if met, would provide sufficient evidence to identify one of the four British transport shipwreck sites as Lord Sandwich (ex-HMB Endeavour). Based on this ‘preponderance of evidence’ approach, the museum’s maritime archaeology team have now identified attributes of RI 2394’s surviving hull that closely or exactly match features of Endeavour’s design and construction recorded in historical sources. These criteria - discussed in detail below - provide definitive evidence that RI 2394 comprises the remnants of Endeavour.

One of the earliest indicators that RI 2394 might be Endeavour was the size of the shipwreck’s visible timbers. It was clear from the outset that RI 2394 represented the remnants of a relatively large, wooden-hulled sailing vessel. Archival accounts of the lead-up to the Battle of Rhode Island revealed Lord Sandwich was the largest of the five British transports intentionally scuttled as blockships in Newport Harbor between Goat Island and the North Battery. Detailed records of Endeavour’s design and construction are contained in surveys of the vessel conducted by the British Admiralty following its acquisition for naval service in 1768, and prior to it being sold out of naval service in 1775. Among this bounty of archival sources is a list of scantlings (height and width measurements) for timbers used in its construction.

The scantling list records the average sided dimension (width) for Endeavour’s floor timbers as 1 foot, 2 inches (14 inches, or 35.6 centimetres), and that for the first futtocks as 11 inches (27.9 centimetres). Floor timbers, or floors, form the lowest component of a ship’s individual frames (or ‘ribs’) and cross the keel at the centreline. A first futtock forms the next section above the floor in a composite timber frame. RI 2394’s surviving hull structure features both floors and first futtocks, and their respective sided measurements—averaging 14 ¼ inches (36.2 centimetres) for floors and 10 ¾ inches (27.3 centimetres) for first futtocks—compare favourably with those listed in Endeavour’s 1768 survey. The average thickness recorded for hull planking on RI 2394 is 3 inches (7.6 centimetres). This correlates exactly to the thickness listed for Endeavour’s ‘plank of bottom from the floorheads to [the] keel’ in the 1768 survey. Finally, the moulded dimension (height or depth) noted in the Admiralty survey for Endeavour’s keel below the rabbet (the notch cut into the top of the keel to accommodate the edge of the first hull plank, or garboard strake) is 11 inches (27.9 centimetres). The maximum recorded moulded depth for RI 2394’s keel at the bow is 14 inches; however, when the height of the rabbet (3 inches, or 7.6 centimetres) is subtracted, the remaining depth—11 inches—is identical to what was recorded for Endeavour’s keel in 1768.

Another significant line of evidence connecting RI 2394 and Endeavour are the types of timber species used to construct the hull. Eighteenth-century British shipwrights preferred to use English oak (Quercus robur) in the construction of framing components, while English or Dutch elm (Ulmus procera or Ulmus hollandica) were desired as keel timber. English oak is an incredibly dense and durable timber that is relatively rot resistant. Old growth trees tend to produce large, robust limbs with curves and angles ideally suited for ‘compass timber’ used to form the framing and upperworks of ships. English or Dutch elm were also ideal for shipbuilding, as they too are dense and durable, but tend to grow tall and straight, and are also rot-resistant when completely submerged. This made both species of elm ideal for ship’s keels and hull planking beneath the waterline.

By contrast, shipbuilders in Great Britain’s North American colonies tended to use a diverse range of domestic timber species that were abundant and much easier to access than the European oak and elm species used in British shipyards. North American timber species that featured in colonial-built British vessels include southern live oak (Quercus virginiana), hackmatack (Larix laricina), Carolina - or shortleaf - pine (Pinus echinate), eastern red cedar (Juniperus virginiana) and American white oak (Quercus alba). Of the five British transports scuttled between Goat Island and the North Battery, three - Lord Sandwich/Endeavour, the 250-ton ship Yowart and 160-ton snow Mayflower - were British built and almost certainly utilised British and European timber in their construction. The remainder, comprising the 200-ton ships Earl of Orford and Peggy, were built in North America, very likely of native timber.

Timber sampling of RI 2394’s hull components in 2018, 2019 and 2021 revealed the majority were hewn from white oak (Quercus sp.). This included five floors, one first futtock, two garboard strakes, two ceiling (internal) planks, one stanchion (pillar), and a treenail (a dowel-like wooden hull fastener). The lone exception was the vessel’s keel in the midships area, which was manufactured from elm (Ulmus sp.). Two samples recovered from the forward section of RI 2394’s keel were identified as white oak but are believed to have been fitted as repairs and are discussed below. The exclusive use of white oak and elm in the shipwreck’s construction is not only indicative of a vessel built in a British shipyard, but also exactly matches information about Endeavour’s construction in the 1768 Admiralty survey, which specifies its framing and keel were hewn from English oak and elm, respectively.

All ships leaked, and consequently were fitted with bilge pumps to ensure seawater that entered the hull could be removed so the vessel remained afloat. Eighteenth-century bilge pumps fitted to English ships typically comprised a minimum of two hollow wooden shafts that extended from the lowest part of the hull (bilge) to the upper (weather) deck. The bottom of the shaft was fitted with a lead or copper sieve to prevent debris from clogging it and the pump assembly, which was fitted to the top of the shaft. Most bilge pumps of the period were manually operated ‘common’ or ‘force’ pumps comprising the shaft, as well as a handle, piston, plunger, and valves that allowed water to be discharged while ensuring it didn’t flow back down the shaft into the bilge. An upstroke on the handle allowed the piston to draw water through an inlet valve into the pump shaft. On the downstroke, the water was discharged through an outlet valve into one or more outlet pipes. As originally built, Endeavour (then known as Earl of Pembroke) was outfitted with two force pumps immediately abaft (behind) the mainmast. These were complemented by two additional force pumps—located just forward of the mainmast - when the vessel was acquired by the Royal Navy, renamed HMB Endeavour, and refitted for Cook’s voyage. The two forward pumps were later removed when the vessel reverted to civilian ownership.

In addition to the sieves fitted to the base of the pump shafts, most vessels featured a pump well that housed the lower sections of the bilge pumps. The well was a square or rectangular partition - normally constructed around the base of the mainmast in the lowest part of the hull - that created a ‘sink’ where bilge water could collect. The well’s walls also prevented ballast stones, cargo, and other items from collecting around or underneath the pump shafts and potentially obstructing them.

 

Remnants of the bilge pump assembly were identified on RI 2394 in 2019. These included the stump of the starboard pump shaft, as well as the bases of two partition walls that formed the aft starboard corner of the pump well. Discovery of the pump well was a significant turning point in the identification of the site: It was a recognisable structural feature that allowed the museum’s maritime archaeologists, Kieran Hosty and Dr James Hunter, to positively identify the wreck’s midships section. It could also be compared with the pump well included on archival draughts (plans) of Endeavour generated during the Admiralty’s survey of the vessel in 1768. When RI 2394’s site plan was superimposed over Endeavour’s 1768 lower hold plan and scaled to the same size, the positions of the surviving pump shaft stump and pump well partitions aligned perfectly with their counterparts on the archival document. Superimposition of the site plan and 1768 draughts also allowed Hosty and Hunter to predict the location of the bow end of the shipwreck’s keel, which was confirmed during subsequent investigations of the site by Dr John Broadwater and Joshua Daniel in 2021. Broadwater and Daniel are two American maritime archaeologists contracted by the museum to assist RIMAP during the 2021 excavation season. They filled in for Hosty and Hunter, who were unable to travel to Rhode Island due to restrictions associated with the Covid-19 pandemic.

During investigation of RI 2394 in 2018, archaeologists noted a small number of floor timbers in the forward part of the hull that exhibited larger-than-average scantlings than floors observed elsewhere throughout the site. The sided (width) dimensions for these timbers averaged 16 inches (40.6 centimetres), rather than the 12-to-14 inches (30.5-to-35.6 centimetres) noted for other documented floors. The larger floors were also unfinished on their upper surfaces and retained the curve of the tree limbs from which they were hewn, rather than flat, squared-off surfaces typical of finished ship timbers. However, the upper sided face of at least one of these floors transitioned to a finished surface as it crossed the vessel’s keel, and the moulded dimension (height) of all unfinished floors diminished from 17 inches (43.2 centimetres) to 13.5 (34.3 centimetres) inches at the centreline. Based on their size and appearance, the unfinished floors appear to be the result of quick and inexpensive repairs to the vessel’s forward hull.

Archival records reveal Endeavour suffered extensive damage to its forward section when it struck an unmarked shoal (now known as Endeavour Reef) on the Great Barrier Reef in June 1770. In an entry recorded in his journal on 22 June while Endeavour was being repaired at what is now known as Endeavour River, Cook observed the biggest ‘Leak was found to be at [the] Floor Heads – a little before the Starboard fore chains – here the rocks had made their way through four planks and even into the timbers [frames]’. The hull was also damaged along most of its length on the larboard (port) side, from the bow end of the keel to just aft of the mainmast in the vessel’s midships section. Some floors in Endeavour’s forward section may have been repaired or replaced at this time, or perhaps later when the vessel arrived at Batavia (modern-day Jakarta, Indonesia) and underwent a more comprehensive refit. Other archival sources note the poor overall condition of Endeavour’s hull at the time it was sold out of naval service in 1775, including that several elements of framing (including floor timbers) were ‘bone rotten’. Under civilian ownership (and renamed Lord Sandwich), additional repairs were made to the hull so that it would be accepted by the Board of Transport for use in the American War of Independence. Given the latter repairs were made hastily as a matter of wartime expediency, this could account for the larger size and unfinished condition of some of the floor timbers.

Discovery of RI 2394’s bow section in 2021 raised the possibility that timber sampling and analysis could reveal evidence that correlated with historical descriptions of the extensive repairs made to Endeavour’s bow section in Batavia following its grounding on Endeavour Reef in 1770. The presence of Australian and/or Indonesian hardwoods among the wreck site’s bow timbers would provide compelling evidence for its identification as Endeavour. An unusual scarph (join) was noted in the forward end of the keel in September 2021 and, intrigued by the possibility it could be a repair, the team collected samples from the two timbers that formed the scarph.

While analysis of the samples did not reveal evidence of non-European (e.g., Australian and/or Southeast Asian) timbers, it did confirm white oak (Quercus sp.) was used in the manufacture of the two keel sections that formed the scarph. This in turn raised interesting questions. Given samples recovered from the keel in the wreck site’s midships area were identified as European elm, the presence of white oak keel sections on either side of a scarph in the extreme forward end of the vessel is strongly suggestive of repair to the hull. Further, eighteenth-century British shipwrights typically preferred elm over oak for keel timber, so the presence of oak in the forwardmost section of RI 2394’s keel hinted its use—like the unfinished condition of some of the floor timbers—may have been influenced by haste and/or cost-cutting measures.

Archival sources note that one or more sections of keel within Endeavour’s bow were replaced over the course of the vessel’s career. Endeavour’s bow section and the lower hull in the vicinity of the starboard forechains (approximately eight feet, or 2.4 metres, aft of the stem) were severely affected when it grounded on the Great Barrier Reef in 1770. According to an entry in Cook’s journal on 22 June, and again on 11 October, the ‘Fore foot and main keel [were] damaged’ and the ‘ship [was] very leaky…Occasioned by her main keel being wounded in many places and the scarph of her stem being very open’. These sections of the hull were repaired in Batavia in 1770 and again in 1775 when Endeavour was surveyed prior to being sold out of service. They were also included in repairs to the vessel - now renamed Lord Sandwich - noted in February 1776 when it was surveyed prior to being accepted by the Transport Service.

Lord Sandwich was one of five British transports intentionally sunk across the northern entrance to Newport’s inner harbor to prevent French warships from accessing the port, where they could more effectively bombard British gun emplacements and land troops. In the late-eighteenth century, the act of scuttling would have involved the creation of multiple openings in a vessel’s hull beneath the waterline to allow seawater to flood in and ultimately sink it. As the possibility existed that a scuttled vessel could later be re-floated and reused, holes used to sink it were relatively small and created with a variety of hand tools, ranging from augers to axes.

 

While excavating around RI 2394’s keel in 2018, Kieran Hosty uncovered a large oval-shaped, jagged edged hole in the port garboard strake. The team immediately suspected it to be a scuttling hole due to its location in the lowest part of the hull, and because it bore the hallmarks of having been executed in haste with a heavy striking or cutting implement, such as a crowbar, axe, or adze. A second scuttling hole was observed by John Broadwater during the 2021 investigations in one of the garboard planks on the starboard side, closer to the wreck site’s stern. This hole had cut sides, indicating an edged tool was used to create it.

 

The presence of one regularly formed and one ragged scuttling hole matches the pattern seen on the wreck sites of other vessels deliberately sunk in North American waters by British forces during the American War of Independence. These include Betsy, which was scuttled in 1781 at the Siege of Yorktown, and RI 2125, a transport scuttled in Newport Harbor in 1778 and investigated by the project team in 2002. In Betsy’s case, a ‘neat, rectangular hole [was] chiselled through the inner planking’ just below the lower deck, followed by a ‘second, irregular hole … cut through the outer planking’. RI 2125 featured a ‘square [scuttling] hole … cut or punched through the outer hull planking’ between two of the vessel’s floors.

 

In September 2021, Broadwater and Daniel located RI 2394’s bow at the wreck site’s southern end. Discovery of the bow was based on the predictive model developed by Hosty and Hunter that utilised superimposition of RI 2394’s 2019-20 archaeological site plan over Endeavour’s 1768 Admiralty plans. Discovery of the bow also revealed a distinctive scarph in the surviving keel timber that attached it to the vessel’s stempost (which is no longer present). The survival of the keel-stem scarph - a highly diagnostic feature - was critical to the identification of the wreck site as Endeavour for two reasons. First, it permitted the project team to obtain a measurement from the stem (bow) end of the keel to the projected location of the mainmast, which almost exactly matched the same distance shown on archival plans of Endeavour. The slight variation noted between the two measurements (approximately 8 inches, or 20.3 centimetres) is likely the result of damage to the end of the keel caused by marine borers and other natural degradation.

Second, documentation of the scarph provided critical details about its design and construction. RI 2394’s example is a rare form of stem attachment known as a ‘half-lap’ scarph joint. It allowed the stem to have a near-vertical rake, an absolute necessity for a vessel requiring the broad, bluff bow typical of a Whitby collier. Discovery of RI 2394’s keel-stem scarph revealed it was significantly different from the ‘table’ and ‘box’ scarphs typically used in mid-to-late eighteenth century British shipbuilding. When compared with the keel-stem scarph shown on Endeavour’s 1768 Admiralty plan, it was an exact match in terms of form and size.

Furthermore, a survey of extant eighteenth-century ship plans held in the collections of the National Maritime Museum in Greenwich, England revealed draughts for 40 individual vessels, ranging from Albion (built 1763; NMM J2579) to Chichester (1785; NMM J5188). Only one of these sets of plans displayed a keel-stem scarph like that observed on RI 2394. That vessel, Marquis of Rockingham (built 1770), was another Whitby collier built by Thomas Fishburn (the owner and master shipwright of the shipyard where Endeavour was constructed), and later commissioned by the Royal Navy and renamed HMS Raleigh. Raleigh is perhaps best known as one of James Cook’s other vessels of exploration, HMS Adventure.

A literature review of comparable historic shipwrecks has also revealed that only one other eighteenth-century site with a keel-stem scarph similar to that of RI 2394 has been archaeologically investigated. That site, known as the Chub Heads Cut shipwreck, is located in Bermuda and has tentatively been identified as the remains of a late-eighteenth century British-built collier.

When the museum’s maritime archaeologists drafted the site plan of RI 2394’s surviving hull in the vicinity of the pump well, they noted a curious arrangement in the pattern of the floor timbers. Most floors documented on the wreck site were spaced at intervals; however, an arrangement of three consecutive floors with no spaces between them were noted around the pump well. When the team examined the superimposed archival and site plans, they noted the arrangement of the ‘tripled’ floor timbers on the shipwreck aligned with the location of Endeavour’s mainmast on the 1768 draught.

While examining forward hull structure following discovery of the bow end of the keel in 2021, Broadwater and Daniel noted two consecutive floor timbers with an arrangement similar to that observed with the ‘tripled’ floors in the midships section. These new timbers were incorporated into the site plan and an interesting pattern emerged: As with the tripled floor timbers, the ‘doubled’ floors in the bow correlated to the location of a mast - in this case, Endeavour’s foremast. Endeavour is known to have been outfitted with a ‘rider’ or ‘deadwood’ keelson atop its regular keelson, a longitudinal centreline timber that sat atop the floors, locked them against the keel, and reinforced the lower hull structure. The rider keelson was an extra baulk of timber incorporated by Thomas Fishburn to further reinforce the hulls of his colliers and allow them to ‘take the ground’ or sit high and dry on the seabed at low tide. Based on archaeological investigation of another Fishburn-built collier, the General Carleton, the bases of the fore- and mainmast were inserted directly into mortises cut into the upper surface of the rider keelson. The findings from General Carleton, as well as the 1768 plans of Endeavour, reveal Fishburn-built vessels do not appear to have been fitted with complex robust structures called mast steps that accommodated and took the weight and strain of their fore- and mainmasts. In Endeavour’s case, it appears consecutive floor timbers were installed beneath the fore- and mainmasts to compensate for the lack of mast steps.

Unique diagnostic artefacts — such as a ship’s bell, name board, or an artefact bearing the name of a crewman, passenger or prisoner associated with Lord Sandwich or Endeavour — have not been encountered on RI 2394. However, given Lord Sandwich was used as a prison hulk and later intentionally scuttled, it would have been regularly cleaned and ultimately stripped of anything of value before ending up on the bottom of Newport Harbor.

Consequently, its wreck site is very unlikely to retain diagnostic artefacts, and this is reflected in the relative dearth of small finds encountered generally on RI 2394 so far. Therefore, identification of the site has hinged on the surviving hull and the evidence it contains. Enough correlations have been drawn between the archaeological and historical records to identify RI 2394 as James Cook’s Endeavour and there is now an urgent need to secure the highest possible level of legislative and physical protection for the site, given its historical and cultural significance to Australia, New Zealand, the United Kingdom, and the United States of America.