Piecing together a puzzle

In September 2018, maritime archaeologists from the museum and its research partner and major sponsor the Silentworld Foundation (SWF) worked with the Rhode Island Marine Archaeology Project (RIMAP) to investigate an 18th-century shipwreck in Newport Harbor, Rhode Island. The wreck site is a strong contender for Lord Sandwich, a merchant vessel contracted by the British government to transport its troops and Hessians (German mercenaries) to North America during the American War for Independence (1775–1783).

Lord Sandwich was one of 13 vessels intentionally scuttled in Newport Harbor ahead of a combined American and French land and naval assault on Rhode Island in August 1778, but its significance to Australia relates to its prior identity as HMB Endeavour – Lieutenant James Cook’s ship during his first voyage of exploration to the South Pacific between 1768 and 1771.

Lord Sandwich and the Siege of Newport

After Cook’s first voyage ended, Endeavour was refitted as a naval store-ship and used to transport soldiers and supplies to Great Britain’s far-flung military outpost at Port Egmont in the Falkland Islands. It made three return voyages to the Falklands, the last of which evacuated the British garrison and most of its arms and equipment in April 1774. Endeavour was paid off five months later, then sold to civilian interests in March 1775 and sent on at least one commercial voyage to the Imperial Russian port of Archangel.

With the outbreak of the American War for Independence in April 1775, the British government began contracting civilian vessels to transport troops and military materiel to its rebellious North American colonies. Endeavour was tendered for consideration, but the Admiralty rejected it due to its poor condition. Following repairs, the vessel – now named Lord Sandwich – was finally accepted for service in February 1776 and assigned as a troop transport. The new name honoured John Montagu, the 4th Earl of Sandwich, who served as First Lord of the Admiralty during the latter half of the 18th century and was an avid supporter of Cook’s voyages. 

Three months later, Lord Sandwich took on a contingent of more than 200 Hessians, joined a fleet of 100 ships (nearly 70 of them transports) and departed Portsmouth for New York. After a particularly rough crossing that scattered most of the convoy, the ship arrived at Halifax, Nova Scotia. The fleet reassembled there and departed for Sandy Hook, New Jersey, arriving on 15 August 1776. Their convoy was joined by several more transports and store-ships. The combined fleet arrived at Staten Island shortly thereafter and supported the British assault on New York.

Following New York’s capture from the Americans, British military leaders turned their attention to Newport, which was held by American forces and represented a threat to British control of New York and its surrounds. In November 1776, Lord Sandwich collected another contingent of Hessians and joined a convoy bound for Rhode Island. British troops and the German mercenaries in their employ quickly seized control of Newport, but were unable to completely subdue the Americans, who controlled the shores surrounding Narragansett Bay and began preparing for a counter-offensive. After the British surrender at Saratoga in October 1777, France entered the war on the side of the Americans, and plans to retake Newport commenced in earnest. 

By the summer of 1778, the Americans and their new allies agreed upon a combined assault that would involve Continental Army and French forces approaching Newport from the north in conjunction with French naval bombardment from the harbour. The overwhelming size of the French squadron (18 vessels, including 11 ships-of-the-line) prompted the British at Newport to intentionally burn all Royal Navy warships present in Narragansett Bay to prevent them from falling into enemy hands. In addition, 13 transports were scuttled in Newport’s Outer Harbor to block access to the Inner Harbor and provide a partially submerged barrier between the city’s land-based artillery batteries and the attacking French warships. Lord Sandwich was one of these vessels, and was sunk in a line with four other ships across the channel between Goat Island and North Battery (a gun emplacement situated at the north end of the city) in early August 1778. The French fleet began its attack on Newport on 8 August, but withdrew the following morning to engage a newly arrived British fleet under the command of Admiral Lord Richard Howe. Ultimately, the British retained control of Newport and a number of the scuttled transports in the Inner Harbor were later re-floated. Lord Sandwich was not among them.

2018 investigations of RI 2394

Two hundred and forty years later, the RIMAP-led team continued archaeological investigations of RI 2394, the official designation of a wooden-hulled historic shipwreck located within Newport Harbor, where Lord Sandwich and four other transports were reportedly scuttled. Better known by its nickname the ‘Kerry Site’ (named after Dr Kerry Lynch, an American archaeologist and the RIMAP Field Supervisor), the wreck is largely buried beneath the seabed, but its visible features include stone ballast, four small 18th-century cannons, a lead scupper and a variety of partially exposed wooden hull components. Among the latter are a line of frames (the ‘ribs’ that formed the skeleton of the ship), as well as the stump of a stanchion (vertical post) and sections of hull (external) and ceiling (internal) planking. 

The scantlings (dimensions) of the visible hull timbers of RI 2394 are substantial, and correlate well to those of Lord Sandwich (Endeavour). However, some discrepancy exists in the archival record regarding the tonnages of the four other ships that were scuttled between Goat Island and the North Battery. Lord Sandwich is believed to be the largest vessel among the five that were sunk, and RI 2394 appears to be the largest of the five shipwreck sites found in this part of the harbour. 

Photogrammetric survey and 3D reconstruction

To further narrow the field, the team tried to collect as much data from RI 2394 as possible – including imagery of the wreck’s visible remnants, which could provide essential clues about its construction. Archival sources – including ship’s draughts – exist that provide great detail about how Endeavour was built and fitted out. These range from individual timber scantlings to the arrangement and spacing of frames and the vessel’s mast steps (timber assemblies into which the ship’s masts were inserted, or ‘stepped’). A composite image of the wreck site could therefore allow its architectural remnants to be correlated with these historical records. When environmental conditions such as water clarity are ideal, creating a photograph-based mosaic is straightforward; however, the waters of Newport Harbor are notoriously turbid in summer, and visibility often does not exceed one metre. This created a unique set of challenges that the team had to overcome to adequately document the site in its entirety.

Thankfully, a relatively new technique available to maritime archaeologists is Photogrammetric 3D Reconstruction (P3DR). P3DR is a cutting-edge algorithmic process in which highly detailed and visually accurate digital 3D models or digital reproductions of real-world objects can be generated from multiple digital still images. The technique is also known as ‘structure from motion’, ‘photogrammetry’ and ‘3D reconstruction’. The term ‘photogrammetry’ is widely used within maritime archaeology to refer to P3DR; however, photogrammetry traditionally refers to the science of obtaining measurements from photographs, and although this occurs at very high density in P3DR, the later stages of digital 3D model development are beyond the scope of traditional photogrammetry.

Taking cues from the photogrammetric survey of submarine AE1 in April 2018 (see Signals 124), the team used underwater camera arrays that included powerful lights to cut through the gloom of Newport Harbor. The cameras themselves were pre-programmed to capture one 12-megapixel image every two seconds. Visible elements of the wreck site were systematically photographed from multiple perspectives, and care was taken to ensure necessary overlap (no less than 60 per cent) among captured images. Because water clarity was generally poor, only 50 square centimetres (or less) could typically be captured within a single photograph at a time; consequently, a single one-hour dive could generate as many as 500 images, but only document a relatively small portion of the site. 

While this technique worked well for hull remains and other site components with unique visual attributes, it proved insufficient for portions of the wreck that were buried beneath sediment or were relatively featureless. To combat this problem, the team placed photogrammetric ‘targets’ throughout areas of sterile seabed. Each target comprised a sheet of white Mylar about 10 centimetres square, printed with a unique geometric pattern. While photographically surveying buried parts of the site, team members swam overlapping transects along the site’s length, made sure to capture no less than two targets in each image, and ensured that at least one target overlapped between successive images. Taken together, the unique pattern on each target provided the photogrammetric processing software with a means of visual recognition that enabled it to combine multiple images into a single digital model. 

More than 10,000 photographs of RI 2394 were collected over the course of the 2018 field season, and the sheer volume of images has meant that generating a composite 3D model of the entire shipwreck has been painstakingly slow. To help combat this, and to test whether the survey was capturing useable imagery, the team created medium-resolution models of specific site features – such as cannons – while still in the field. The test models confirmed the efficacy of P3DR in the documentation of RI 2394 and other historic shipwrecks in Newport Harbor, and have formed the basis of a much higher-resolution model of the entire Kerry Site that is currently in development. 

Dr James Hunter and Kieran Hosty, with Irini Malliaros (Silentworld Foundation).

Further reading

  • Abbass, D K, 2001, ‘Newport and Captain Cook’s ships’. The Great Circle, Vol 23.1: 3–20.
  • Connell, M and Liddy, D, 1997, ‘Cook’s Endeavour bark: Did this vessel end its days in Newport, Rhode Island?’. The Great Circle, Vol 19.1: 40–49.
  • Erskine, N, 2017, ‘The Endeavour after James Cook: The forgotten years, 1771–1778’. The Great Circle, Vol 39.1: 55–88.
  • Knight, C, 1933, ‘HM Bark Endeavour’. The Mariner’s Mirror, Vol 19.3: 292–302.
  • Marquardt, K H, 1995, Captain Cook’s Endeavour: Anatomy of the Ship. Conway Maritime Press, London.
  • McGowan, A P, 1979, ‘Captain Cook’s ships’. The Mariner’s Mirror, Vol 65.2: 109–118.

James Hunter

Dr James Hunter is the inaugural Curator of RAN Maritime Archaeology at the Australian National Maritime Museum. He received his MA in historical archaeology from the University of West Florida, and holds a PhD in maritime archaeology from Flinders University, where he is an Associate Lecturer in the Department of Archaeology. James has worked in the field of maritime archaeology for nearly two decades, and during that time has participated in the investigation of shipwrecks and other archaeological sites ranging from prehistory to the modern era. He was a member of the archaeological team that investigated the American Civil War submarine H.L. Hunley, and a staff archaeologist with the U.S. Naval History and Heritage Command’s Underwater Archaeology Branch. He has been published widely and is also an accomplished archaeological illustrator whose work has been featured in a number of scholarly books and articles.