Shelburne Bay, Vermont, USA

Mission director: K. Crisman & C. Kennedy Texas A&M University

Shelburne Shipyard Steamboat Graveyard Project

by Carolyn Kennedy

The Shelburne Shipyard Steamboat Graveyard Project began in June 2014, under the direction of Dr. Kevin Crisman and Carolyn Kennedy. The project entails the study of four steamboat hulls that currently lie in Shelburne Shipyard, Lake Champlain, Vermont. These wrecks are the remains of passenger steamboats once owned by the Champlain Transportation Company (1826) and were sunk intentionally in Shelburne Shipyard after they were deemed unsafe for passenger transport. The four wrecks, Wrecks 1, 2, 3 and 4, are believed to be the steamers A. Williams (1870), Phoenix II (1820), Burlington (1837), and Whitehall (1838). The latter three steamboats were built in the period just post-Fulton, when commercial passenger steamboats were a brand new phenomenon. For this reason, these wrecks are particularly of interest to nautical archaeologists examining the dynamic period of steamboat construction that followed the first steamboats. The shipwrights that built these early steamers were figuring out the best construction techniques to create larger, faster boats, but still sturdy enough to travel on Lake Champlain.

Shelburne NPR with potree
click to open the 3D model with potree viewer

We can see these changing designs in the three early Shelburne Shipyard hulls reflected in the narrowing sided dimension and increased spacing of the framing timbers, and the progression to additional longitudinal support timbers. The earliest Wreck 2 was built with heavy, square-cross-sectioned framing timbers spaced fairly close together. This wreck also contains only two pairs of longitudinal engine support timbers. Wreck 3’s frames were built narrower, but deeper, and spaced further apart in an effort to relieve the overall weight of the vessel. The longitudinal support timbers on Wreck 3 increased to three per side, and were longer overall. Wreck 4, built only a year later than Wreck 3, again shows these features becoming more pronounced, with even wider spacing between frames strategically at the ends of the hull, and longitudinal support timbers spanning the length of the boat. These changes were indicative of the experimental nature of steamboat construction in these first few decades of commercial passenger steamboats. They demonstrate the shipwrights’ desire to increase the speed of the boats by decreasing the weight of the framing timbers, and support the longer vessels with longer and more numerous longitudinal support timbers.

Over the course of two seasons, June 2014 and June 2015, these four wrecks were studied by Texas A&M University graduate and undergraduate students of the Nautical Archaeology Program. In the first season, the focus of the work was to identify the wrecks and produce preliminary site plans of all four wrecks. With additional funding received from the National Park Service, we were able to expand the scope of work in the second season by including photogrammetric recording of the four wrecks, under the leadership of Kotaro Yamafune. Yamafune, paired with prospective TAMU Ph.D. student, Dan Bishop, were able to fully record over 700 ft. of shipwreck in less than three weeks. Since the second season focused specifically on Wreck 2 (Phoenix II), this was where the majority of the photogrammetric work took place as well. Thanks to Yamafune, Wreck 2 was recorded pre-disturbance, as well as throughout our archaeological work on the site.

Shelburne 3D restitution
3D rendering by Kotaro Yamafune, Texas A&M University.

In 2015 the TAMU team returned to Shelburne Bay and recorded Wreck No. 2, previously believed to be the steamer Winooski, but now believed to be Phoenix II (1820), using photogrammetry. The recording process, carried out under the guidance of Kotaro Yamafune, entailed the acquisition of thousands of images, which were used to produce a tridimensional mesh using off-the-shelf software Agisoft PhotoScan. An animation was produced to illustrate the work. These data will be used by the GROPLAN team to develop a set of informatics tools to extract and treat tridimensional artifacts and work toward their automatic identification.