These dimensions of the _Savannah_ required the researchers to investigate the method of taking register dimensions in 1818. It was found that the customhouse rule then in effect measured length between perpendiculars above the upper deck, from "foreside of the main stem" to the "after side of the sternpost." The beam was measured outside of plank at the widest point in the hull, above the main wales. If a vessel were single-decked, the depth was measured alongside the keelson at main hatch from ceiling to underside of deck plank; if double-decked, one-half the measured beam was the register depth.[5] However, inspection of the register of a number of ships of 1815-1840 showed that, in practice, double-decked ships commonly were measured as single-decked ships; this obviously was the case in the _Savannah_. Also, due to the lack of precise measuring devices, the register dimensions were not always accurate, particularly those of the length, which often were in error as much as one foot in a hundred, as was found by investigation of various classes of vessels. Because of inherent difficulties in measuring to the required points, this condition lasted even after steel tapes were introduced late in the 19th century. The Museum's researchers next turned their attention to examination of the Marestier work, a French report on early American steam vessels that had become known to some American marine historians in the 1920's. The author was a French naval constructor who, on orders from his government, had spent two years in the United States between 1819 and 1822 studying American steam vessels, schooners, and naval vessels. The published report contained only material on steam vessels and schooners. The portion dealing with naval vessels was not published, and the manuscript has not been found to the present time (1960). The publication, a rare book, was available in only a few collectors' libraries or public institutions in the United States. In 1930 the writer translated the chapter on schooners,[6] and in 1957 Sidney Withington translated most of the remainder.[7] As a result of these publications and earlier published references, the Marestier material became widely known to persons interested in ships. [Illustration: Figure 3.--Marestier's sketch of the _Savannah_ (from plate 8 in Withington's translation of the Marestier report). Heights of lower masts are excessive by all known American masting rules; and, according to Marestier's drawing of the engine (see figure 4), the deckhouse is too short.]