Reconstructing the Plans
The first step in the reconstruction of the Savannah's plans was to block out the register dimensions on a scale of one-quarter inch to the foot in a drawing and then to work out the profile, using the Ohio plan as a general guide. This produced a hull about 100 feet 9 inches in length at main rail to inside of plank, or "moulded"; 25 feet 6 inches moulded beam, allowing 3 inches for plank (as usual in a ship of this size and date); and about 15 feet 4 inches moulded depth at side, keel rabbet to underside of upper deck. The bulwarks were drawn at 28 inches height. Next, the mast positions were decided by prorating from the plan of the Ohio the position of each mast from the fore perpendicular and then modifying these positions slightly by use of masting rules contained in M'Kay's book[20] of 1839.
Since it appears that the Savannah may not have been purchased for conversion to a steamer until near the date of her launch and because of the lack of identification of the lithograph referred to by Collins, the statement that the mainmast was placed farther aft than normal was rejected. At launch her mast partners would have been in place and the deck laid. Any alterations in the position of the mainmast then would have made it impractical for the owners to demand them of the builders without heavy additional expense. In addition, the plan, as it was developed, indicated no need for such alteration.
The plan of the engine, drawn to the same scale as the profile plan, was shifted about on the lower deck in the hull profile to determine where the engine and side paddle wheel shaft might be located. A little experimentation and study made it certain that the proper location could be estimated within a foot or so, to scale, as to fore and aft positions. The after end of the cylinder, and its piping, had to clear the mainmast by at least 9 to 10 inches to allow removal of the cylinder head for inspection and repair. The position of the wheels, stack, and masts in Marestier's sketch of the ship make it certain that the engine was on the lower deck, abaft the paddle wheel shaft. Due to differences between the dimensions stated by Marestier and in the Vail account books and what the graphic scale in Marestier's engine drawings produce, the exact dimensions of the engine are uncertain. Nevertheless, they can be approximated with enough accuracy for our purpose. As a result of this treatment, it seems fully apparent that the engine was abaft the paddle wheel shaft, with frame extending abaft the mainmast on the lower deck; there does not appear to be a practical alternative in the light of the available evidence. This matter will be referred to again.
The size of the cylinder and its valve chest and the inclined position of the cylinder indicate conclusively that the valve chest was in the mainhatch, which would normally be just forward of the mainmast. Even then, the after flange of the cylinder would just clear the lower deck, allowing 6 feet between decks, as in the Ohio. The cylinder would have been about 6 feet long; the graphic scale indicated 6 feet 3 inches. The diameter of the cylinder plus height of valve chest seems to have been 5 feet 9 inches to 6 feet. Because of the use of the crosshead and a connecting rod, pivoted at crosshead, the oscillating rod (or pitman) and piston together equalled twice the stroke plus allowance for stuffing box, crosshead, and pitman bearings. Therefore, the engine's over-all length, from head of cylinder to the centerline of the side paddle wheel shaft, could not have been much less than 15 feet 9 inches, and probably as much as 16 feet 2 inches, thus making the length at extreme clearance of crank throw as much as 19 feet. These dimensions indicate that the centerline of the side paddle wheel shaft must have been from 38 to 39 feet from the forward perpendicular. It is not clear how the wheel shaft was mounted in the vessel. Taking into consideration her depth and her reported draught, light and loaded, the Marestier sketch, and the hull structure then used, it seems reasonable to place the centerline of the shaft (which seems to have been about 7 to 8 inches square) about 12 inches above the upper (or spar) deck to allow proper dip of the blades. This position would have given proper blade immersion at the mean draught of 13 feet.
In order to get the engine below deck, and to get the boiler or boilers placed, it was necessary to cut a large opening in the two decks. It may be assumed that this opening was big enough to take the cylinder, without valve chest, and also the boilers, which went into the hold. Taking the proportions of other boilers as shown by Marestier, it has been estimated that the Savannah might have had a boiler about 18 to 20 feet in length, 7 to 8 feet wide, and 6 to 61/2 feet high at firebox. The form might be the same as that of Fulton the First, illustrated in the translation of Marestier's report.[21] However, since the Russian descriptions[22] indicate there were two boilers, each measuring 6 feet in diameter and 27 feet in length, the two boilers would have reached past the mainmast if they were located in the same manner and in the same place as the boilers shown in the illustration of Fulton the First. Consequently, if the Russian description is accepted, there would have been a need for longer fuel (coal) spaces in the wings.
The boilers, then, were the largest piece of equipment to be passed through the decks; for this an opening (estimated to have been about 101/2 feet wide and 81/2 feet long) probably was cut through both decks about 3 feet forward of the main hatch, which was commonly a little forward of the mainmast. The boilers could then have been lowered, after end first, into the hold. The opening in the lower deck could then have been closed, except for a small hatchway perhaps, and the steam cylinder let down to the lower deck and moved aft into position. To allow the crosshead to reach its maximum travel, the opening in the upper deck would have been about 101/2 feet wide—the over-all width of the engine frame—and would have been left open, inside the deckhouse.
The width of the boilers might be particularly important because it would determine the deadrise at floor in the hull. The apparently precise dimensions of the boilers given in the Russian description were utilized to arrive at a suitable hull form. Both a single boiler and a double boiler (as described in the Russian accounts) were placed in the hull to assure the correct space estimates.
Since the engine, as shown by Marestier, had an air-pump cylinder alongside the steam cylinder (with the pistons of both attached to the crosshead), it is evident that a condenser was employed. This condenser would not have been much larger than the air-pump cylinder. It may have been placed under the side paddle wheel axle on the lower deck, but its mode of operation is unknown. Possibly it was of the jet type, with pumps operating off the paddle wheel axle and with a return of condensate from a hot well into the feed water line. A number of possibilities could be mentioned, all speculative. However, there was no doubt that this equipment could be properly installed in the reconstructed hull, either on the lower deck or in the hold.
Two questions have been raised as to machinery arrangement—whether the engine, and boilers also, might have been forward of the wheel shaft, and whether the wheel shaft was above or below deck. If the engine were placed forward of the wheel shaft, the wheels might be farther aft than is proposed in the reconstruction. However, the smokestack could not then be forward of the wheel shaft as shown by Marestier because it would have had to pass through the engine frame, thus interfering with the movement of the large crosshead. If the engine were abaft the wheel shaft, the stack could have been only as shown by Marestier. The boilers might then have been forward of the wheel shaft only if the stack were at the end away from the firebox. However, the length of the boilers as indicated by the Russian description would then have required them to pass through the bows!
Models have been built of the Savannah in which the engine and boilers are forward of the paddle wheel shaft, and the shaft below the main deck. This was accomplished by placing the engine off center so that the stack came through the decks alongside it. This is an impractical arrangement because it would have created an impossible ballasting problem. The weight of the engine, to port in the models, would have to have been counteracted by ballast to starboard. Due to the coal bunkers, and the possibility of two boilers below the engine in the hold, there would not have been room for sufficient ballast. In addition, were such ballasting possible, the combined weights were too far forward to give proper trim, and a great deal more ballast would have been required far aft, a most impractical proceeding.
The position of the wheel shaft was determined as described earlier. The ship was apparently well-advanced in construction at the time of purchase. Her clamps and shelves supporting her upper deck beams, which then would have been in place, were important strength members. In reconstructing, to place the wheel shaft below these members would not only bring the engine nearly level—it is described and shown inclined by Marestier—but also would immerse the paddle blades too deeply for the draft and depth of the hull. To place the shaft below or through the lowest clamp member would require the shaft centerline to be at least 3 feet below the upper deck, and this would contradict Marestier. These questions indicate the importance of a scaled drawing when deciding arrangement in the reconstruction of a ship under the circumstances existing in the Savannah. Some models have been built with the shaft below deck by disregarding the structural and dimensional objections just outlined.
The question of the number of boilers originally was raised by Braynard. A single boiler with double flues was a common boiler design in American steamboats of 1818–1828, and this form of boiler is shown in a number of Marestier's drawings. In general descriptions, "boiler" and "boilers" are often used interchangeably, and this probably came about through confusion over the number of flues. A "single boiler, double flues," would thus become "boilers," apparently. The Russian description specifically states there were two boilers, and gives specific dimensions; though these probably are not exact. Either a single boiler with double flues, or double boilers, each with a single flue, could have been fitted in the reconstruction. However, fuel space is affected and, with double boilers, the cross-sections of the bunkers are reduced to about 20 square feet each; therefore, the bunkers would have to become much longer. It may be said that the boiler capacities in relation to dimensions of the steam cylinder as indicated in the Russian description far exceed those given by Marestier. As a practical matter of ship design, it seems that the single boiler would have been a more logical fitting than double boilers. The boilers were apparently of copper, and expensive. However, this matter does not affect the hull-form and dimensions established for the reconstruction, as the drawings proved. The Russian description does show that the cargo space was extremely small and practically nonexistent, indicating the effect of the large boiler capacity.
All requirements that have been given can be approximated for space necessary in the hull. It is established that the ship carried about 75 tons of coal and 25 cords of wood. The coal would take up from about 1,700 to 1,850 cubic feet of space, and because of its weight it would have to be bunkered alongside the boilers in the lower hold, where there would be ample room, in the reconstruction, for two bunkers, each in excess of 30 square feet in cross section and about 28 feet in length for a single boiler; one third more bunker space, in length, would be required for double boilers. Such bunkers would together hold about the required tonnage or cubic footage. The cord wood would have required, say, two bunkers each of about 60 square feet in cross section and 20 to 24 feet in length. Because of the light weight, the cord wood could have been stowed in the wings on the lower deck. There is room for the required stowage on the lower deck in the reconstructed hull, leaving ample passages under either side of the engine frame.
Marestier shows the location of the stack as being abreast the buckets on the forward side of the paddle wheels, and it has been so placed in the reconstruction. The deckhouse shown in Marestier's sketch extends from a little forward of the mainmast to a little forward of the paddle wheel axle. Probably this house actually covered the main hatch and the crank-connecting-rod hatchway; therefore, Marestier shows it too short. In the reconstruction, the deckhouse works out as between 17 and 18 feet long. Its width can only be guessed at, but it probably would have been as wide as the opening cut in the upper deck for machinery—say 11 feet. Perhaps this house contained the engineer's stateroom and that of his assistant, as well as a ladderway to the engine room. Doors on the sides of the house gave access to these spaces and to the inboard shaft bearings. Bunker hatches were probably forward of the house and outboard; these are taken as being about 2 feet 6 inches wide and 3 feet 6 inches long—large enough to allow coal baskets to be lowered through them, as well as to allow cord wood to be passed below.
A fidley hatch, in which the stack passed through the upper deck, would have been a square hatch forward of the deckhouse. This hatch, about 21/2 to 3 feet square, would have been fitted with an iron or iron-bound fidley grating, with solid cover over. The stack could have been swivelled, to bring the elbow to leeward. The upper portion of the stack probably overlapped the lower portion at least 3 to 4 feet above the fidley coaming, and the upper stack rested on a collar bearing at the bottom of the overlap. Perhaps straps were bolted to the side of the upper stack to take heaving bars athwartships, by which two men could rotate the upper stack to turn the elbow to leeward.
The bearings of the paddle wheel axle were perhaps four in number. Two, one either side of the crank, may have been secured to the engine frame just inside the deckhouse walls. Two were certainly outboard, one on each side, fastened to the topsides, as shown in Marestier's sketch of the wheel construction. The axle, probably square in cross section, turned only at the bearings and wrist pin. It may have been cast in two parts, each with a crank arm, and then joined by the wrist pin, after the latter had been turned.
The wheels, shown in much detail in Marestier's sketches of the engine, had flanged hubs to which the pivoted arms or spokes were bolted. The fixed arms were integral parts of the outer hubs. The inner flanges were cast with the hubs. To fold the blades, the fixed arms were brought parallel to the rail, then the chain span between each pair of the pivoted blades on top of the wheel was disconnected and a pair of the blades, each way, were dropped on top of the fixed arms, or blades, and lashed there. The wheel was then given a half-revolution and the process repeated. The wheel could then have been unshipped from the hub by sliding it off the square shaft end after removing, let us suppose, a bolt or pin in the hub. Some writers, like Collins, refer to a "jointed" or "hinged" axle, but Marestier makes no mention of such an arrangement; indeed, his sketch makes a "broken" axle impractical. The wheels could have been removed from the axle and lifted aboard by use of tackles from the main yard ends, or from a fore spencer gaff if it were made long enough. However, as stated in the Russian description, the pivoted blades were removed and stowed aboard, leaving only the two fixed arms in a horizontal position outboard. This is a far more convenient treatment than unshipping the whole wheel, as might be supposed from logbook mention of "shipping" or "unshipping" the wheels.
There remain some other matters to be explored. The ship was fitted with 32 passenger berths in staterooms. The passenger accommodations for first class passengers in the early (1820–1830) packets were aft, on the lower deck. The berths would have been about 6 feet 2 inches long, and 21/2 feet wide. With berths placed athwartships and allowing for cabin bulkheads, there would have remained a space at least 10 to 12 feet wide down the centerline of the ship. This space would have provided space for a mess table and a lounge area. Each stateroom would then have been about 7 feet long fore and aft and could have contained four athwartship berths. The space available abaft the middle of the after cargo hatch would have allowed four staterooms on each side and room at the extreme stern for a small master's cabin, with toilets on each side. The cabin of the mates and stewards, containing two berths each, would then have been about abreast of the fore end of the after cargo hatch.
The galley would have been on the lower deck, just abaft the foremast and forward of the fore cargo hatch. Food would have been carried aft along the lower deck to the cabin, by way of passages on either side of the engine frame. Cabin stores would have been in the hold below the passenger accommodation, and here food, water, and other stores would have been kept. A small cargo space, say of about 1,500 to 2,500 cubic feet, depending on bunkers, would have been possible in the after hold. A fore cargo hold of about 1,000 to 1,500 cubic feet of contents could be expected; forward of this would have been sail locker, spare rigging gear, and a cable tier. On the lower deck, above these spaces, a forecastle might have had berths for 12 to 14 men. The cables and chain would be passed through the forecastle to the cable tier below by chutes leading from cable scuttles in the upper deck abaft the windlass on each side of the centerline of the ship.
The upper deck, abaft the mainmast, was reserved for use of the passengers and officers of a packet. The low, 28-inch bulwarks were insufficient to give proper protection there, so they were increased by employing a 16-inch rail made of a cap supported by iron stanchions above the main rail. This rail was closed in by a tarred netting extending from the main rail upward to the quarter-deck rail cap and running from the mainmast aft to the stern. This is plainly shown in Marestier's sketch of the Savannah as well as in some portraits of early packet ships.
Though the passenger accommodations described were far from palatial by modern standards, they were considered adequate in the 1820's and for almost 15 years afterwards. The staterooms had no individual toilets. Usually there were two small toilets, one on each side of the stern cabin, at the extreme stern on the lower deck, in the quarters. Usually the master's stateroom and toilet were to starboard, with a public space and toilet to port. Sometimes toilets for the crew were placed forward, on either bow abaft the catheads on the upper deck. These were small cabinets accommodating one person each, and with the door closed for privacy there was not room to stand. To enter the user backed in, crouching. Such cabinets are not shown by Marestier, so probably the crew used the headrails, as then was usual in merchant vessels.
The hull-form to be chosen had to enclose all spaces that have been described or listed. Since the Savannah is known to have sailed quite fast for her length, her lines had to equal those of the Ohio; however, her smaller size and other factors indicated a somewhat different hull-form, with harder turn of the bilge and a little less deadrise. Due to the position of the machinery, the effect of its weight and that of the necessary fuel had to be considered. The midsection, or cross section of greatest area, would have to have been only a little abaft the paddle wheel axle to allow proper trim with a minimum of ballast. It was found by this criterion that the midsection of the reconstructed hull was located in proportion to length in a comparable manner to that of the Ohio. The run could have been made about as long and easy, in proportion, as that of the Ohio; likewise, the entrance could have been equally well designed for sailing. Probably a little ballast—stone, gravel, sand or pig iron—was required under the temporary flooring of the cargo holds, most of it abaft the mainmast. Some ballast would normally have been placed under the cabin stores, in the run. The boilers, engine, and fuel weights were relatively important. To trim the ship, with minimum ballast, the location of the machinery weights would have to have been about as shown in the reconstruction drawings. It may be observed that the engine and fuel weights are relatively great for the recorded hull dimensions and resultant displacement limitation, indicating only a small quantity of ballast would have been employed under any circumstance.
Using the Ohio as a guide, the midsection was formed to comply with the dimensions of the boilers and with due regard to the small dimensions of the Savannah. The result was a section having very moderate rise of straight floor, carried farther out in proportion to beam than in the Ohio, but with rather easy turn of the bilge and moderate tumble-home in the upper topsides. This section has a form found in plans of some American freighting ships of 1815–1830, but with slightly slacker bilge.
The stern used in the reconstruction was the "square stern and round tuck" seen in the Ohio and referred to in the Savannah's register. Collins' "round stern," shown in Hudson's drawing, did not come into use in America until about 1824, and then in naval ships only, so far as existing plans of American vessels show.
The reconstructed hull-form ([figure 6]) shows the man's bust figurehead mentioned in the register, and the supporting head and trail mouldings employed in the packets and other American ships of the period. The figurehead may have had some relation to the original or intended name of the ship prior to her purchase for conversion. No detailed description has been found. A ship built to the drawing would at least sail well and would carry her machinery, fuel, etc., as indicated in the descriptions that exist. Whether or not the hull is precisely like that of the original ship can never be determined until the original plan, or model, is found. The proposed deck arrangement is shown in dotted lines, in plan view.
The rig shown in [figure 7] is based upon Marestier's sketch and his incomplete description. Since the ship had long royal poles on her topgallant masts it is highly probable she crossed royal yards, like the later packet ships. The proportions for the length of spars are based upon the masting rules given by M'Kay[23] in 1839. The fore spencer gaff, used as a crane for handling coal and cargo if the fore or main yards were not available, may have been long enough to be used also as a crane to handle the side wheels. The stack and mainstays may have made the fore spencer sail a nuisance, so it may not have been set while the vessel had her engine. In general, aside from the use of the spencers on fore and main, the sail plan shown is of standard proportions and arrangement of 1815–1825. For rigging, Darcy Lever's book[24] was consulted. The drawing of the reconstructed Savannah's sail plan agrees with contemporary sail plans of ships in the author's collection. The log shows she set studding sails and had all the light canvas of a ship of her type.
There remain a number of matters that do not directly concern the reconstruction project but which are of sufficient technical importance to warrant comment. Apparently the engine was mounted on a wooden frame consisting of two large oak timbers on each side, say about 10" × 10", one above the other, that probably supported iron saddles in which the two cylinders rested. Between each pair would have been the iron track, or channel, in which the ends of the crosshead travelled, along the axis of the engine in elevation. These frames measured about 9 feet 2 inches, outside to outside, and reached from the beams of the upper deck on either side of the crank hatchway to abaft the mainmast on the lower deck. It is probable that the fore and after ends of the frame were supported by stanchions stepped on the lower deck at the fore end and in the hold at the after end. The crosshead was of iron and probably had shoes at the ends to work in the tracks or channels in the frame. To help steady the crosshead, these shoes probably were a foot or more long, for the loading of the crosshead is spread out. The pitman to the paddle wheel shaft is to starboard of the centerline of the engine; the steam cylinder piston is slightly off center of the frame and crosshead; and the piston of the air cylinder is close to the port engine frame. The steam lines to the valves of the steam cylinder come in horizontally over the frames. As has been mentioned, the frame may also have supported the paddle wheel axle bearings at the crank.
This engine has been criticized by some writers (see Tyler's[25] résumé of Gilfillan's[26] comments), but the Savannah logbook shows it gave no trouble, and should be compared with the logs of Sirius and Great Western as summarized by Tyler. The relatively slow piston speed and small power put little strain on the moving parts. Tallow was probably used for lubrication, being introduced into the valve chest by pots on top of the casing, where radiated heat would melt the tallow. From the valve chest the melted tallow was carried into the cylinder, and from there probably passed into the jet condenser. No doubt the lubricant became a sludge that had to be removed from the condenser at least once every 48 hours. There is no real evidence that the engine and boilers suffered any great strains; the operating pressure of steam must have been low at all times. The boilers were probably of very low efficiency and made steam slowly. Fuel consumption was high, and, according to the logbook, the vessel ran out of coal when she reached the English coast; however, she had enough fuel left to steam up the Mersey to Liverpool, probably using wood. At the time she ran out of coal she had used her engine about 80 to 83 hours. While this indicates a fuel consumption of almost a ton per hour, it must be remembered that the intermittent operation of the engine required expenditure of fuel to raise steam in cold boilers over and over again. This was one of the weaknesses in the auxiliary steamship, particularly, as in the case of the Savannah, when the engine was used a number of times during a voyage without long periods of continuous operation. Also, there is doubt that the vessel carried as much as 75 tons of coal; she probably had no more than 55 to 60 tons aboard, if the figure of 1,500 bushels is correct. It is impossible to establish exact weight-cubic measurements with the available data.
Though the authorities quoted seem to agree that the Savannah could steam only 41/2 to 51/4 knots in smooth water, her logbook credits her with 6 knots under steam alone at sea. However, this is probably an approximation affected by current and sea rather than a truly logged speed.
Judging by references in the logbook, the Savannah carried one boat on the stern davits. The davits, shown in Marestier's sketch, would handle a boat of about 16 to 18 feet in length. At sea the boat was probably carried on top of the deckhouse. The vessel obtained a new boat during her European trip. It is probable that the lack of passengers is why a second boat, which could have been stowed on the deckhouse roof, was omitted.
There is no record of how the Savannah was painted, except that the logbook refers to her "bright" strake. Packets appear to have followed what once was a Philadelphia practice in having a varnished band along the topsides. Marestier's sketch indicates that there may have been four or five bands of color, beginning at or a little above deck and wide enough for the top band to be up about two-fifths the height of the bulwarks. The hull was commonly black. The bands were red, white, and blue and there was a "bright" strake, or alternate black and varnished bands. These bands were about 3 to 5 inches wide. Sometimes the "bright" band, as mentioned in the Savannah logbook, was along the topside just above and adjacent to the top of the wale, or belt of thick planking, or might be the uppermost strake of the wale. Perhaps the Savannah had a wide bright band above the wale and multicolored bands just above the deck. The headrails were painted black, with mouldings at top and bottom of rails and with knees picked out with very narrow bands of yellow, or "beading." The figurehead was then commonly painted in natural colors, to suit the form of head if a figure or a bust. The bowsprit and davits probably were black. Deck structures were probably white, the neck natural, with waterways and inside of bulwarks white, the stack black, and rail caps varnished.
In this period it was unusual to copper a wooden ship before launch, so it is doubtful that the Savannah was copper sheathed. Since her voyage occurred during a period of financial depression, it is probable that her bottom was "white" (tallow and verdigris).
The reconstruction described herein produced a plan for a model that complied to the fullest extent with all the known dimensions and descriptions of the Savannah that have yet been found. The result showed that the United States National Museum's old model could not be altered to agree with the known features of the Savannah and that a new model was therefore necessary. So that the new model would be comparable to other models of early American steamers, existing or intended, in the Watercraft Collection, it was constructed on the scale of one-quarter inch to the foot. The new model (figs. [2], [8], and [9]) is now on exhibition at the Smithsonian Institution.
Figure 8.—Stern-quarter view of the new model of the Savannah, showing one wheel partially folded and the iron frames for canvas wheel-boxes in place.
Figure 9.—Bow-quarter view of the new model of the Savannah, showing deck arrangement details.
FOOTNOTES:
[1] Carl W. Mitman, Catalogue of the Watercraft Collection in the United States National Museum, U.S. National Museum Bulletin 127, 1923.
[2] Jean Baptiste Marestier, Mémoire sur les Bateaux à Vapeur de Etats-Unis d'Amérique, Paris, 1823.
[3] A memorandum dated April 20, 1899, in the manuscript file on the watercraft collection shows that the Museum had both the rigged model and the original logbook at that time. Also in the collection were a coffee urn and miniature portrait of the Savannah's captain, Moses Rogers, that had been presented to him abroad; later, these items were returned to the donor. A cup and saucer belonging to Captain Rogers also had been given to the Museum, and they are now in its historical collection.
[4] Robert Greenhalgh Albion, Square Riggers on Schedule, Princeton, New Jersey, 1938. Between the years 1817 and 1837 the yard of Fickett and Crockett also operated at various times under the name of S. & F. Fickett and the name of Fickett and Thomas. The yard appears to have specialized in the construction of coastal packet ships, because only 4 ocean packets, against 24 coastal packets, were built by the various partnerships in which Fickett was a member.
[5] L. M'Kay, The Practical Shipbuilder, New York, 1839.
[6] Howard I. Chapelle, The Baltimore Clipper, Salem, Massachusetts, 1930, pp. 112–134.
[7] Sidney Withington, translator, Memoir on Steamboats of the United States of America by Jean Baptiste Marestier, Mystic, Connecticut, 1957.
[8] Ibid., pl. 7, figs. 32, 33, 35.
[9] Ibid., pl. 3, fig. 10.
[10] Geo. Henry Preble, A Chronological History of the Origin and Development of Steam Navigation, 1543–1882, Philadelphia, 1883.
[11] John H. Morrison, A History of American Steam Navigation, New York, 1930.
[12] David Budlong Tyler, Steam Conquers the Atlantic, New York and London, 1939.
[13] S. C. Gilfillan, Inventing the Ship, New York, 1935.
[14] Charles Frederich Partington, An Historical and Descriptive Account of the Steam Engine, London, 1822.
[15] J. Elfreth Watkins, "The Log of the Savannah," in Report of the U.S. National Museum for the Year Ending June 30, 1890, 1891, pp. 611–639.
[16] Previously, the author had assumed there was one boiler with two flues.
[17] Op. cit. ([footnote 4]).
[18] William Morgan and Augustin Creuze, eds., Papers on Naval Architecture, London, n. d., no. 12, p. 387.
[19] Letter from Carl C. Cutler to the author, November 24, 1958.
[20] Op. cit. ([footnote 5]).
[21] Withington, op. cit. ([footnote 7]), pl. 9, figs. 55, 56.
[22] Report of Malcolm Bell, Jr., and Frank Braynard.
[23] M'Kay, op. cit. ([footnote 5]).
[24] Darcy Lever, Sheet Anchor, London, 1808–1811. (Reprint, Providence, Rhode Island, 1930.)
[25] David Budlong Tyler, Steam Conquers the Atlantic, New York and London, 1939.
[26] S. C. Gilfillan, Inventing the Ship, New York, 1935.