Returning to England, Fulton published a pamphlet on steam navigation (1796), but failed to interest the people. Then he became interested in marine torpedoes and submarine navigation. With these ideas he went again to Paris (1802), where he made experiments in working a boat under water, but failed to convince Napoleon, who was then the despotic ruler of France. While thus engaged, however, he met Robert R. Livingston, the American minister to France. In connection with John Stevens and Nicholas J. Roosevelt, of New Jersey, Livingston had already made experiments looking toward steam navigation on the Hudson, and in aid of his enterprise had secured from the legislature of New York the exclusive privilege of navigating New York waters with steamboats. This privilege had expired by limitation, and the experiments had come to an end, so far as he was concerned, but on meeting the enthusiastic Fulton he at once became interested again, and furnished money for further experiments. Thereupon Fulton built a steamboat that, when launched upon the Seine, in 1803, instantly broke in two and sank. The frames were not strong enough. Fishing up the engine, Fulton built a stronger hull, and this time the engine worked, though the speed was so slow that only Fulton and Livingston believed the experiment to be a success. With the optimism that breeds success, Fulton, backed by Livingston, now made drawings of an engine to be used in a boat which he purposed building upon the Hudson, and these plans he carried to England, where he ordered his engine of Watt, who was building the most efficient engines in use. In the meantime, William Symington, a Scotch engine builder, had been experimenting with steamboats. In 1789 he had installed an engine upon a double-hulled yacht, with which one Patrick Millar navigated Lake Dalwinston, in Scotland. Although Millar spent $150,000 in experiments with steam, nothing came of them. In 1801 Symington induced Lord Dundas, president of the Forth & Clyde Canal Company, to build a steam-tug for towing barges. This tug, named the _Charlotte Dundas_, had an engine with a 22-inch cylinder (stroke four feet), and the piston was connected directly with the shaft of a stern paddle-wheel by means of a piston rod and a connecting rod--a plan of such simplicity that it came into universal use later on. The rudder of the boat was handled by means of a wheel placed near the bow, and this plan, too, received universal approval. In March, 1802, this tug towed two 70-ton barges nineteen miles and a half, against a strong wind, in six hours. As a demonstration of the feasibility of steam navigation on smooth water, that passage should have been entirely convincing, but the fear that the wash from the wheel, or wheels (for Symington thought to try side-wheels), would injure the banks of the canal, prevented the adoption of the tug.