"The next important point was to combine each pair of wheels by some simple mechanism, instead of the cog-wheels which had formerly been used. My father began by inserting each axle into two cranks at right angles to each other, with rods communicating horizontally between them. An engine was made on this plan, and answered extremely well. But at that period (1815) the mechanical skill of the country was not equal to the task of forging cranked axles of the soundness and strength necessary to stand the jars incident to locomotive work; so my father was compelled to fall back upon a substitute which, though less simple and less efficient, was within the mechanical capabilities of the workmen of that day, either for construction or repair. He adopted a chain which rolled over indented wheels placed on the centre of each axle, and so arranged that the two pairs of wheels were effectually coupled and made to keep pace with each other. But these chains after a few years' use became stretched, and then the engines were liable to irregularity in their working, especially in changing from working back to forward again. Nevertheless, these engines continued in profitable use upon the Killingworth Colliery Railway for some years. Eventually the chain was laid aside, and the front and hind wheels were united by rods on the _outside_, instead of by rods and crank-ankles _inside_, as specified in the original patent; and this expedient completely answered the purpose required, without involving any expensive or difficult workmanship. [Illustration: SECTION OF KILLINGWORTH LOCOMOTIVE, 1815.] "Another important improvement was introduced in this engine. The eduction steam had hitherto been allowed to escape direct into the open atmosphere; but my father, having observed the great velocity with which the waste-steam escaped, compared with the velocity with which the smoke issued from the chimney of the same engine, thought that by conveying the eduction steam into the chimney, and there allowing it to escape in a vertical direction, its velocity would be imparted to the smoke from the engine, or to the ascending current of air in the chimney. The experiment was no sooner made than the power of the engine became more than doubled; combustion was stimulated, as it were, by a blast; consequently, the power of the boiler for generating steam was increased, and, in the same proportion, the useful duty of the engine was augmented.