This locomotive, like the first, had two vertical cylinders, which communicated _directly_ with each pair of the four wheels that supported the engine, by means of a cross head and a pair of connecting rods. But in attempting to establish a direct communication between the cylinders and the wheels that rolled upon the rails, considerable difficulties presented themselves. The ordinary joints could not be employed to unite the parts of the engine, which was a rigid mass, with the wheels lolling upon the irregular surface of the rails; for it was evident that the two rails of the line of way—more especially in those early days of imperfect construction of the permanent road—could not always be maintained at the same level,—that the wheel at one end of the axle might be depressed into one part of the line which had subsided, whilst the other wheel would be comparatively elevated; and in such a position of the axle and wheels, it was obvious that a rigid communication between the cross head and the wheels was impracticable. Hence it became necessary to form a joint at the top of the piston-rod where it united with the cross head, so as to permit the cross head to preserve complete parallelism with the axle of the wheels with which it was in communication. In order to obtain that degree of flexibility combined with direct action, which was essential for ensuring power and avoiding needless friction and jars from irregularities in the road, Stephenson made use of the “ball and socket” joint for effecting a union between the ends of the cross heads where they united with the connecting rods, and between the ends of the connecting rods where they were united with the crank-pins attached to each driving-wheel. By this arrangement the parallelism between the cross head and the axle was at all times maintained and preserved, without producing any serious jar or friction on any part of the machine. Another important point was, to combine each pair of wheels by means of some simple mechanism instead of by the cogwheels which had formerly been used. And, with this object, Stephenson made cranks in each axle at right angles to each other, with rods communicating horizontally between them.