The crank being placed by the eye upon its forward dead center, and the eccentric connected to the top of the link being moved round on the axle (in the direction in which the wheels will run when the engine is going forward) until the steam port at the front end of the cylinder is open to the amount of the lead, we fasten the eccentric to hold in that position. We then throw the reverse lever over into the last notch at the other end of the sector, lifting the link up so that the eccentric connected to the lower end of the link may be approximately adjusted, which is done by moving the eccentric round upon the axle (in the direction in which the axle will revolve when the engine is running backward) until the crank stands upon the same dead center, and the front port is open to the amount of the lead. This being done, we have the eccentrics approximately adjusted and may proceed to the final adjustment, in which the first thing to do is to find the exact dead centers of the crank. It is obvious that a line drawn through the center of the crank pin and the center of the wheel axle, will stand horizontally true and level when the crank is on either of the dead centers, but the presence of the crank pin makes it impracticable to draw such a line. We can therefore draw one which will be parallel to those centers; and to do this we draw a circle upon the end of the wheel axle (and from its center) of the same diameter as that of the crank pin, and then resting a straight-edge upon the bearing of the crank pin (taking care to avoid the round corner upon the pin, if there is one), we place the other end of the straight-edge even with the top of the circle drawn upon the axle; and then, using the straight-edge as a guide, we draw a line across the end of the axle and the wheel face. When this line is level the crank will be upon its dead center. This plan is sometimes employed, but is not a very accurate one, because the length of the line is very short as compared to the circumference of the driving wheel; hence, an error of the thickness of the line becomes one equal to several thicknesses of the line when carried out to the wheel circumference. Furthermore, if the line of the cylinder does not stand horizontally level, as is sometimes the case, the result of the whole proceeding will be inaccurate. Again, the connecting rod end and the coupling rod is in the way, rendering it awkward to both draw and level the line.