The two pistons were first pinned together into a single unit which was probably ringless, since it is believed the walls of the outer piston were too thin to admit rings. Because the piston no longer struck the exhaust valve, a short rod had to be screwed into the pistonhead; this pushed the valve shut at the completion of the exhaust stroke. The remaining problem, the opening of the exhaust valve, was solved by screwing a device to the side of the cylinder which operated from the sidewise motion of the connecting rod. This device shifted a small spacer between the piston and the striker arm of the exhaust-valve rod, permitting the piston to push open the exhaust valve. On alternating strokes the spacer shifted back out of the cylinder; therefore, no contact was made between piston and striker arm. Sometime in February 1893, the altered engine was successfully started. At last the transmission could be tested. Will Russell had come upstairs to watch the trial, and according to a statement by him, given April 30, 1926, Frank, standing to the right of the engine and behind the rear axle, reached forward and with the combination tiller-belt-shifter, moved the belt into driving position. The carriage started forward, but as it approached the wall of the building Frank discovered that he could not get the belt back into the neutral position. In desperation, he grasped the rear axle with both hands and was dragged a short distance, attempting to stop the machine, before it struck the wall. He had, however, sufficiently retarded it so that no damage was done. This short trial demonstrated some of the weaknesses in the friction transmission. Since the speed of the surface of the flywheel, in feet per second, increased in proportion to the distance of the point of contact from the center, the outer edge of the belt attempted to run faster than the inner edge. This conflict of forces not only put an undue load on the motor causing a great loss of power, but it also created a tendency for the belt to work towards the outer edge of the flywheel. Conversely, when the operator desired to return the belt to neutral, it strongly resisted any efforts to slide it toward the center of the wheel, as Frank had learned from the wall-bumping incident. Furthermore, the rubber belt on the friction drum had worn so badly that it had to be replaced at least once during the brief experiments. [Illustration: FIGURE 18.--IGNITION CHAMBER, switch, and breaker contacts of the present Duryea engine.]