In August 1893, possibly as a result of indoor experiments, Frank discovered that the chains running from the small 5-tooth[22] jackshaft sprockets to the large, bronze, wheel sprockets were tight at some times and loose at others. This caused considerable unnecessary noise. The difficulty apparently was the result of the sprockets being cast and not machined. The patternmaker had said he believed he could make the pattern accurately enough so that no machining of the castings would be necessary. Nice castings were produced, but "these sprockets were the reason why an unusual construction was put on the crankshaft [meaning jackshaft]," explained Frank Duryea during an interview at the National Museum on November 9, 1956. Elaborating further, in reply to the queries of E. A. Battison, of the Museum's division of engineering, Duryea told of the problem and the solution when he explained that the sprockets had places where the shrinkage was not even. The hot metal, contracting as it cooled, did not seem to contract uniformly, creating slightly unequal distances between teeth. This resulted in the chain hanging quite loose in some places and in others the tightness prevented adjustment. He contacted Will Russell, foreman of the Russell shop, where the automobile was made, and Russell showed him a device, built by George Warwick, who had made the Warwick bicycle. It was an internal-cut gear, according to Duryea's description, with sprocket teeth on its periphery. With sprockets outside and normal teeth inside, the wheels were about 6 inches in diameter, externally. These little internal-gear sprockets were hung on double-shrouded pinions secured to each end of the jackshaft. A solid disc or housing fitted against both ends of the pinion to prevent the internal gear from working off sideways. Duryea explained the function of these unique little parts: "as soon as tension came on that ring gear that we talked about, it not only tightened the chain hanging on this sprocket on the upper side, but it tightened it on both sides. [The sprocket] rocks right out: both sides of the chain are tight." This feature is one rarely encountered elsewhere, and Duryea, later in the interview said, "To tell you the truth, I think I was just a little bit ashamed about the thing, because I had to pull it off. I didn't like the looks of it after I got it on." Two small tanks, each with a capacity of approximately two gallons, were mounted over the engine in the positions they still occupy, the one on the left for gasoline,[23] the other for water. The small fitting under the gasoline tank has a thumbscrew shutoff and a glass-sight feed tube, leading to the carburetor. The water tank, an inch longer than the gasoline tank, communicates with the water jacket of the engine through two pieces of half-inch pipe, entering the jacket from above and below. The overflow tank, holding just over a gallon, is suspended between the rear axle and the flywheel.