The first screw experimented with was _a_. This screw was run at a high velocity--about 2,500 revolutions per minute--until a screw thrust of 14 lbs. was obtained, and then the governor of the engine was set so that all screws of the same diameter could be run at the same speed. Wishing to ascertain the efficiency of the screw and how much was lost in skin friction, I multiplied the thrust in pounds by the pitch of the screw in feet and by the number of turns it was making in a minute. This, of course, gave the exact number of foot-pounds in energy that was being imparted to the air. I was somewhat surprised to find that it corresponded exactly with the readings of the dynamometer. I thought at first that I must have made some mistake. Again I went very carefully over all the figures, tested everything, and made another experiment and found, even if I changed the number of revolutions, that the readings of the dynamometer were always exactly the same as the energy imparted to the air. This seemed to indicate that the screw was working very well and that the skin friction must be very small indeed. In order to test this, I made what we will call, for the moment, a screw without any pitch at all--that is, the blades were of wood and of the exact thickness and width of the blades of the screw _a_, but without any pitch at all. The extremity of the blade is shown at _f_. I placed this screw on my machine in place of _a_, and although my dynamometer was so sensitive that the pointer would move away from the zero pin by simply touching the tip of the finger to the shaft, it failed to indicate, and thus the screw appeared to consume no power at all. These experiments were repeated a considerable number of times. I then obtained a sheet of tin the same diameter as the screws, 18 inches, and upon running it at the same speed, I found that it did consume a measurable amount of power, certainly more than the two blades _f_. This no doubt was due to the uneven surface of the tin. Had it been a well-made saw blade without teeth, perfectly smooth and true on both sides, it probably would not have required power enough to have shown on the dynamometer. However, it is quite possible that there is a little more skin friction with a polished metallic surface, than with a piece of smooth evenly lacquered wood. The screws which I employed were of American white pine such as used by patternmakers. This wood was free from blemishes of all kind, extremely light, uniform, and strong. When the screw had been formed, it was varnished on both sides with a solution of hot glue, which greatly increased the strength of the wood crosswise of the grain. When this glue was thoroughly dry, the wood was sand-papered until it was as smooth as glass; the whole thing was then carefully varnished with shellac, rubbed down again and revarnished with very thin shellac something like lacquer. In this way the surface of the screw was made very smooth.