[Illustration: Fig. 13.--Apparatus for testing the direction of air currents caused by a rapidly rotating screw. Silken threads were attached to the wire _c_, _c_, which indicated clearly the direction in which the air was moving.] In many of the treatises and books of that time it was stated that a screw propeller, working in the air, was exceedingly wasteful of energy on account of producing a fan-blower action. Some inventors suggested that the screw should work in a stationary cylinder, or, better still, that the whole screw should be encased in a rotating cylinder, to prevent this outward motion of the air. In order to ascertain what the actual facts were, I attached a large number of red silk threads to a brass wire, which I placed completely around my screw (see Fig. 13). Upon starting up I found that, instead of the air being blown out at the periphery of the screw, it was in reality sucked in, as will be seen in the illustration. I was rather surprised to see how sharp a line of demarkation there was between the air that was moving in the direction of the screw and the air that was moving in the opposite direction. The screw employed in these experiments was 18 inches in diameter and had a pitch of 24 inches. It was evident, however, if the pitch of the screw was coarse enough that there would be a fan-blower action. I therefore tried screws of various degrees of pitch, and found when the pitch was a little more than three times the diameter, giving to the outer end of the blade an angle of 45°, that a fan-blower action was produced--that is, part of the time when the screw was running, the air would alternate; sometimes it would pass inwards at the periphery and sometimes outwards. The change of direction, however, was always indicated by a difference in the pitch of the note given out, and also by the thrust. In Fig. 14 I have shown the extremities of the blades of some of the different forms of screws experimented with, in which _a_ shows a plain screw, the front side being straight and of equal pitch from the periphery to the hub; _b_ is a screw of practically the same pitch, but slightly curved so as to give what is known as an increasing pitch; _c_ shows the extremity of a screw in which the curve is not the same throughout--that is, it is what is known as a compound increasing pitch; _d_ is the shape of the screw that gave the angle of 45° above referred to. [Illustration: Fig. 14.--This drawing shows the ends of screw blades in which _a_ is a plain screw; _b_, screw with increasing pitch; _c_, screw with compound increasing pitch; _d_, end of screw blade 45°; _e_, screw with very thick blade; _f_, blade with no pitch at all; _g_, blade which gave a thrust in the direction of the convex side, no matter in which direction it was revolved; _h_, screw said to have been used in the French Government experiments.]