In the drawing (Fig. 18) I have shown screw blades of a proper shape to give the best results--that is, providing a metallic screw is employed. Instead of having the arm of the screw on the back of the blade to offer resistance to the air, the arm should be tubular, flattened, and covered on both sides with sheet metal. This particular formation not only prevents the air from striking the arm, but, at the same time, prevents the pressure of the air from deforming the blade, so, if a metallic screw is to be used, the form of blade which I have shown will be found much superior to that employed at the present time on continental flying machines. We should not lose sight of the fact that weight tells very seriously against the success of a flying machine, and that no expense should be spared to reduce the weight, providing that it is possible to do so without reducing the factor of safety. Suppose, for example, that we use an ordinary hub secured to a solid shaft by a common key. All the parts have to be made heavy in order to be sufficiently strong to withstand the strain. In the drawing (Fig. 20) I have shown a hub which I think is quite as light and strong as it is possible to make it. The action of the motor is often spasmodic and puts very great strain upon the parts, and there is a very strong tendency for the shaft to turn round in the hub. If a key is used, the hub has to be large and strong, and the key of considerable size, otherwise the parts would be deformed. In my own experiments, I have found considerable difficulty in securing a shaft to wooden screws. However, it will be seen in the drawings that a series of grooves is cut in the shaft and that the hub has internal projections, so that the one fits the other. This makes a very strong connection and is of extreme lightness. Both the hub and the shaft should be of tempered steel. The spokes should be hard drawn steel tubes with long fine threads, so as to withstand centrifugal force. To prevent them from rotating in the hub, the nuts _d_, _d_ are provided, which compress the arms of the steel hub so as to grip the tube with any degree of force required. It will be seen that with this system the pitch of the screw may be adjusted to some extent; however, it is better to have all parts of the screw, from hub to centre, of the same pitch. A slight deviation from this is admissible in the experimental stage, so long as the deviation from a true screw, caused by rotating the arm, is not greater than one half of the slip while in flight.