The doubt as to the actual position of the resultant center of pressure, then, renders the application of the rule uncertain. In practice, we are compelled (unfortunately) after first calculating the balance, by such rules as the above, and after it has been thus found with approximate correctness, to try a preliminary flight. Having witnessed the actual conditions of flight, we must then readjust the position of the wings with reference to the center of gravity, arbitrarily, within the range which is necessary. This readjustment should be small. [Illustration: FIG. 6. Diagram showing effect of Pénaud tail.] In the preceding discussion it has been assumed that, if there is a flat tail or horizontal rudder, it supports no portion of the weight. This is not an indispensable condition but it is very convenient, and we shall assume it. In this case the action of the so-called Pénaud rudder becomes easily intelligible. This is a device, already referred to in Chapter II, made by Alphonse Pénaud for the automatic regulation of horizontal flight, and it is as beautiful as it is simple. Let ‹AB› (Fig. 6) be a schematic representation of an aerodrome whose supporting surface is ‹Bb›, and let it be inclined to the horizon at such an angle α that its course at a given speed may be horizontal. So far it does not appear that, if the aerodrome be disturbed from this horizontal course, there is any self-regulating power which could restore it to its original course; but now let there be added a flat tail ‹AC› set at an angle −α with the wing. This tail serves simply for direction, and not for the support of the aerodrome, which, as already stated, is balanced so that the ‹CG› comes under the ‹CP› of the wing ‹Bb›. It will be seen on a simple inspection that the tail under the given conditions is horizontal, and that, presenting its edge to the wind of advance, it offers no resistance to it, so that if the front rises and the angle α increases, the wind will strike on the under side of the tail and thereby tend to raise the rear and depress [p051] the front again. If the angle α diminish, so that the front drops, the wind will strike the upper surface of the tail, and equally restore the angle α to the amount which is requisite to give horizontal flight. If the angle α is not chosen originally with reference to the speed so as to give horizontal flight, the device will still tend to continue the flight in the straight line which the conditions impose, whether that be horizontal or not.