Economical and practicable starting of an aeroplane thus requires a free launching space, along which the machine may accelerate with nearly flat planes: a downward slope would be an aid. When the planes are tilted for ascent, after attaining full speed, quick control is necessary to avoid the possibility of a back-somersault. A fairly wide launching platform of 200 feet length would ordinarily suffice. The flight made by Ely in January of this year, from San Francisco to the deck of the cruiser _Pennsylvania_ and back, demonstrated the possibility of starting from a limited area. The wooden platform built over the after deck of the warship was 130 feet long, and sloped. On the return trip, the aeroplane ran down this slope, dropped somewhat, and then ascended successfully. If the effort is made to ascend at low velocities, then the motor power must be sufficient to propel the machine at an extreme angle of inclination--perhaps the third of a right angle, approximating to the angle of least velocity for a given load. According to Chatley, this method of starting by Farman at Issy-les-Moulineaux involved the use of a motor of fifty horse-power: while Roe's machine at Brooklands rose, it is said, with only a six horse-power motor. Descending [Illustration: TRAJECTORY DURING DESCENT] What happens when the motor stops? The velocity of the machine gradually decreases: the resistance to forward movement stops its forward movement and the excess of weight over upward pressure due to velocity causes it to descend. It behaves like a projectile, but the details of behavior are seriously complicated by the variation in head resistance and sustaining force due to changes in the angle of the planes. The "angle of inclination" is now not the angle made by the planes with the horizontal, but the angle which they make with the path of flight. Theory indicates that this should be about two-thirds the angle which the path itself makes with the horizontal: that is, the planes themselves are inclined downward toward the front. The forces which determine the descent are fixed by the velocity and the angle between the planes and the path of flight. Manipulation of the rudders and main planes or even the motor may be practised to ensure lancing to best advantage; but in spite of these (or perhaps on account of these) scarcely any part of aviation offers more dangers, demands more genius on the part of the operator, and has been less satisfactorily analyzed than the question of "getting down." It is easy to stay up and not very hard to "get up," weather conditions being favorable; but it is an "all-sufficient job" to _come down_. Under the new rules of the International Aeronautic Federation, a test flight for a pilot's license must terminate with a descent (motor stopped) in which the aviator is to land within fifty yards of the observers and come to a full stop inside of fifty yards therefrom. The elevation at the