It will be seen at once that the relation of center of gravity to center of support called for here is in direct contradiction to the requirements for eliminating vibration. Either one requirement or the other must be sacrificed, or else a compromise made in which neither delicate response to inclination of the plane nor fully satisfactory freedom from vibration is attained. This is a very serious objection to the pendular support. But the really vital objection to the pendular support is that it performs its function only very partially. It is entirely satisfactory only under conditions of steady flying, as in a uniform climb or glide, with the plane tail or nose heavy, or in flying with one wing down. As soon as we introduce any acceleration, as in making a turn, the camera follows the plane and not the earth. It is true that mapping photography is done from a plane flying as level as possible, and that except under bad air conditions it holds its course with very little turning, if handled by a skilled pilot. Nevertheless, a surprisingly small deviation from straight flying causes quite serious variations from the vertical. It is of interest to calculate how large may be the horizontal accelerations that accompany swervings from a straight course which one might think insignificant. For instance, consider the horizontal acceleration due to a turn having a radius of a kilometer when the plane is moving at 100 kilometers per hour. If _a_ is the acceleration, _v_ the velocity of the plane, and _r_ the radius, we have from elementary dynamics that _v_^2 _a_ = ————— _r_ Substituting the values chosen, we have— 100,000^2 meters _a_ = ————————————— = .77 —————— 3600^2 × 1000 sec^2 The acceleration of gravity being 9.80 meters/sec^2 we have that the ratio of the horizontal acceleration to the vertical is .77 ———— = .078 9.80 This is the tangent of the angle of deviation from the vertical, from which the angle turns out to be about 4½ degrees, a very considerable error, rapidly multiplied as the speed of the plane is increased. It is, indeed, open to question whether the average deviations from the vertical are not apt to be less with the camera rigidly fixed to the plane, if guided by a skilled pilot who will hold the ship level at the expense of “skidding” the slight turns he must make to hold his direction.