Brief consideration will show that only the latter—the rotational motions—are of any importance when the small displacements due to vibration are in question. To illustrate the negligible effect of vibrations which merely move the camera parallel to itself in any direction it is only necessary to imagine the camera moved parallel to the ground through a large distance, such as 10 centimeters. Now 10 centimeters motion of the camera at 3000 meters elevation means, with a 25 centimeter camera lens, .25 1 ———— × 10 = ———— centimeter 3000 1200 motion on the plate, which would be only a tenth the distance separable by a good lens. If we reduce this motion to the small fraction of a centimeter which vibration would actually produce, it is evident that such vibration is of absolutely no importance. Similarly, if we imagine the camera, under the same conditions, moved vertically with reference to the ground by ten centimeters, the scale of the picture would merely be changed by 1/12000 or by 1/1000 centimeter on a 12 centimeter plate, again quite negligible. When we consider motions of rotation, however, the case is quite different. If the camera is mounted so that the effect of any vibration is to rotate it around a horizontal axis, this is exactly equivalent to rotating the beam of light from the lens so that it sweeps across the plate. Thus a millimeter displacement of the lens of the camera with the plate remaining fixed gives approximately a millimeter motion of the image. Consequently, a rotation producing only a fraction of a millimeter's relative motion of lens and plate during the period the curtain aperture is over a given point would cause fatal blurring—and the visible vibration of plane longerons and cross members is easily of half millimeter amplitude or more. Reduced to angular units it is easily shown that a rotation of one degree per second—which is quite slow as plane oscillations go—is beyond the limits of toleration. Translational motions of large amplitude may be allowed, but the mounting of the camera must not permit these translations to be at all different for different parts of the camera. [Illustration: FIG. 77.—Diagram showing possible motions of the airplane camera: three of translation and three of rotation, and their combinations.]