[Illustration: FIG. 58.--Complete apparatus. P, plant; S, micrometer screw for raising or lowering the plant; C, clockwork for periodic oscillation of plate; W, rotating wheel. V, cylindrical plant-chamber. (From a photograph.)] _Connecting Links._--Another puzzling difficulty lay in the fact that the magnification actually obtained was sometimes very different from the calculated value. This unreliability I was able to trace to the defects inherent in thread connections, employed at first to attach the plant to the first lever, and the first lever to the second. These flexible connections were found to undergo a variable amount of elastic yield. Hence it became necessary to use nothing but rigid connections. The plant attachment, A, of triangular shape is made of a piece of _navaldum_; its knife-edge rests on a notch at the short arm of the lever, L. There are several notches at various distances from the fulcrum. It will be understood how the magnification can be modified by moving A, nearer or further from the fulcrum. The lower end of the attachment is bent in the form of a hook. The end of the leaf of the plant P, is doubled on itself and tied. The loop thus formed is then slipped over the hooked end of A. The link, C, connecting L and L′, consists of a pin pointed at both ends, which rests on two conical agate cups fixed respectively to the upper and lower surfaces of the levers L and L′. This mode of frictionless linking is rigid and allows at the same time perfectly free movement of the levers. _The fulcrum._--The most serious difficulty was in connection with frictionless support of the axes of the two levers. The horizontal axis was at first supported on jewel bearings, with fine screw adjustment for securing lateral support. Any slight variation from absolute adjustment made the bearing either too loose or too tight, preventing free play of the lever. When perfect adjustment was secured by any chance, the movement of the levers became jerky after a few days. This I afterwards discovered was due to the deposit of invisible particles of dust on the bearings. These difficulties forced me to work out a very perfect and at the same time a much simpler device. The lever now carries two vertical pin-points which are supported on conical agate cups. The axis of the lever passes through the points of support. The friction of support is thus reduced to a minimum. The levers are kept in place under the constant pressure of their own weight. The excursion of the end of the recording lever, which represents magnified movement of growth, was now found to be without jerk and quite uniform. EXPERIMENTAL ADJUSTMENTS.