A glass rod _a_, _a′_, Fig. 33, is sustained by two wheels _b_, _b′_, and kept in contact, with them by a third friction roller _c_, pressed downward by a spring. This rod carries a circular frame _d_, _d′_, upon which at _e_, _e′_, _e″_, are three glass holders and platinum catches. A spring _f_ holds the sensitive plate in position, by pressing against its back. The circular frame _d_ is kept in one plane by a fourth friction roller _g_, which runs on a glass rod _h_, and is kept against it by the inward pressure of the overhanging frame _d_. The cord _i_ is attached to the arm _k_, and pulls in the direction of the glass rod _a_. From _m_ to a fixed point near _b_, a strip of elastic India-rubber is stretched, to keep the cord tight. The ring of brass _n_, _n′_ carries the whole, serving as a basis for the stationary parts, and in its turn being fastened to the eyepiece holder, so as to allow the glass rod _a_ to change direction, and be brought into coincidence with the apparent path of the moon. At _o_ is a thumb-screw or clamp. Through the ring _n_, _n′_, a groove _p_ is cut, into which a piece of yellow glass may be placed, when the actinic rays are to be shut off from the plate. Since this contrivance has been completed, all the previous difficulties have vanished. The moving of a plate can be accomplished with such precision, that when the atmosphere was steady, negatives were taken which have been enlarged to three feet in diameter. The length of time that such a slide can be made to run is indefinite, depending in my case on the size of the diagonal flat mirror, and aperture of the eyepiece holder. I can follow the moon for nearly four minutes, but have never required to do so for more than fifty seconds. At the mouth of the instrument, where no secondary mirror is necessary, the time of running could be increased. The setting of the frictionless slide in angular position is accomplished as follows: A ground glass plate is put into it, with the ground face toward the mirror. Upon this face a black line must have been traced, precisely parallel to the rod _a_. This may be accomplished by firmly fixing a pencil point against the ground side, and then drawing the frame d and glass past it, while the rest of the slide is held fast. As the moon passes across the field, the position of the apparatus must be changed, until one of the craters runs along the line from end to end. A cross line drawn perpendicular to the other, serves to adjust the rate of the clepsydra as we shall see, and when a crater is kept steadily on the intersection for twice or three times the time demanded to secure an impression, the adjustment may be regarded as complete. It is necessary of course to expose the sensitive plate soon after, or the apparent path of the moon will have changed direction, unless indeed the slide is set to suit a future moment. b. _The Clepsydra._