the slit to the correct width, the lines can be got fairly sharply. Slide your front so that the spectrum falls on the ground glass in just such a position as a quarter plate glass would occupy when in the dark slide, and arrange matters so that the red comes to your left, and the violet to the right, and invariably adopt that plan. It is advisable to include the double H lines in the violet on the right hand edge of your plate. They afford an unerring point from which you can calculate backward, finding G, F, E, etc., by their relative positions to the violet lines. Otherwise you may be mistaken as to what portion of the spectrum you are really photographing. The red should just be seen along the left edge of the quarter plate. When all is arranged thus, you utilize three-fourths of your plate with the spectrum, with just a little clear glass at each end. Before disturbing the arrangement of the apparatus, it is desirable to scratch a mark on the sliding tube, and make a memorandum of the position of all the parts, so that they may be taken away and replaced exactly and thus save time in future. To take a photograph of the spectrum, put a quarter plate in the dark slide and place in camera; point the camera toward a bright sky, or white cloud, near the sun--not at the sun, as there is considerable difficulty in keeping the direct rays exactly in the axis of the spectroscope--draw the shutter, and give, say, sixty seconds. On development, you will probably obtain a good spectrum at the first trial. The duration of exposure must, of course, depend upon the brightness of the day; but if the experiments are to have relative values, the period of exposure must be distinctly noted, and comparisons made for a normal exposure of sixty seconds, ninety seconds, two minutes or more, just according to whatever object one has in view in making the experiments. With a given exposure the results will vary with the light and the width of the slit, as well as being influenced by the character of the instrument itself. Further, all such experiments should be made with a normal developer, and development continued for a definite time. The only exception to this rule would be in the event of wishing to ascertain the utmost that could be got out of a plate, but, under ordinary circumstances, the developer ought never to vary, nor yet the duration of development. To try the effect of various developers, or varying time in development, a departure must be made of such a nature as would operate to bring out upon each plate, or piece of a plate, the utmost it would develop short of fog, against which caution must be adopted in all spectrum experiments.