parallel to the collimator. The axis is rotated (with a period of 24 hours) by clockwork, and is tilted so as to be parallel to the earth's axis. It was adjusted so that the collimator pointed to the south pole and the camera to the north pole.
The tube of the collimator is prolonged beyond the plane of the slit, and is arranged to carry at its end a mirror of speculum metal and an object-glass, by means of which an image of the sun can be thrown between the two slits.
The whole arrangement thus consists of a spectroscope combined with a polar heliostat, and in virtue of the fact that the spectroscope is rotated together with the mirror, the image of any celestial object thrown on the slit does not rotate relatively to the slit. The mirror is mounted in such a manner that the axis about which it is tilted can be oriented relatively to the collimator tube, so that the slits may be set parallel to any diameter of the sun. On the present occasion the slits were set parallel to the axis of the sun.
In India the slits were set at too great a distance apart, and the light of the corona was not strong enough to impress the photographic plate ; the slits were set so as to catch the corona nearly 8' from the sun's limb on each side. In Algiers the instrument was used with only one slit, and the spectrum of the corona was photographed in two regions to a distance 3}' from the limb with an exposure of 49 seconds, and the resulting negative was a good strong one. In Sumatra I expected to be able to get an exposure of 270 or even 300 seconds, arid to avoid risk of disappointment the slits were set at such a. distance apart as to catch the corona 3' from the sun's limb on each side. This distance was as small as it seemed wise to adopt, account being taken of the apparent diameter and movement of the moon during totality.
The instrument was fairly well adjusted on May 2, and I then took the opportunity afforded by a clear night to throw an image of the full moon on to the slits. It was brightly seen and easily adjusted ; but to make matters more certain, I painted the slit-plate white, so that there might be no possible difficulty in setting the image of the corona between the slits after the exposures had been made for the " flash " spectrum with the crescent across one of the slits. The sequel will show that even this precaution was not enough under the circumstances.
The adjustment of the axis of the instrument to parallelism with the earth's axis was accomplished in the same way as in India and Algiers by moans of a theodolite with declination circle and level, which was attached to a part of the frame of the spectroscope specially prepared for it.
The rotating plate-holder, charged with twelve small plates, was again used for the exposures as in Algiers.
