Page:Proceedings of the Royal Society of London Vol 69.djvu/169

region in the blue almost entirely. Meanwhile the curvature of the spectrum increases in a most remarkable manner, and the entire red end is lifted high above the green-blue end. As the density of the vapour increases the red gradually fades away, leaving only the yellow and green and the remote blue and violet, the curvature increasing all the while. The fluted or channelled spectrum was described by Roscoe and Schuster about twenty-five years ago, but so far as I know no work has been done on it since. I have recently secured excellent photographs of it with a Eowland concave grating, from the extreme red to the violet, and find that it is much more extensive than has been supposed, for the flutings run right up to the absorption band at the D lines on both sides, though they are very faint on the side of shorter wave-lengths. This spectrum will be described in a subsequent paper.

Very satisfactory photographs of the dispersed grating spectrum were secured, some of which are reproduced. It was found impossible to maintain a sufficiently uniform density, at the low temperature, for a sufficient length of time to enable a negative to be secured showing the appearance before the light between the D lines vanished. I therefore went back to the old plan of using a prismatic flame. After some experimenting it was found that the most satisfactory flame was secured by passing hydrogen through a tube containing metallic sodium, strongly heated, and burning the gas at a flat jet piece made of platinum foil. An exceedingly dense and very uniform sodium flame is obtained in this way, which can be maintained almost indefinitely. The arrangement of the lamp is shown in fig. 3, the diagram requiring no description.