exploring this region, by means of a narrow slit filled with phosphorescent calcium sulphide, I ascertained, without any difficulty, that, in certain azimuths, the glow of the spark diminished under the action of the rays, and increased, on the contrary, when they were intercepted by a wet screen. These were, in fact, the looked-for radiations; I will call them "N," rays.

Although the aluminium prism of 27° 15′ I used previously is suitable for these experiments, nevertheless, in order to increase the dispersion, I used an aluminium prism of 60°, and afterwards another of 90°. With the help of the latter, I very carefully studied the feebly deviated part of the spectrum. The prism was arranged so that the angle of incidence was 20°; for each radiation, the deviation was measured and the refractive index deduced; then the wave-length was determined by means of a Brunner grating of 200 lines to the millimetre, by the process already described (see p. 57). The following table gives the numbers which result from this study, and were used for constructing the diagram (Fig. 10), in which

the abscissæ stand for the wave-lengths and

the ordinates for the indices diminished by unity.

Each of the divisions marked on the axis of abscissæ corresponds to o'001, and each of the divisions marked on the ordinate axis corresponds to an excess over unity equal

to Ο ΟΙ.

In spite of all the care with which the experiments were executed, the deviations are so small, and, consequently, the indices so near to unity, that the table and diagram can only be regarded as a preliminary indication of the behaviour of the dispersion in the very slightly deviated part of the spectrum. An important consequence arises from these measures, viz. points corresponding to "N" rays, and those corresponding to N, rays, are all situate on the

same curve, within the limits of experimental error. The study of radiations still less refrangible than those I have dwelt on appeared to me impracticable. To avoid confusion, I was obliged to adopt a very large scale for the ordinates; this is why I could not plot on the diagram the results of my former measurements of the more refrangible "N" rays (loc. cit.). These results give points situated on a branch of the curve, starting from the topmost point on the right, and rising almost vertically, with a feeble inclination, from bottom to top, and from right to left, and a slight convexity turned upwards.

Certain sources seem to emit N, rays exclusively, or, at least, these rays predominate in the emission. This is the case with copper and silver wire, and with hard-drawn platinum wire. M. Bichat has observed that ethylic ether, when brought to the state of forced extension, by the process discovered by M. Berthelot, emits N, rays. When this state of strain ceases, whether spontaneously or under the action of a slight blow, the emission of N1 rays immediately disappears.

up

N1 rays can be stored like "N" rays. For instance, one need only bring a bit of stretched copper wire in proximity to a lump of quartz to make the quartz emit N, rays for some time after.

On Peculiarities presented by the Action exercised by "N" Rays on a Dimly Lighted Surface (February 2, 1904).

Consider a phosphorescent screen, or, more generally, a dimly lighted surface. If this surface is viewed normally, one notices that the action of "N" rays is to render it more luminous; if, on the contrary, the surface is viewed very obliquely, nearly tangentially, the action of "N" rays is to render it less luminous. In other words, the action of "N" rays increases the quantity of light normally emitted, while it diminishes the light emitted in a very oblique direction. If one looks at it in an intermediate position, no appreciable effect is observed. This explains the fact, observed