indeed what occurs. The interposition of a sheet of mica, set so that its axis makes an angle of 45° with the plane of action of the radiations emitted by the tube, destroys their rectilinear polarization, for their action on a small spark remains sensibly the same, whatever be the direction of the spark-gap. If a second sheet of mica is interposed, identical with the first, so that the axes of the two sheets are perpendicular to each other, rectilinear polarization is re-established. This result can also be obtained by the use of a Babinet's compensator. Consequently, we are dealing with elliptic polarization.
Now, if the sheet of mica changes rectilinear into elliptic polarization, such a sheet must be doubly refractive for the radiations thus transformed. But if double refraction exists, a fortiori simple refraction must exist; and I was thus led to examine whether, in spite of the fruitless attempts to discover the refraction of "X" rays, I could not obtain a deviation by a prism. I then arranged the following experiment: a focus tube sends through an aluminium screen a pencil of rays, limited by two vertical