But according to the curious discovery of Malus this is not what happens. What does happen is that the second glass, after being given a quarter turn (though always kept at the same angle), seems to lose its power of reflection almost altogether. The light which comes from it now is diminished enormously, and yet nothing is distorted or displaced; everything is seen correctly if enough light remains to see it by at all, and the ray is said to have been "polarized by reflection." It would be out of place to enter here on the cause of the phenomenon; the fact is certain, and is a very precious one, for the astronomer can now diminish the sun's light till it is bearable by the weakest eye, without any distortion of what he is looking at, and without disturbing the natural tints by colored glasses. In practice, a third and sometimes a fourth reflector, each of a wedge shaped, optically plane piece of unsilvered glass, are thus introduced, and by a simple rotation of the last one the light is graded at pleasure, so that with such an instrument, called "the polarizing eyepiece" (Fig. A), I have often watched the sun's magnified image for four or five hours together with no more distress to the eye than in reading a newspaper. With this, in favorable moments, we see that the sun's surface away from the spots, everywhere, is made up of hundreds of thousands of small, intensely brilliant bodies, that seem to be floating in a gray medium, which, though itself no doubt very bright, appears dark by comparison. What these little things are is still uncertain; whatever they are, they are the immediate principal source of the sun's light and heat. To get an idea of their size we must resort to some more delicate means of measurement than we used in the case of the watch. The filar micrometer consists essentially of two excessively fine strands of cobwebs (or, rather, of spider's cocoon), called technically "wires," stretched parallel to each other and placed just at the focus of the telescope. Suppose one of them to be fixed and the second to be movable (keeping always parallel to the first) by means of a screw, having perhaps one hundred threads to the inch, and a large drum shaped head divided into one hundred equal parts, so that moving this head by one division carries the second "wire" 1/10000 part of an inch nearer to the first. Motions smaller than this can clearly be registered, but it will be evident that everything here really depends upon the accuracy of the screw. The guide screw of the best lathe is a coarse piece of work by comparison with "micrometer" screws as now constructed (especially those for making the "gratings" to be described later), for recent uses of them demand perhaps the most accurate workmanship of anything in mechanics--the maker of one which will pass some lately invented tests is entitled at any rate to call himself "a workman." [Illustration: Fig. 11]