70
SECTION B.
which there is a considerable difference in path between the two pencils.
The instrument which $ had the honor of describing to the section at the last meeting is free from all the objections mentioned. It is simple in construction ; with a little familiarity it is easily adjusted ; it may be used with a broad luminous surface ; the pencils may be separated as far as desired ; and when properly adjusted the position of the fringes is perfectly definite.1 As an additional advantage it may be stated that this is probably the onty form of instrument which permits the use of white light (and therefore of the identification of the fringes) without risk of disturbing the position of either surface by contact or close approximation. It is chiefly this property which renders this instrument peculiarly adapted to the comparison of standards of length.
As this form of refractometer has already proved its value in several experiments already completed and in the preliminary work of others now under way, I may be permitted to recall the chief points of its construction and theory. A beam of light falls on the front surface of a plane parallel piece of optical glass at any angle—usually forty-five degrees—part being reflected and part transmitted. The reflected portion is returned b}' a plane mirror, normal to its path, back through the inclined plate. The second or transmitted portion is tilso returned by a plane mirror and is in part reflected by the incited plate,/tlms coinciding with the transmitted part of the first pencil; and the two pencils are thus brought to “ interfere.”2 A little consideration will show that this arrangement is exactty equivalent to an air-film or plate between two plane surfaces. The interference phenomena are therefore the same as for such an air-plate.
If the virtual distance between the plane surfaces is small, white light ma}^ be employed and we have then colored fringes like Newton’s rings or the colors of a soap-film. If the distance exceeds a few wave lengths, monochromatic light must be employed.
«•
1 It may Incidentally be mentioned that extraneous reflections—such as usually accompany most ol' the phenomena of interference—may be almost entirely avoided by ft transparent Him of silver on the front snrfacc of the glass plate where the rays separate; and accordingly the fringes in white light present a purity and gorgeousness of coloration that is only rivaled by the colors of the pulariscope.
* A second plane parallel plate of the same thickness and Inclination is placed (for compensation) in the path of the first pencil.
ADDRESS BY ALBERT A. MICHELSON.
71
TVe may confine our attention to the case of two parallel surfaces. Here it can readily be shown that the fringes are concentric circles, the common axis of the rings being the normal passing through the optical centre of the eye or telescope. Further they are most distinct when the eye or the telescope is focussed for parallel rays. In any other case we are troubled with the same perplexing changes of form and position of the fringes as already noted.
If now one of the mirrors have a motion normal to its surface the interference rings expand or contract; and by counting the fringes as they appear or disappear in the centre, we have a means of laying off any given distance in wave lengths.
Should this work of connecting the arbitrary standard of length — the yard or the metre — with the unalterable length of a light wave prove as feasible as it is hoped, a next step would be to furnish a standard of mass based upon the same unit. It may seem a little like exaggeration to say that the solution of this problem may admit of almost as high a degree of accuracy as the preceding.
Suppose a cube, ten centimetres on a side, with surfaces as nearly plane and .parallel as possible. Next suppose a testing instrument made of two parallel pieces of glass, whose inner surfaces are slightly farther apart than an edge of the cube.
The parallelism and the distance of these surfaces can be verified to a twentieth of a wave. Now apply this testing instrument to the three pairs of surfaces of the cube and determine their form, parallelism and distance to the same degree of accuracy. We have thus the means of measuring the volume of a cubic decimeter with an error less than one part in a million.
A very convenient and accurate method of making the determination of the weight of this volume of water at its maximum density has been suggested by Professor Morley, which consists in making the cube hollow, so that it will have almost exactly the same density as the water. On weighing the cube in water the excess of weight may be as small as required and may be most accurately measured by a very light and sensitive balance.
It does not seem extravagant to say that by some such plan as this we may obtain a standard kilogram which will be related to the standard of length with a degree of approximation far exceeding that of the present standard.
In the manufacture of plane surfaces, the only practicable method
