the position of any ray produced by a substance is affected by no cause but the motion of the substance. How and when this hypothesis may fail is a very important question. It is found, for example, that the position of a spectral ray may be altered by compressing the gas emitting or absorbing the ray, and it may be inquired whether the results for motion in the line of sight may not be vitiated by the absorbing atmosphere of the star being under heavy pressure, thus displacing the absorption line.
To this it may be replied that, in any case, the outer layers of the atmosphere, through which the light must last pass, are not under pressure. How far inner portions may produce an absorption spectrum we cannot discuss at present, but it does not seem likely that serious errors are thus introduced in many cases.
These measures require apparatus and manipulation of extraordinary delicacy, in order to avoid every possible source of error. The displacement of the lines produced by motion is in fact so minute that great skill is required to make it evident, unless in exceptional cases. The Mills spectrograph of the Lick Observatory in the hands of Professor Campbell has, notwithstanding these difficulties, yielded results of extraordinary precision. Quite a number of investigators at some leading observatories of Europe and America are pursuing the work of determining these motions. The determinations have almost necessarily been limited to the brighter stars, because, owing to the light of the star being spread over so broad a space in the spectrum, instead of being concentrated on a point, a far longer exposure is necessary to photograph the spectrum of a star than to photograph the star itself. The larger the telescope the fainter the star whose spectrum can be photographed. Vogel, of Potsdam, who has made the most systematic sets of these measures that have yet appeared, included few stars fainter than the second magnitude. With the largest telescopes the spectro of stars down to about the fifth magnitude may be photographed; beyond this it is extremely difficult to go. The limit will probably be reached by the spectrograph of the Yerkes Observatory, which is now being put into operation by Professors Hale and Frost.
THE MOTION OF THE SUN.
When a star is found to be seemingly in motion, as described in the last section, we may ascribe the phenomenon to a motion either of the star itself or of the observer. In fact no motion can be determined or defined except by reference to some body supposed to be at rest. In the case of any one star, we may equally well suppose the star to be at rest and the observer in motion, or the contrary. Or we may suppose both to have such motions that the difference of the two shall represent the