It had long been felt, however, that there was an unfortunate gap in the evidence respecting stellar motions. The astronomer could tell how much or how little the stars were shifting on the heavens, but he could obtain no measure whatever of other motions which nevertheless must exist among the stars. If a star were receding or approaching, no trace whatever of such motion could be recognised. No instrumental means could enable the astronomer to measure the change of brightness due to the star's change of distance, since such changes must needs be infinitely small compared with the actual lustre of the star.

So that it seemed as though the astronomer must for ever remain ignorant of one most important portion of the stellar motions. All he could do, as it appeared, was to watch the aspect of the heavens, and, as it slowly changed, to infer in what way the stars were moving athwart the line of vision; and even this he could only do most imperfectly, since his knowledge of the distances of the stars is so limited that he can form but inexact notions of the rate at which the stars are so moving. They may be very far away and moving very swiftly, or they may be at a less (though still enormous) distance and moving with a correspondingly reduced velocity.

This source of difficulty was very strikingly illustrated when the subject of the stellar motions was treated in connection with the ideas respecting the sidereal universe promulgated by Sir W. Herschel. In the hypothesis which regarded the stars as spread with a certain general uniformity through a stratum or slice of space, there was no feature which afforded any promise that by the study of the stellar motions the mysteries of the sidereal universe might be interpreted. The very basis of Sir W. Herschel's own researches into the subject is the vague supposition that it is as likely a priori that any given star will move in one direction as in another. Later we find Struve presenting his results in the following form : 'One may wager four hundred thousand to one that a portion of the seeming motions of the stars is due to the sun's motion, and it is an even wager (on peut parier un contre un) that the latter motion takes place at the rate of between 135 and 175 millions of miles per annum.' The whole question had become one of probabilities, based on more or less trustworthy assumptions. We cannot wonder greatly that, when Sir G. Airy undertook the complete reexamination of the matter twenty years ago, the result he obtained, while indicating the general probability of the inferences before obtained, nevertheless exhibited the whole problem as one needing further investigation.1

It will be seen presently that we cannot too attentively regard those earlier researches, if we would fully estimate the importance of the results which have recently been obtained. Let it be carefully noticed that the earlier results flowed directly from the hypothesis respecting the stars which have so long maintained their ground in our text-books of astronomy If these hypotheses are sound, the results flowing from them, even though only based on the general principles of probability as applied to those hypotheses, might be expected to be somewhat near the truth. If, on the contrary, an independent and trustworthy series of results should show that those earlier results are not correct-are indeed very far from correctness-then pro tanto the hypotheses which led to those earlier results would be invalidated.

Let it then be clearly understood that, according to the results in question, the stars were held to be in motion at rates comparable in general with the velocity of our sun, this velocity being estimated at about four and three-quarter miles per second. We do not include here the result that the sun is moving towards Hercules, because that may be regarded as established, whatever

1 This part of my subject is fully discussed in a paper called ' The Sun's Journey through Space,' which appeared in Fraser's Magazine for September 1869, and will be found among my ' Essays on Astronomy.' opinion we may form as to the distribution of the stars in space.

Before proceeding to indicate the bearing of recent observations on these theoretical conclusions, I would invite some degree of attention to the circumstance that the view I am here advancing as to the bearing of new facts on the old hypotheses, is not a new one framed to account for the new facts in a way agreeing with my own theories respecting the stars. More than three years ago in Fraser's Magazine, and earlier still in the proceedings of scientific societies, I indicated my belief that the real facts are precisely such as have now been demonstrated.

Already when I so wrote, promise had been afforded that the astronomer might come in time to know,1 not merely whether certain stars are approaching or receding, but at what rate (in miles per second) these motions are taking place. I need not here enter into an explanation of the method by which this was to be accomplished, inasmuch as a full account of the principle on which the method is based is given in the paper called 'News from Sinus,' in my Essays on Astronomy. Suffice it to say, that it depends on the observed displacement of some known dark line in the rainbow-tinted streak forming the spectrum of a star, and that when such a line is displaced towards the red end of the spectrum it is known that the star is receding, while when the displacement is towards the violet end it is known that the star is approaching.

1 See the closing words of the last paragraph but three in the essay mentioned.

Dr. Huggins, our great spectroscopist, had successfully applied this method to the star Sirius, and he had found that that star is receding from the earth at the rate of upwards of twenty-five miles per second. But Sirius was the only star which could then be examined by this method. The light of Sirius exceeds more than five times that of the next star in order of brightness, at least of those visible in our hemisphere; and with the instrument then at Dr. Huggins' disposal (his own eight-inch refractor) it was found impossible to see the dark lines of any other star-spectrum with a spectroscope dispersive enough to give any measurable displacement of the lines.