same at these times, then it is to be inferred that there is a sensible difference in the direction of the lines E'S and E"'S.
Now you will observe, that there is a good deal assumed in this. In the first place, when we observe the North Polar distance of the star, the thing to which we refer it, is the position of the Pole in the heavens. The Pole of the heavens is in reality that point of the heavens defined by continuing the earth's axis in a straight line to the region of the stars, so that when we observe the North Polar distance of the star, we do really determine the position of that star in relation to the position of the earth's axis. We assume, therefore, that we can account fully and accurately for the change of position which the earth's axis has undergone, and which arises from precession and nutation. Now there is always a very minute uncertainty about these, which it is desirable to get rid of. Another cause of uncertainty arises from aberration; on this account there is always an uncertainty of a fraction of a second, which enters into the observation. There is also in most cases another cause of uncertainty; it is that of which I have spoken so frequently, refraction, which is such a trouble to astronomers. Nearly every observation which we make upon the positions of the stars is affected by refraction, and after making all proper allowance, we cannot always answer for the results. These considerations serve to cast some doubts on the observations made for determining the distances of the stars. Still there is one star of which the parallax seems to be determined with considerable accuracy, and that is the bright star of the Centaur, Alpha Centauri. It appears certain that this star has an annual parallax of one second (using the term "annual
