In and before the time of Bradley, the refraction of light was not well determined, which would throw a doubt over any observations made to detect small quantities, unless the star which furnished them were situated in that part of the observer’s heaven in which there is no refraction, or next to none, that is, in or near his zenith. For the purpose of measuring annual parallax, therefore, stars had always been chosen which passed very nearly over the spot of observation, and instruments called zenith sectors (now almost out of use) were employed, which measured small angles of the meridian near the zenith, the latter point being ascertained by a plumb-line. Mr. Molyneux, a friend of Bradley, and a wealthy man, had caused the celebrated Graham to erect a large instrument of this kind at his house in Kew, afterwards the palace. Bradley and Molyneux observed with this instrument the star γ in the Dragon, which passed nearly through the zenith of that place, in December 1725. The star was found to pass the meridian more and more to the south of the zenith, until the following March, when it was about twenty seconds (about the sixty-five thousandth part of the whole circuit of the heavens) lower than at first. It was afterwards traced back again to its first position in the following December, allowing for the precession of the equinoxes. Other stars were examined in the same way, and the result was, that all stars were found to describe small[2] ovals in the course of the year. But on comparing the situations of the stars in their small orbits with the corresponding places of the sun, it was evident that the cause of the phenomenon could not be the change of place arising from the orbital motion of the earth. Various hypotheses proposed by Bradley were found insufficient. In 1727 he erected a zenith sector for himself at Wanstead; and by further observations, and using different stars, he came at length to this fact, that instead of the star being in the place which annual parallax would give it, it was always in the position which it should have had a quarter of a year later: or that if the observer could measure the oval with sufficient exactness, and were to find the time of the year from the star, on the supposition of annual parallax being the cause of the star’s orbit, he would suppose himself in March instead of December, and so on. Footnote 2: The original memorandum of Bradley, on the first night on which a decided result had been obtained, was accidentally found among his papers. There is a fac-simile of it in Professor Rigaud’s work.