On an Inaccuracy (having its greatest value about 1″) in the usual method of computing the Moon's Parallax

Extract

When, as in the usual operation, the moon's observed zenith distance is corrected for the effects of atmospheric refraction, the zenith distance so obtained is that of the rectilineal part of the ray of light between the planet and the upper surface of the air; and on applying that correction, as at the Observatory, we do not obtain the direction of the moon as it would have been seen if there had been no atmosphere, but that of a line drawn parallel to the first part of the ray, and therefore passing below the moon.