I have read with interest the note by Dr. Morize on pp. 29, 30, of the last number of the Journal, with your note thereon. I understand Dr. Morize to mean that when a ray of light is transmitted through both the windows of his magnet‐box there is a refraction error of 7″ 11′, but that he is uncertain whether this arises wholly from one window or partly from each. He explains, however, that refractive error from the window nearest the telescope would not matter, so far as the declination is concerned, supposing use made as usual of a distant mark.It is easy to hold the magnet box in such a position that the line of vision of the telescope, directed to the distant mark, cuts the center of either window (of course at an angle) and passes through an open side. In this way, by interposing and removing the box, each window can be tested separately, without taking the box to pieces. A second method is to fix a piece of looking‐glass on an independent support inside the magnet‐box, and view the image of the wire in a telescope reflected by this with the box away and in situ. The second method allows of normal incidence of the light ; the first if less accurate is simpler, and it easily shows whether there is a refraction error of the order of 10′. Magnetometer verniers usually show only 20′, so that quantities less than 10′, need hardly be considered.