124 instance, be less than as determined by observations above the pole. A.mong the causes to which this tnay be ascribed, the fol-]o,ving deserves notice :-On comparing Dollond's baro1neter (by the height of which the refractions have always been calculated) 'vith one constructed by Troughton and Simms, it appeared that the reading of the former was steadily less than that of the latter by 0·1 inch. Assu1ning Troughton and Simms' baron1eter to be correct, the calculated refractions ought to be increased by about