88 ANNUAL RECORD OF SCIENCE AND INDUSTRY. 



the air is in movement as when it is at rest. Barometric 

 measures made in the tower of the cathedral at Anvers when 

 the air w^as calm gave results sensibly accordant with the 

 true heiofht as measured bv the level. On the other hand, 

 measurements made when the west wind prevailed gave re- 

 sults too high, and those made when the east wind prevailed 

 gave results too low, the difference being fourteen feet. It 

 does not appear, however, that Montigny has taken account 

 of the relative directions between the wind and the open- 

 ing of the windows and doors of the room in which he was 

 placed a point which, as is well known from recent investi- 

 gations, is of the highest importance if so great accuracy 

 would be attained. Bull. Acad, de J3elgique^ JSruxelles, 

 1875, 820. 



ON THE ACCUKACY OF THE ANEROID BAEOMETER. 



The increasing accuracy of the aneroid barometers as man- 

 ufactured by the best makers, and the numerous applications 

 of this instrument to the determination of altitudes, have led 

 to several excellent investi2:ations into the errors and relia- 

 bility of the instrument, among which one of the best is due 

 to G. Grassi, recently published in the supplement to the il/e- 

 teorologia Italiana. The first instrument examined by him 

 was made by Casella, of London, which was compared with 

 an excellent siphon-barometer. Subsequently a number of 

 others, some of them made by different makers, were ob- 

 tained, and subjected to the same course of investigation. 

 Grassi summarized his results as follows : First, aneroids that 

 are subject to great variations of pressure need corrections 

 which have a regular progression, and are generally of very 

 similar form. Second, each aneroid must have its own proper 

 corrections. Third, there exist pressures for which the cor- 

 rections are nearly constant. Fourth, there exist other press- 

 ures at which the corrections experience sudden changes, 

 sometimes an increase, sometimes a decrease. Fifth, if the 

 corrections are plotted as curves, the ascending or descend- 

 ing portions of these will be sensibly parallel. Sixth, sudden 

 jumps are in a positive direction if the preceding period is 

 one of increasing correction, and vice versa ^ the jumps are 

 negative if the preceding period is one of diminishing cor- 

 rection. Seventh, the magnitude of a sudden jump is pro- 



