Dr. Roxburgh on the Cartesian Barometer. 411 



was emptied, when it was found that the salt having crystalUzed 

 between the mercury and the glass, had so allowed the air to 

 enter. 



A solution of muriate of lime, not being crystallizable, was 

 next tried ; and this seems to stand best, as yet having sunk in 

 2| years only -03 of an inch, the greater part of this depression 

 having occuiTed in the first few months, giving rise to the sui*- 

 mise that the air which has caused it was left in at the time of 

 filling, and has not crept in since. The addition of the salt to 

 the water, besides removing to a great extent the air, has the 

 effect of diminishing the correction required for the force of 

 vapour ; the last named solution has its boiling-point at 234° F., 

 and, as has been shown by experiments, the tension of vapour 

 from water and watery solutions of salts is the same at an equal 

 number of degrees below their boiling-points, the correction to 

 be applied is lessened to that of pure water 22° lower than the 

 observed temperature. This correction, which is to be added, 

 and that for the expansion of the fluids, which is to be subtracted, 

 thus nearly neutralizing each other at low temperatures, I have 

 applied by means of a moveable scale, in the same way as is used 

 in the sympiesometer. Among the slighter variations shown by 

 this barometer, may be mentioned the oscillations during a gale 

 of wind ; these are quite as conspicuous in this barometer as they 

 were observed by Professor Daniell in the water barometei*, 

 amounting frequently to 0-03, and once to 0*4 of an inch ; 

 they vaiy in duration from 5 to 7 seconds ; they begin with a 

 short, quick rise, followed by a slower and much greater descent, 

 and tlien a return to the point of rest, which is much nearer the 

 top than the bottom of the oscillation. Previously to a gale of 

 wind, the column descends by jerks and with irregular rapidity ; 

 but on one occasion, on which no wind followed for two days, 

 the column fell without the slightest jerk more than half an inch ; 

 there was, however, a heavy and long-continued fall of rain. 

 During heavy and sudden showers the column rises, and falls 

 again on the cessation of the shower ; on one occasion the I'ise 

 was '02 of an inch. In a room with a fire, with door and win- 

 dow shut, the column is lower than when the window is open ; 

 the difference is usually "005, but with a good fire '01 of an inch. 

 The last two ca'ses are veiy likely to give rise to error, and the 

 better the barometer the greater will be the error. 



38 Gloucester Place, Portman Square. 



