NO. 17.] BAROMETERS. 



The above-mentioned 16 comparisons in 1893 gave, as a mean correction 

 of this barometer, -f- 0*153 mm. The mean temperature of the attached 

 thermometer was 9'4. But the corrections found were apparently greater in 

 lower temperatures than in higher. Dividing the 16 comparisons into two 

 groups, each containing 8 observations, and taking the means for each group, 

 I obtained 



Temperature t of Attached Thermometer Correction M. E. 



5-3 + 0-214 mm. 0*084 mm. 



13-4 + 0-092 0-026 



Mean 9*4 + 0*153 0'055 



The correction comes out greater with lower temperatures, and vice versa. 

 Correction = + 0*153 0*0151 (t 9*4) = -f 0*295 0*0151 t. mm. 



On board ship, Adie 763 was repeatedly compared with Fuess 362, from 

 October, 1893, to December, 1895. The observations were made in the cabin, 

 the barometers being suspended on the wall, side by side and at the same 

 level. Fuess 362 was frequently observed with vacua of different volume. 

 The observations with different vacua gave the following results for the 

 difference between the correction found for Adie 763 with the lower vernier 

 of Fuess 362 put at mm. of the scale, and with the vernier put at n mm. 

 (less volume of vacuum) 



n= 10 20 30 40 50 



Diff. + 0*015 00 0*00 - 0'04 + O'Ol Weighted mean 0*00. 



These figures indicate that neither the standard barometer Fuess 362 

 nor Adie 763 contained any air. The correction found for Adie 763 was, 

 from the observations with the auxiliary scale (n = 0), + 0*078 mm. 



reduced vacuum, + 0*078 - 



The observations in the cabin were numerous, but the circumstances 

 were not so favorable for obtaining good results as at the Meteorological 

 Institute. They were made by lamp-light, and the temperature of the attached 

 thermometer was steadily rising during the series of comparisons. The 

 attached thermometer of Fuess frequently showed from 1 to several degrees 

 higher than that of Adie at the same moment. So great a difference was 

 not caused by a difference in the errors of the thermometers themselves. 

 Taking only those comparisons in which the difference between the thermometers 

 was less than 1, the resulting correction was as follows: 



