188 PROCEEDINGS OF THE AMERICAN ACADE>IY 



I am not able to say how far my own experiments are at variance 

 with those to which Mr. Lloyd refers. Neither am I able to say how 

 much of the error manifested by my comparisons is fairly to be 

 charged to the general character of the new barometer, and how 

 much is peculiar to the single instrument with which I experimented. 

 I intend, as soon as an opportunity offers, to subject other specimens 

 of the aneroid barometer, both of French and English construction, to 

 the same trial. 



" My next series of experiments consisted in a comparison of the 

 aneroid barometer, day by day, with the common barometer, under 

 the ordinary changes of atmospheric pressure. The mercurial ba- 

 rometer used for this purpose was made by W. and S. Jones, of 

 London, and is the same as that used by Professor Farrar in the 

 barometric observations published by him in Volume III. of the Me- 

 moirs of this Academy. This instrument is furnished with an ad- 

 justment for level, an attached thermometer, and a scale of correc- 

 tions for temperature. This correction, as well as that for capillarity, 

 has been applied to my observations. In this series of experiments it 

 was necessary to know how much the aneroid barometer was affected 

 by a change of temperature. Only a partial compensation is aimed 

 at in the construction of the instrument. An increase of temperature 

 will make the air in the reservoir expand, in the same way as a 

 diminution of pressure. The same increase of temperature, by 

 enlarging the metallic surfaces of the reservoir and increasing its 

 capacity, may sometimes even over-compensate for the increased 

 elasticity of the contained gas. In the instrument which I used, the 

 compensation was deficient, and the amount of the deficiency was 

 determined by exposing the barometer, side by side with a ther- 

 mometer, to a temperature of 32° Fah., and reading the index, and 

 then exposing it to a high temperature (in some instances as high as 

 140° Fah.), and then again reading the index. The difference of the 

 two readings divided by the difference of the two temperatures was 

 adopted as the correction for one degree, and was applied to the daily 

 observations. The value of this correction, as obtained from the 

 mean of five experiments, was .0021 of an inch, with the same sign 

 as in the mercurial barometer. To accommodate the scale of the 

 mercurial barometer, the standard temperature adopted was 55° 

 Fah. The aneroid barometer which I used was not provided (as is 



