OF ARTS AND SCIENCES. 189 



sometimes the case) with an attached thermometer. A thermometer 

 by the side of it, and not under the same inclosure as the air-chest, 

 does not indicate the exact temperature of the working parts of the 

 instrument. The slowness with which the index returned to its old 

 mark after the barometer had been subjected to excessive heat or cold, 

 and was then restored to the temperature of the room, manifests the 

 importance of having the thermometer inclosed as the test of the 

 instrument. The result of this series of comparisons is given in 

 Table II. Although the agreement is much better than with low 

 ranges, it falls far below the requirements of nice scientific investiga- 

 tions. 



" Mr. David Purdie Thompson, in his very recent work on Meteo- 

 rology, has the following paragraph : — ' Upon comparison of indications 

 made with the aneroid barometer — not corrected for the particular 

 temperature — and a very perfect mercurial barometer, given by Mr. 

 Dent, we find that, from forty-nine observations made between the 

 6th of January and the 23d of February, 1848, the mean difference 

 was 0.037 of an inch, the aneroid being in excess; and from sixty 

 similar observations made with a standard barometer, during Decem- 

 ber, 1848, and between the 3d and 31st of January, 1849, the mean 

 difference amounted to 0.0-26 of an inch, the mercurial being in this 

 case in excess over the aneroid barometer. Combining these obser- 

 vations (109 in number), a mean difference amounting to 0.0025 inch 

 is found to exist, the indications of the aneroid being in excess. For 

 general use, the instrument is thus shown to be well suited ; for the 

 measurement of heights, it is peculiarly adapted, from its portability 

 and comparative strength ; and for nautical purposes we know of no 

 better instrument.' — p. 448. 



"Now it will be observed that the mean difference in the twenty- 

 eight comparisons which I have given of the two barometers amounts 

 to only 0.040 of an inch. So far as can be known from such means, 

 the comparison was as satisfactory as in the first set given by Mr. 

 Thompson. Still, the differences in single comparisons are large : 

 whether larger or smaller than in Mr. Dent's observations, I am not 

 able to say, as Mr. Thompson has not given the individual differences. 

 Provision has been made in the construction of the instrument for 

 diminishing the mean difference, as we may alter the rate of the chro- 

 nometer. This mean difference has been eliminated from my com- 



