56 PROCEEDINGS OF THE AMERICAN ACADEMY 



be traced, and most analysts are probably not aware of the extent to 

 which their weighings may be influenced by changes in the density 

 of the air due to variations of temperature and pressure. In seeking 

 to fix the weights of certain absorption tubes, (in connection with my 

 work on the revision of the atomic weights,) I have been led to a 

 method of correcting the weights of such tubes for variations of buoy- 

 ancy ; which, while it does not involve the determination of any data 

 except the temperature and tension of the air in the balance-case, and 

 is as simple in its application as the calibration of a flask, also gives a 

 clear conception of the effect of each variable on the weight. 



It is assumed that the air of the balance-case is dry ; and with one 

 of Becker's balances I have not been able to trace any effect on the 

 weight of a glass vessel from variations of hygrometric condition, 

 when two open dishes of sulphuric acid (three inches in diameter) 

 were kept in the case, which has a volume of about thirty-seven cubic 

 decimeters. Under such circumstances, the only causes which sensibly 

 modify the weight of a small glass vessel (like a closed potash bulb-tube) 

 are the variations of temperature and pressure. The relative effect of 

 these two variables will appear from the following considerations, 

 which suggested the method I am to describe. 



If we assume thirty inches of mercury as the standard of barometric 

 pressure, it is obvious that the variation of each tenth of an inch from 

 this standard will determine a change of ^J^ in the resultant effect of 

 the buoyancy of the air on the load and its equipoise. Again, if we 

 assume 27° C. as our standard of temperature, — that is, 300° on the 

 so-called '• absolute scale," — then, according to the law of Charles, the 

 variation of each degree from this point will also cause a change of 

 -^^■Q in the same resultant. In other words, counting from these 

 standards, a variation of one degree in the Centigrade thermometer 

 indicates the same effect on the density of the air, and therefore on its 

 buoyancy, as the change of ^ of an inch in the mercurial barometer. 

 In our climate the barometer changes slowly, and its fluctuations do not 

 ordinarily exceed one inch. On the other hand, the balances in our 

 chemical laboratories are liable to rapid changes of temperature, which 

 often exceed twenty degrees, the equivalent of two inches. Hence, 

 of the two variables the temperature is by far the more important. 



If we select the two standards of temperature and tension here 

 assumed, we can easily correct for temperature by simply adding 

 to the observed height of the barometer (in tenths of an inch) 

 the difference between 27° C. and the temperature observed. — Of 

 course the correction becomes negative if the temperature exceeds 



