220 PROCEEDINGS OP THE AMERICAN ACADEMY 



Volume of brass weights = 8.55 4- 8.2 1.04 c. c. 



Uncompensated volume = 4.77 — 1.04 3.73 " 



Weight of 3.73 c. c. of air at 300 tenths inch and 



27° C. = 1.176 X 3.73 = corr. required . . 4.39 m.g. 



In this calculation we have necessarily left out of the account both 

 the small amount of carbonic acid absorbed by the soda lime and the 

 water carried by the current from the potash bulb to the connecting 

 tubes ; for the increased weight of these tubes results from both these 

 causes, whose relative effects or whose influence on the correction we 

 canuot estimate! That they wuuld tend to raise slightly the value 

 of the correction is obvious ; and if we assume, as we have in the 

 following calculations, that the value is 4.5 milligrams we shall ap- 

 proximate as nearly to the precise amount as circumstances will per- 

 mit. For all these corrections the data are only known within certain 

 narrow limits ; but the total uncertainty thus arising does not exceed 

 two or three tenths of a milligram in the final result. 



In weighing the potash bulb and U tubes it is of course essential 

 that they should be left in the balance case until a perfect equilibrium 

 of temperature has been reached, and then they must be weighed 

 when open so that a perfect equilibrium of pressure on the inside and 

 outside of the glass may be secured. And since we were here aiming 

 at very great accuracy, we felt it important to inquire further whether 

 any gain of weight from hygroscopic moisture was possible during 

 the time necessary to verify the weight while the tubes remained 

 open on the scale pan. We therefore carefully investigated this 

 point. The air in the balance case was kept as dry as possible by 

 an open dish of sulphuric acid, aud the experiments were made under 

 such conditions. It was found that the weight both of the potash 

 bulb and of the soda lime tube remained absolutely constant, hour 

 and hour together, except so far as they were influenced by changes 

 of temperature and pressure, but the weight of the phosphoric anhy- 

 dride tube slowly but uniformly increased at the rate of 0.025 m.g. 

 per hour. We followed the change on one occasion from I0 h - 45 m - a. m. 

 to 6 h< 15 m - p. m., and during these seven hours and a half the gain was 

 19 m.g., while the changes of temperature and pressure meanwhile 

 were not enough to cause a sensible difference in the atmospheric 

 buoyancy. Such experiments were continued for a sufficient length 

 of time to enable us to confirm the values of the corrections above 

 given, and they gave us great confidence in our weights. They 

 showed that the only precaution necessary for accuracy was to close 



