OF ARTS AND SCIENCES. 207 



all the nas left in the connecting tubes will soon be replaced by air. 

 At the freezing point the internal volume of the balloon, as nearly as 

 we have been able to measure it, is 4961.5 cubic centimeters, and the 

 weight of the glass 570.5 grams. After much experimenting, we have 

 concluded that as great accuracy in determinations of gas density ran 

 be obtained with a balloon of this size as with a larger one. The ad- 

 vantage to be derived from the greater mass of gas experimented on 

 is more than compensated by the greater liability to error which the 

 weighing of larger vessels and the longer duration of all the processes 

 involve. 



The extent to which this balloon is actually compressed by the ten- 

 sion of the atmosphere when a vacuum is made in the interior has been 

 carefully measured,* and amounts to 1.66 cubic centimeters for a 

 difference of tension of 760 mm. between the interior and exterior of 

 the glass vessel ; and this corresponds to a loss of buoyancy of 1.96 

 milligrams in the air at 700 mm. and 22° C. 



When the balloon was not hanging on the balance, it was always 

 protected by a metal case (see Plate II.), from which the neck ami 

 connecting tubes alone protruded, and in transferring the balloon to 

 and from the balance it was handled only by the neck. 



In Regnault's experiments the balloon was filled with the gas und< r 

 examination at the temperature of 0° C. by surrounding the glass 

 with melting ice. This necessitates a careful cleaning of the globe at 

 least twice during each determination, and thus may arise slight varia- 

 tions of condition to impair the accuracy of the results. We have 

 found with Lord Rayleigh that it is preferable to fill the balloon at the 

 temperature of the laboratory, but we must then observe the tempera- 

 ture with very great accuracy. It is important to know the tempera- 

 ture to the one hundredth of a centigrade degree, for even so small a 

 difference of temperature as this causes a change in the amount of 

 carbonic acid gas which our balloon will hold under ordinary atmos- 

 pheric conditions corresponding to three tenths of a milligram. 



We have found it possible to secure even such constancy of tem- 

 perature as this delicate measure implies by placing the balloon while 

 filling in a calorimeter case, and observing the temperature with a 

 calorimetric thermometer divided to the fiftieth of a centigrade degree. 

 The calorimeter case consists of a cylindrical vessel of sheet zinc wide 

 enough to receive the balloon with its tube and leave a play of lesa 

 than an inch on either side. This vessel is fastened by metal Btaya 



* These Proceedings, vol. xxiii. \>. 182. 



