16 



Dr. E. P. Perman and Gr. A. S. Atkinson. 



Chief Possible Sources of Error. 



(1) Error in filling and emptying Globe. — To avoid error in filling, the 

 globe was exhausted and filled with bromine seven times, as men- 

 tioned above. Each exhaustion being to at least one-sixth atmosphere, 



the volume of the residual air would be theoretically , i.e.^ less 



than 1/250,000 of the volume of the globe. Similar remarks apply to 

 the emptying. 



As a practical test, a glass globe was filled in the manner described, 

 and on opening it under water, only a negligible amount of air was 

 found to be present. 



(2) The Low Temperature of the Stem. — An approximate correction for 

 this was made in the following manner : — 



Let V = volume of globe. 



8 = vapour-density of bromine in globe. 



T = absolute temperature of globe. 



v\ 8', T' = corresponding quantities for the stem. 



D = apparent vapour-density. 



Then 8 =^ D - — - dY 



V \ i J 



The maximum correction thus calculated is only 0*4. 



(3) Loss of Bromine by Lncomplete Absorption. — The potassium iodide 

 solution in the small bulbs of the absorption apparatus E remained 

 uncoloured, showing that the absorption was practically complete. 



(4) Leakage at the Stopcocks. — In the earlier experiments ordinary 

 stopcocks were used, and were found to le'ak to such an extent as to be 

 useless for the purpose. Well-made diagonal stopcocks were then 

 employed and were found quite trustworthy. 



(5) Error in Temperature Determinations. — It is very difficult to say 

 what the maximum error is ; Ave think it to be not more than 10°, if 

 so much. The chief source of uncertainty was the drift of the zero of 

 the galvanometer ; this was reduced to a minimum by switching in 

 just before taking a reading, 



(6) Expansion of the Porcelain Globe. — A correction was made for 

 this, the coefficient of expansion found by Deville and Troost, 

 0*0000108, being used. This gives practically the same result at the 

 temperatures considered as the formula of T. G. Bedford (British 

 Association, 1899). 



General Conclusions. — The results at atmospheric pressure have been 

 plotted against temperature (see Curve I). From this it appears that 

 the vapour-density of bromine is normal up to about 750° 0. ; at this 

 point dissociation becomes appreciable and gradually increases with 

 rise of temperature. At 1050° C. the density falls to 75*25. 



