Glass Bulbs in Liquids. 31 



Table VI. 



Gas. y found. y. t. 



Carbonic dioxide... 21*81 22'00 72*5 



Oxygen 17-17 16-00 38*2 



Air 14-13 14-50 30-6 



Water-gas 6-44 6'00 23*0 



The equation is 7 = 26-Q32a/*~' 21 ' 589 . 



Probable error of a single comparison, 0*51. 



Great accuracy in work of this kind is, of course, out of the 

 question ; but for approximate determinations, when only very 

 small volumes of a gas are available, the method will probably 

 be advantageous. 



11. The following set of experiments has reference to a 

 medium consisting of two mixed liquids, the rate of ascent of 

 a bulb in each of which separately is known. One of the 

 liquids was castor-oil, the viscosity of which is very consider- 

 able ; to 100 parts of this, successive quantities z of linseed-oil 

 were added so as to gradually reduce the viscosity. The uni- 

 form temperature was 12°'4. By separate experiments with 

 the oils it was found that, if t were the rate of ascent of the 

 bulb in seconds per centimetre, \ the rate for linseed, « the 

 unbalanced unit pressure in castor, and ft the change in that 

 pressure for every unit of linseed added, then 



(*-X)(« + /fr)»=c, X 



(£--32335)(-09122 + -00169£) 2 =-034376,J 



are equations which fairly represent the experiments. 



Table VII. 



t. z taken. z calc. 



4-45454 0-0 0-0 



3-10027 11-1 11-9 



2-21798 25-0 25-7 



1-52758 42-9 46*0 



1-22140 66-7 61*8 



0-32335 oo co 



Probable error of a single comparison, 1-7. 



Sum of the errors, +0'3. 



12. Castor-oil presents the remarkable property of recover- 

 ing its normal viscosity very slowly after heating. A sample 

 of the oil, through which a bulb rose in a known number oi' 

 seconds per 22 centim., was heated to a temperature not ex- 

 ceeding 80°, and cooled to its previous condition. The bulb 



