'276 Messrs. F. Sodcly and T. D. Mackenzie on the 



tap, and between this and the tap of the mercury-pump are 

 placed a soda-lime tube and a phosphorus-pentoxide tube, 

 separated by a tap. A side tube connects with a small bulb 

 of 22 c.c. capacity closed at each end with taps, for admitting 

 small quantities of air to the uranium solution. The flask is 

 stood on the water-bath, and is heated until vigorous boiling 

 occurs. It is then allowed to cool, a volume of air admitted 

 and pumped out into the mercury reservoir. This process is 

 repeated twice to ensure the complete removal of emanation. 

 The gasholder is then connected to the electroscope through 

 a soda-lime tube and the gas run in, a small additional 

 quantity of air being admitted to displace the gas in the 

 soda-lime tube. The exit and the outlet tubes of the electro- 

 scope are then closed with rubber stoppers and observations 

 of the rate of leak taken over about half-an-hour, noting the 

 time of admission of the gas and the time of each observation. 



This method, which may be designated as the "pump 

 method " in distinction to the earlier " bubbling test," has 

 many disadvantages, the chief being that our uranium solu- 

 tions tend to generate nitric oxide when heated. So long as 

 an excess of air is present this does no harm, for it is con- 

 verted into the peroxide and absorbed by the soda-lime. 

 But the leaf-system of the electroscope has been occasionally 

 destroyed by the formation of the peroxide in the electroscope. 

 The method is gradually being abandoned, and that described 

 in the section on the calibration of the electroscope used in 

 its stead. 



It is of interest to give the result of a comparison of this 

 method of testing with the old bubbling method, in the case 

 of a kilogram of uranyl nitrate purified by two extractions 

 with ether. By the bubbling method, iu which the uranium 

 solution was stored in a bottle of rather over one litre capacity, 

 with an air space of about 100 c.c. the leak in the electro- 

 scope* was found to be 0'45. The same solution tested by 

 the mercury-pump method gave 4*5. The comparison is 

 certainly unfavourable to the old method, as with larger 

 amounts of radium it is probable the discrepancy would be 



* Rates of leak in the electroscope throughout are expressed in divi- 

 sions of the eyepiece-scale per minute, corrected for the natural leak of 

 instrument, and unless otherwise stated, are reduced to represent the 

 equilibrium or maximum accumulation of emanation, by means of the 

 formula Le=L t /1 — e — M, where Le is the equilibrium leak, L T the 

 observed leak, t the time in seconds of the accumulation of the emanation, 

 and \ is 2-16xl0~6 (Rutherford and Soddy, Phil. Mag. 1903, [6] v. 

 p. 447). The tables of Gruner (Jahr. Had. u. Elekt. 1906, iii. p. 120) 

 for the exponential function with negative exponent have proved of great 

 use in this work. 



