On Units and the Theory of Relativity. 431 



an assumption of the same kind will enable the passivity of 

 metals in acids to be regarded as due to a layer of anions 

 firmly held on the surface of the metal. 



The law of Mass Action holds good in the case of strong 

 electrolytes if we represent active mass by a momentum 

 term, as is certainly necessary in all applications of the law. 

 Bates (J. Am. C. S. xxxvii. p. 1421, 1915) has suggested that 

 the abnormality of strong electrolytes is due to the abnormal 

 osmotic behaviour of the ions, and this suggestion is in 

 entire agreement with the fundamental assumption of this 

 investigation. Perhaps the departure of osmotic pressure 

 from all known gas laws may be explained on a similar 

 assumption. 



Bedford Modern School, 

 May 16, 1921. 



XL VII. On Units and the Theory of Relativity. 



To the Editors of the Philosophical Magazine. 

 Gentlemen, — 

 N the January number of this Magazine, Sir George 



i 



Greenhill asked for a short explanation of the state- 

 ment made in some works on Relativity, that the mass of 

 the sun is equal to 1*47 kilometres. I accordingly sent 

 to him the following, which he suggests may be of interest 

 to others besides himself ; and I venture to forward it for 

 publication — without wishing to express any opinion on the 

 main problem as to the value of the Relativity theory. 

 It gives me, however, an opportunity to say that the use 

 of these special types of units has (almost always) led to 

 much confusion when used by those who are not expert 

 mathematicians. 



Supposing that the earth's orbit round the sun is treated 

 as a circle of radius a, described in periodic time T, and take 

 the constant of gravitation to be denoted by y ; then if the 

 sun's mass is taken to be S, we have the equations 



7 S/a 2 = a(2<7r/T) 2 , 



or 7 ST 2 = 4ttV. 



Let c denote the velocity of light ; then we have the 



result 7 S 4^3 



(1) 



c 2 (cT) 2 



On inserting the usual astronomical data, the value of 



