1900 ] on Beccnt Studies in Gravitation. 293 



But in neither case could the experiments he taken as showing 

 a real couple. They only showed that, if it existed, it was incapable 

 of producing an effect greater than that observed. 



Perhaps the best way to put the result of our work is this : 

 Imagine the small sphere set with its axis at 45° to that of the other. 

 Then the couple is not greater than one which would take 5^ hours 

 to turn it through that 45° to the parallel position, and it would 

 oscillate about that position in not less than 21 hours. 



The semicircular couple is not greater than one which would 

 turn from crossed to parallel position in 4^ hours, and it would 

 oscillate about that position in not less than 17 hours. 



Or, if the gravitation is less in the crossed than in the parallel 

 position, and in a constant ratio, the difference is less than 1 in 

 16,000 in the one case and less than 1 in 2800 in the other. 



We may compare with these numbers the difference of rate of 

 travel of yellow light through a quartz crystal along the axis and 

 perpendicular to it. That difference is of quite another order, 

 being about 1 in 170. 



As to other possible qualities of gravitation, I shall only mention 

 that quite indecisive experiments have been made to seek for an 

 alteration of mass on chemical combination,* and that at present 

 there is no reason to suppose that temperature affects gravitation. 

 Indeed, as to temperature effect, the agreement of weight methods 

 and volume methods of measuring expansion with rise of tempera- 

 ture is good, as far as it goes, in showing that weight is independent 

 of temperature. 



So while the experiments to determine G are converging on the 

 same value, the attempts to show that, under certain conditions, it 

 may not be constant, have resulted so far in failure all along the line. 

 No attack on gravitation has succeeded in showing that it is related 

 to anything but the masses of the attracting and the attracted bodies. 

 It appears to have no relation to physical or chemical condition of 

 the acting masses or to the intervening medium. 



Perhaps we have been led astray by false analogies in some of 

 our questions. Some of the qualities we have sought and failed to 

 find, qualities which characterise electric and magnetic forces, may 

 be due to the polarity, the + and — , which we ascribe to poles 

 and charges, and which have no counterpart in mass. 



But this unlikeness, this independence of gravitation of any 

 quality but mass, bars the way to any explanation of its nature. 



The dependence of electric forces on the medium, one of 

 Faraday's grand discoveries for ever associated with the Royal 

 Institution, was the first step which led on to the electromagnetic 

 theory of light now so splendidly illustrated by Hertz's electro- 

 magnetic waves. The quantitative laws of electrolysis, again due 



* Landolt, Zeit. fur Phys. Chem. xii. 1, 1891. Sanford and Eay, Physical 

 Review, v. 1897, p. 247. 



