Researches on the Elementary Law of Hydrodiffusion. 523 



column in order to compare it with that obtained from the 

 column preceding. In the case of iron the two values acci- 

 dentally agree exactly. 



43. As regards the concluding section of the former paper 

 (the suggestions for future experiment), I have seen no reason 

 to modify it in any way. I still think the rods should be ex- 

 amined by a thermoelectric compensation method (i. e. with 

 the use of a bridge wire and slider, and with fixed thermo- 

 electric joints on the rod) after their surfaces have been uni- 

 formly coated with stove black-lead. The exhaustion of the 

 chamber containing the rod should proceed to the " neutral- 

 point " observed by Mr. Crookes, and employed by Mr. Poyn- 

 ting in his delicate weighing experiment (Proc. Roy. Soc), 

 " when convection currents have ceased and the radiometer 

 effect has not yet begun." It is probable that at such a point 

 (if it really exists, as I suppose it does) Dulong and Petit' s 

 law of cooling, assumed all through this paper, would hold 

 with perfect accuracy ; and I think that with very careful 

 observations of temperature over a sufficient range on long 

 thin rods, there would be no insuperable obstacle to obtaining 

 rapidly the absolute conductivity and the coefficient of its 

 variation with temperature of as many metals and alloys as 

 one pleased. 



I have taken considerable care ; and I hope that all the cal- 

 culations in the present paper are free from error ; but they 

 are rather long and unsymmetrical and therefore tiresome, 

 and I have not ventured to ask any one to read over the proof 

 for this very reason. 



LXI. Researches on the Elementary Law of Hydrodiffusion. 



By H. F. Weber. 



[Concluded from p. 500.] 



Second Method. 



Description. — The above-described first method requires 

 rather long periods of time for the observations, in consequence 

 of which the investigation is very tedious and, on account of 

 the absolute necessity of a constant temperature, in most 

 localities impracticable. It certainly appears possible to 

 shorten the duration of the observations considerably by 

 diminishing the length L ; but L can scarcely be allowed to 

 fall much below the value employed in the above experiments 

 if the initial state of the diffusion-vessel is to remain sharply 

 definable. 



This propertv of the method used induced me to seek other 



2N2 



