Conduction of Heat in Liquids. 9 



obtain values of a moderate degree of accuracy. In every 

 case, however, a certain amount of error will exist, for glass 

 is a very bad conductor ; and so the inner glass vessel, how- 

 ever thin, will vary in temperature throughout its thickness, 

 and the thermometer thus will not give the temperature of the 

 inuer layer of the liquid. The connexion between the outer 

 and inner glasses, and the divergence of the apparatus from 

 the form assumed in the mathematical investigation, have also 

 some effect. 



The question of convection-currents was carefully considered 

 by Beetz. He introduced some lycopodium-seed into the 

 liquid under investigation and watched it during the experi- 

 ment through a microscope. He had no difficulty in detect- 

 ing currents travelling down the one glass surface and up the 

 other. Experimenting with water at the lower temperature 

 he gradually thickened it with meal till no currents appeared, 

 but found scarcely any change in his value for the conduc- 

 tivity. He even boiled the water and meal and let it cool, so 

 as to form a thick paste, without producing much effect. He 

 thus concluded that at low temperatures the convection- 

 currents were negligible. With our amended formula for 

 the conductivity his conclusion would scarcely be justified ; 

 but without exact knowledge of the change in the density and 

 specific heat it is impossible to deduce trustworthy results. 

 At higher temperatures Beetz found convection-currents to 

 play a decided part. As the thickness of his liquid layer was 

 the same as in Winkelmann's smallest apparatus, these facts 

 go far to justify Weber's objections to Winkelmann's method. 



Beetz examined a very large number of solutions of differ- 

 ent strengths, in all cases stating the density ; thus, where the 

 specific heats are known, his results can be easily modified 

 according to the amended formula. His two series of obser- 

 vations answered to mean temperatures of about 11° and 32° 

 respectively, and thus measures of the increase of the conduc- 

 tivity with the temperature might be deduced. As there is a 

 comparatively small change in the specific heat of unit volume 

 of most liquids for a temperature-variation of 21°, the numbers 

 actually given by Beetz would suffice. The values so deduced 

 are, however, generally several times greater than those ob- 

 tained by more trustworthy methods, which points distinctly 

 to an increase of convection-currents with the temperature. 

 Beetz introduced in the liquids colouring^matter variously 

 affecting radiant heat, but could detect no change in the con- 

 ductivity. He is thus at one with Guthrie in this matter. 



The following Table gives a summary of Beetz's results for 

 such of the liquids and solutions as I can find the specific 



