210 GENERAL DISCUSSION OF RESULTS. 



the association of water and other associated solvents, such as the alcohols, and in 

 turn these probably diminish the association of the glycerol. When the association 

 of a solvent is diminished, so also is its dissociating power. 



The viscosities of solutions of the above substances in glycerol and in mixtures 

 of glycerol with water, with methyl alcohol, and with ethyl alcohol, at 25 and 35, 

 were measured. This enables us to calculate the temperature coefficients of 

 viscosity or its reciprocal fluidity, and to compare the coefficients of fluidity with 

 those of conductivity. The fluidities of tenth-normal solutions were measured and 

 were in nearly all cases less than those of the corresponding solvents. Negative 

 viscosity coefficients were, however, found for potassium iodide in water and in the 

 25 and 50 per cent mixtures of glycerol and water at 25 and 35. The measuring 

 of negative viscosity coefficients has already been discussed by Jones and Veazej^. 1 

 While the salt does not lower the viscosity of glycerol itself, it docs lower the vis- 

 cosity of a 50 per cent mixture of glycerol and water. 



The relation pointed out by Jones and Veazey is that only salts of metals with 

 large atomic volumes lower the viscosity of the solvent in which they are dissolved. 

 Although no salt was found which decreased the viscosity of pure glycerol, yet a 

 relation was discovered which was analogous to that found in aqueous solutions. 

 The effect of the several salts on the viscosity of glycerol is inversely proportional 

 to the molecular volumes of the salts in question. Thus, potassium iodide increases 

 the viscosity of glycerol less than lithium bromide, and the former has a much 

 larger molecular volume than the latter. Cobalt chloride increases the viscosity of 

 glycerol more than either of the other salts named, and cobalt chloride has the 

 smallest molecular volume of the three. These relations are strictly analogous to 

 those already discussed by Jones and Veazey 1 for solutions in water as the solvent. 



Although glycerol is more than 1,000 times more viscous than methyl alcohol, 

 yet the same relations seem to hold here as for the less viscous solvents. The fluidity 

 curves for glycerol and water, and glycerol and the alcohols, resemble very closely 

 the conductivity curves in these solvents. These curves show the same sagging 

 below the straight line of averages, and have no minima. 



The temperature coefficient of fluidity in pure glycerol, between 25 and 35, 

 is 11.5. This is slightly greater than the temperature coefficient of conductivity 

 over this range of temperature, which is about 10.5 per cent. The larger value 

 of the temperature coefficient of fluidity is probably due in part to the decrease in 

 dissociation with rise in temperature, the molecules having less frictional surfaces 

 than the ions into which they dissociate. 



A comparison of the conductivity and fluidity curves in glycerol show, then, that 

 the two phenomena run nearly parallel; glycerol, therefore, resembles water much 

 more closely than it resembles the alcohols. 



The molecular conductivities in glycerol do not reach the limiting value at any 

 of the dilutions studied. At four-hundredth normal the molecular conductivities 

 increase very slowly, showing that complete dissociation is reached at moderate 

 dilution. Glycerol is, then, a strongly dissociating solvent, as would be expected 

 from its large dielectric constant. 



'Carnegie Institution of Washington Publication No. 80. 



