NOTES AND COOLIDGE. — ELECTRICAL CONDUCTIVITY. 211 



to those with potassium chloride at 18°. The Kohlrausch formula ex- 

 presses the results for both salts at all temperatures without great error,* 

 and the same is true of the Barmwater formula, except at the highest 

 temperature, where the deviations with both salts are large. The van't 

 Hoff and Rudolphi formulas do not accord at all with the observed values 

 at y06°, the deviations in the case of the latter formula being especially 

 large, while at the lower temperatures of 140, 218, and 281° the van't Hoff 

 formula is far less satisfactory than those of Kohlrausch and Barmwater. 

 On the whole, therefore, the simple Kohlrausch formula furnishes the 

 best representation of the results, and the Barmwater the next best, — facts 

 which are directly indicated by the final means at the foot of the last 

 table. "Whether within the range of concentration in question (0.0005 — 

 0.1 normal) the deviations corresponding to the former are really less 

 than the experimental errors cannot be decided with certainty : the 

 greatest deviation (see Table X) from the very accurate values of Kohl- 

 rausch and Maltby at 18° is 0.13 per cent f in the case of sodium chloride, 

 and 0.06 per cent in that of potassium chloride ; the greatest deviations 

 at the highest temperatures are 0.55 per cent at 281° and 0.40 per cent 

 at 306° in the case of sodium chloride, and only 0.20 per cent at 306° in 

 that of potassium chloride. It seems improbable that the experimental 

 errors are as large as these deviations in the case of the sodium chloride ; 

 but it is perhaps not impossible. 



It may be of interest to state also the percentage deviations of our 

 straight line corresponding to the Kohlrausch function from the points 

 representing the conductivities of sodium and potassium chloride at 18° 

 in the still more dilute solutions investigated by Kohlrausch and Maltby. 

 These deviations are — 0.53 and —0.42 per cent, respectively, in case of 

 the 0.0001 normal solutions, and — 0.36 and —0.25 percent, respectively, 

 in that of the 0.0002 normal solutions. It is to be noted with reference 

 to the significance of this disagreement, that the conductivity of the water 

 used for the 0.0001 normal solutions formed from 7 to 10 per cent of that 

 of the salt, and that the results were corrected for it under the assump- 

 tion that it was uninfluenced by the addition of the salt. Kohlrausch and 

 Maltby, however, consider it almost impossible that from this source an 

 error of the magnitude of these deviations can arise. 



* Compare also Fig. 9, which is an accurate plot of the values for sodium 

 chloride at 218°. 



t Assuming tlie first two as well as the last two deviations to have been com- 

 pletely equalized, which could have been done by sUghtly displacing the represen- 

 tative straiglit Une. 



