CONSTANT FOR ELECTROLYTES HEBB. 421 



others If an estimate could be made of the relative value of the 

 various determinations, the weighted mean would give a closer 

 approximation to the true values of the depression constant", 

 than can be given by a mere average. I have not attempted, 

 however, to attach weights to the determinations, being unable 

 to do so with any confidence. 



Assuming that averaging will eliminate the greater part of 

 the errors we get as the values of TO and i, 1.932 and 1.851 

 respectively. The value for i is undoubtedly the more accurate 

 of the two ; and, as we assume that undissociated molecules have 

 the same effect as free ions, the value of TO is to be taken as 

 1.851 also. That the average value of TO is so much greater 

 than that of i may be due not only to the defects of the method 

 employed in determining it, but to the fact that the bend of the 

 curve of an electrolyte is more likely to be to the right than to 

 the left, as concentration increases. For it is only where associa- 

 tion of molecules takes place that it bends to the left. 



Some of the sets of observations given in the paper, from 

 which I have taken my data, I did not use. A few of these 

 sets were so erratic that I could make nothing out of them. 

 The other cases, however, had been worked out by Dr. Mac- 

 Gregor, and, hence, I did not think it worth my while to do so. 

 If, now, I combine my results with those obtained by him, I will, 

 with the exception of the few sets mentioned above, have made, 

 use of all available data. The mean values of TO and i, as 

 given by him, are 1.895 and 1.850 respectively and hence the 

 values of TO and i as given by averaging his mean values with 

 mine are 1.913 and 1.851. 



The above result is in agreement with that reached by Prof. 

 MacGregor by means of the second method referred to above 

 The conclusion he arrived a.t was that, for all the electrolytes 

 examined, the curves of his diagram were consistent with the 

 depression constant having a common value of about 1.85. That 

 the two methods should give results which are not only in close 

 agreement with one another, but are also in close agreement with 

 the value of the depression constant given by Van 't Hoff's 

 theoretical expression for it, must be regarded as of considerable 

 interest. 



