• Freezing-point Depressions for Electrolytes. 511 



improvements in freezing-point determinations had been 

 recognized as necessary. 



In cases in which there was bat one series of observations 

 available, I have plotted the actual observations in the dia- 

 gram, though sometimes smoothing the curves a little. In 

 cases in which two or more series were available, I first 

 plotted the various observations and then drew mean curves, 

 making them represent all the observations as well as I could, 

 but giving greater weight to recent observations than to 

 those of earlier date, and to long series of consistent obser- 

 vations than to short series or to series which were more 

 erratic. These mean curves were then entered in the 

 diagram. 



The curves are labelled with the initial letters of observers' 

 names (R for Raoult, &c, A for Archibald, Ab for Abegg), 

 so as to show both the depression-observations on which they 

 .are based, and the ionization-coefficients used in plotting 

 them. Thus the inscription KC1 (J-W) means that Jones's 

 depressions and Whetham's coefficients were used ; H 2 S0 4 

 (JLB-B), that the curve is a mean curve based mainly at 

 least on depression-observations by Jones, Loomis, and Barnes, 

 and plotted with Barnes's coefficients. The limits of concen- 

 tration for the curves are indicated also in gramme-equi- 

 valents per litre. The equivalent depressions plotted are in 

 all cases depressions per gramme-equivalent in one litre of 

 solution. 



Some of the curves are entered on an inset drawn on four 

 times the scale of the main diagram. 



In interpreting the curves we must not only bear in mind 

 what has been said above about the probable accuracy of the 

 ionization-coefficients, but must in addition note the tendency 

 exhibited by the curves of the various observers, as dilution 

 increases, to run off at great dilution in directions, character- 

 istic of the observers, to the left or right relatively to the 

 course pursued by them at moderate dilution. Thus Abegg's 

 curves (see NaOl, KC1, K 2 S0 4 ) and Jones's (see NaCl, KG1, 

 NH 4 C1, and HC1) run off to the right. So do Arrhenius's 

 (not plotted), in a marked manner. Raoult's tendency is 

 also to the right (see NaCl ; his K 2 S0 4 , not plotted, shows it 

 also ; his most dilute KC1 observation he himself clearly 

 regards as accidentally out). On the other hand, Loomis's 

 curves (see HC1, KN0 3 , NH 4 C1, BaCl 2 ) go to the left. So 

 do Ponsot's and probably Wilder mann's (not plotted). And 

 I gather from Ponsot's diagrams of Pickering's observations, 

 to which I have not access, that Pickering's also have the 

 leftward tendency. Archibald's and Barnes's curves show 



2 N 2 



