1510 
It is clear that a steeper course is favourable to the explanation 
given in $ 11. If namely A decreases more quickly with the tem- 
perature, the rate of dissociation will become smaller at the tran- 
sition point, and O’ approaches more to OQ; this favours the coin- 
eiding of OB and O/D’, (see fig. 3). The equation of the line 
dotted in fig. 6 is: 
; 3770 
log K = — —— + 4.274. 
i 7 
If from this we calculate the values for loy K at the temperatures 
of table 8, and from this the ratio of dissociation we find the valnes 
recorded in table 10. 
TABLE 10. 
t P eit | 1032 
280 | 135.0 | 0.652 | 1.454 | 1.808 
290 | 185.3 | 0.647 | 1.599 | 1.776 
| | | 
300 | 252.5 | 0.641 | 1.742 | 1.745 
310 | 341.3 | 0.638 | 1.878 | 1.715 
320 458.1 | 0.632 | 2.014 | 1.686 
| 
330 | 610.6 0.627 | 2.146 | 1.658 
The curve log p= — = + 9.790, which satisfactorily represents 
the observations of the columns 4 and 5 of table 10, yields log p 
— — 0.287 for t=184.5°. This value is given in fig. 5. It is seen 
that the modification is much too small to make a satisfactory 
solution of this question possible. 
A line of still greater slope in fig. 6 must be considered impro- 
bable in my opinion. It is true that the observation at the lowest 
temperature lies appreciably lower than the lines drawn, but then 
this is the least accurate, and must certainly be omitted when the 
most probable line is drawn. Besides the angle at the point of 
transition would become much too large, if we wished to take this 
observation into account. I have already demonstrated before that 
the line which represents log K as function of — 
= 
can present no per- 
ceptible departure from a straight line—I shall come back to this 
in a following paper— and besides the curvature if it were percep- 
