274 
has been added to (a) from 400°.) The calculation teaches that the 
deviations from the found values are least, when the exponent of 
the first factor (viz. ¢:100) amounts to 4, and that of the second 
factor to */,. Every modification in one of the exponents immediately 
gives not only greater, but much greater deviations. *) Here follows 
a survey of the values of the two parts (a) and (6), from 400° C. 
(a) (6) 
400° C. | 12,6486 + 0,0112 = 12,6598 = 12,66 
500° 12,4225 + 0,0230 = 12,4455 = 12,45 
600° 12,2004 + 0,0376 = 12,2380 = 12,24 
700° 11,9825 + 0,0490 = 12,0305 = 12,03 
800° 11,7687 + 0,0334 = 11,8021 = 11,80 
900° 11,5592 — 0,0807 = 11,4785 = 11,48 
1000° _ | 11,3538 — 0,2290 = 11,1248 = 11,12 
11009 _ | 11,1526 — 0,4598 = 10,6928 = 10,69 
1200° _ | 10,9557 — 0,8063 = 10,1494 = 10,15 
1300° 10,7627 — 1,3050 = 9,4577 = 9,46 
1400° | 10,5740 — 1,9985 = 85755 = 8,58 
= 8,30 
1427° | 10,5238 — 2,2254 = 8,2084 
The first table shows clearly that the thus calculated values are 
in perfect concordance with the found values. As we have calculated 
the coefficient 16,4.10-® of the correction term (6) exclusively from 
observations up to 1300° (inclusive), the agreement at 1400° C. is 
the more valuable. We may, therefore, safely assume the calculated 
value 8,30: 2—= 4,15 to be accurate for the critical density. 
6. The Value of D,—D, near T, and that of y at different 
Temperatures. 
We have another means to control the approximate correctness 
of the values of D, and D, e.g. above 900°, and of that of the 
1) Below 400° the correction term (b) is no longer valid. For 300° it would 
yield + 0,0041; for 200° + 0,0009 and for 100° + 0,00006, which values are 
too great. 
*) We point out that the exponent %3 is assumed not to influence the sign of 
8,35 — t/jo9, so that this remains negative for ¢ > 835°. 
