512 



'-?=/(? 



when /' represents -r^: 



We can put in approximation 0,51:486 = 0,00105 lor /'', so 

 that /V would become 



fs" = (578 X 0,4343 (0,00860 ^ 0,00074) = 295 X 0,00934 = 2,76. 



According to the above table, this value is too small, as it would 

 be still smaller than the value of ƒ'" at 397°, viz. 3,09. We expect, 

 indeed, a lower value than 3,60, as the latter represents the chord 

 in the curve ?/=/'(,i;), and 2,76 tl)e tangent — but not a value so 



much smaller. The value — = 0.153 given by Heijcock is therefore 



dp 



probably too high ') — or else the value assumed for /' is too low. 



Also the value 7'^. = 1970 can have been assumed too low. 



4. Potassium. After the above explanations we can be briefer, 

 and simply repeat the same calculations as above. 



The melting point lies at 62°, 5 C. = 331°, 6 abs. The boiling point 

 at 762°,2 C. = 1035°,3 abs. (Hetjcock and L.). Ruff and Joh. give 

 757°, 5 C The critical temperature lies, therefore, at 335,6 X ^Vj = 

 = 1790, 1035,3 X 1,7 = 1750, mean 1770° ; or at 335,6 X 47, = 

 = 1566, 1035,3X1,5=1554, mean 1560° abs. These will again 

 appear to be the limiting values. 



Thus [/Tjc becomes = 42,08 or 39,50, hence 2y = 2,599 to 

 2,501, 7 = 1,30 to 1,25. Tl^is gives the value from 0,918 to 0,926 

 for the factor /, and the value 74,29 to 74,90 for y. For X 

 we have used y : fy -j- 1) = 0,565 to 0,556. 



M Or would 0,153 be a printer's error for 0,135 V See Tables Annuelles of 1912, 

 which are full of misprints. Then 860 would become 975, and 0,00934 would 

 become 0,0105, because of which f.s^'> would become from 2,76 to 3,09. And this 

 value is very well possible and would — like the expansibility — point to a value 

 of Tk which would be slightly higher than 1970, e.g. 2000" abs. 



