40 ART. 12. H. HIEOBE : OX THE FUSION SURFACES OF THE 



tables are too large as well as too regular to he referred to 

 experimental error, and seem to show that our starting supposi- 

 tion is not quite correct, perhaps other molecular associations 

 are not strictly excluded in the temperature range studied. 



We have not extended the comparison over other fusion 

 curves and isotherms, because it is almost certain that the de- 

 viations would not be so great as along the two curves we have 

 chosen. 



Yet it is of considerable interest to observe that the fusion 

 surface is expressed within the maximum error of l.ofo in com- 

 position and of 1° of temperature over the tempereture range of 

 50° by the equation thermodynamically deduced starting from 

 only the two boundary curves. Hence it may not be too bold 

 to affirm that within the temperature range given tlie following 

 reaction mainly takes place in the solutions 



(CGHoO)3Tl3(CeHeO). 



The next question is " What side reaction can take place 

 besides the above ? " The above calculation as well as the values 

 of Ü in Table 13, shows that the convexity of the fusion surface 

 towards the edge of the naphthalene is too large to be explained 

 by our simple hypothesis (see Plates I and II). Suppose that 

 there exist double molecules besides triple molecules of phenol 

 and let qu represent the number of these double molecules. Then 



N' = r/i + 2î?,, + 3?î., 

 3Î = n^ + n.2-{-7h. 



... 3dl-N = 2n, + n, 

 2 31-N = ii,+ ri.,. 



Put Ç3Ti-N) = a, (2dl-N) = ß 



