38 THE TRUE VALUE OF a OF VAN DEE WAALS' 



From the molecular weight and valences, taking 14 valences to 

 the molecule, a is computed to be 6.60 X I0 12 , which agrees very 

 well with the value from the (d) formula. Van LaAii gives for a 

 the value 5.33 X 10 12 again about 20% 1)elow tncse values. 



11. Chlorine. Cl v 



(a), a = 2.03 BT C V C 1.03 — P C V C \ T c = 417; P c = 76.10; d c = 

 0.573 ( Pellaton) ; V c = 123.7; M = 

 70.90 ii = 7.28X10 12 



(6). a = 6.bP c F ( 2 a=7.66X10 12 



(c). a = 3MN il *CT c /d . C =2.02 (Johnson and 

 MoInto.sh). #=3.64. d taken as Sd c ; 

 d = 2.086 â = 7.27X10 12 



These values agree fairly well. The mean is 7.40 X Ï0' 2 . K there 

 are valences in the molecule a computed from the valence and 

 molecular weight would be 6.65 X Ï0 12 ; if there are 7 valences, «would 

 be 7.37, which is practically the mean value found ; with S valences 

 a would be S.00. Chlorine everywhere else is trivalent so that 

 with G valences the computed value would be about .1 °/ below 

 the found value. I do not believe the uncertainty in the found 

 value is as great as this. It is not probable that the value is less 

 than the surface tension computation of 7.27. If the truth of the 

 dependence of cohesion on the valence electrons be admitted, then 

 the computation shows 7 valences in the chlorine molecule. We 

 might interpret this to mean that in a chlorine molecule one chlorine 

 atom is oxidized to its maximum and has lost 7 negative electrons, 

 while the other has its maximum number of 7. Of course actually 

 it is to be presumed that they will be distributed between the two 

 atoms. A similar result is found in bromine. The formula might 

 be written as follows : 



: CI : CI : 



Van Laar's value for Cl 2 is 5.S5. This is very much worse than 

 the computation from the valences assuming each atom to be tri- 

 valent so that there are 6 valences to the molecule. It is about 

 20 °j below what I believe to be the true value. If we com- 

 pute a from van deb Waals' equation by the usual method of 

 a = 27 2 7 c 2 /64 X 273 2 i J ( . we find the value 0.3S X 10 12 , which 

 is certainly nearer the true value than van Laar's. 



