FLUIDITY AND THE CHEMICAL COMPOSITION 123
There is then a somewhat regular decrease in the apparent
value of chlorine as the number of atoms in the molecule are
increased. How much of this is due to constitutive influence
directly and how much can be explained on the ground of asso-
ciation? Ramsay and Shields and Traube agree that carbon
tetrachloride is very little associated if at all, Ramsay and
Shields giving the value 1.01 and Traube 1.00i5o. If then we take
the average of the closely agreeing values of the two compounds
containing four chlorine atoms we obtain as the value of the
chlorine constant 109.5 and with this we can calculate the asso-
ciation of the other compounds. The values thus obtained are
given in the fifth column of Table XXXII. Ethylene chloride
is seen according to this method of calculation to be highly
associated, but Traube has given a still higher value for the asso-
ciation at 15° of 1.46. Data for the other chlorides is lacking,
but calculating the association of propyl chloride by the method
of Traube, the author obtains the value of 1.11 which agrees
excellently with our value of 1.105. The mono-halides seem to
be usually associated according to Traube for he gives for methyl
iodide 1.30, for ethyl iodide 1.19 and for ethyl bromide 1.28. It
is greatly to be regretted that our available data is so meager,
but for the present we can only conclude that the effect of con-
stitution upon the value of the chlorine atom is too small to be
detected.
In reference to the lack of constancy in the value of a methyl-
ene group in Table XXV, it seemed desirable to take the average
of as large number of values as possible, but with the limited
data on hand this made it necessary to include a number of
compounds which are certainly associated. This does not mean
that the value of the methylene group is therefore certainly in
error because associated compounds can give this as well as others,
provided the homologues are equally associated; and even if they
are unequally associated, the average value for the methylene
grouping may not be greatly in error although the individual
differences may be large. Finally the fact that the calculated
values in Table XXV differ from the observed values by less than
1 per cent seems to put a maximum limit upon certain kinds of
constitutive influences.
Hitherto it has been deemed necessary to give oxygen a differ-