CHANGE IN WEIGHT DURING CHEMICAL REACTION. 
235 
After mixing the contents of B, IG independent determinations of the difference, 
A —B, were made ; from these was calculated the final mean difference 
A—B = +'022 mgr. 
During this set of measurements the observed maximum and minimum values for 
the difference A —B were respectively equal to +’049 mgr. and +’003 mgr. Of the 
16 determinations, 9 showed an average deviation from the mean value = +’009 mgr. ; 
for the remaining 7, the average deviation was = — 'Oil mgr. ; the final mean variation 
may therefore be taken as = +‘010 mgr. 
Concisely stated, the final conclusion reached by means of Series II. is, 
Before mixing the contents of B, the difference A —B = +'026 mgr. 
After „ „ „ „ „ A-B=+-022 „ 
Apparent change in mass = —'004 ,, 
The apparent change here observed is of the order of about 1 in 32,000,000 parts. 
We now proceed to review, very briefly, the results yielded by the two series of experi¬ 
ments, and to indicate the probable causes and sources of the observed variations. 
(G) Of the Possible Errors associated tvith the Experirne^its of Series I. and II.-— 
In itself, and from a certain point of view, the final result secured by means of the 
experiments of Series II. was both excellent and gratifying. We may not, however, 
ignore the important fact, that this result is not in harmony with the final result 
obtained from Series I. As the two sets of measurements were effected under precisely 
similar conditions, the two final values are, so far as can l^e known, equally reliable. 
If we combine the two final values, then the apparent change in mass is equal to 
i(-033--004) = -015 mgr. 
Or, the apparent change in the whole weighed mass is equal to 1 in 8‘5 million 
parts ; this degree of accuracy was frequently attained by Landolt and by him 
sometimes surpassed. 
In our opinion it was now definitely established that the observed apparent changes 
in mass arise, not from any appreciable inherent or unallowed-for defects in the 
balance itself, but that the fluctuations were the natural outcome of certain unsuspected 
variables associated with the glass reaction vessels and their contents. It is believed 
that at least three such variables have been unwittingly allowed to creep, not only 
into my own experiments, but also into the majority of those of other observers. These 
three variables may be generated by— 
(a) Very small yet sufficiently distinct, irregular and unequal convection currents 
in the air near the balance pans. 
(/3) Slight differences in the total external areas of the two vessels. 
2 H 2 
