CHANGE IN WEIGHT DURING CHEMICAL REACTION. 
229 
both the probable error and the limit of accuracy in weighing, as estimated by Landolt, 
are just five times as large as the final value for the apparent decrease in weight 
during chemical reaction. The only legitimate conclusion that could be drawn was 
that which Landolt himself drew, namely, that if there occurs any real change in 
the total mass of reacting substances, then in the cases investigated any such change 
is less than +1 in 10,000,000 parts. This then was Landolt’s final verdict after 
much tedious and highly refined work conducted with patience, skill and resource¬ 
fulness, during the lengthy period of 17 years. 
For the sake of brevity and to avoid repetition, certain other very important 
points connected with Landolt’s investigations are considered in later portions of this 
paper ; their influence upon the final result is also discussed. 
In concluding these introductory remarks, we may observe that a number of 
experiments bearing upoji the apparent want of strict constancy of mass have been 
conducted by Heydweiler.* The reactions chosen were carried out along the lines 
set down by Landolt ; but Heydweiler’s estimated possible experimental error was 
larger than Landolt’s, being = + '04 mgr. ; therefore for the purpose wliich we have 
in view further comment is unnecessary. 
In an earlier communication to the Royal Society, the present author suggested 
that the difierences in weight recorded by Landolt were in all probability due, at 
least in part, to the presence of slight and unsuspected disturbing factors within the 
balance beam itself In the belief that this was the correct view, in June, 1910, one 
of Landolt’s experiments was repeated under circumstances still more favourable 
than those, which obtained during the German chemist’s investigations As the 
conditions, precautions and refinements which are absolutely indispensable when 
weighings of the highest order of accuracy are to be undertaken have been described 
and discussed in two former communications, the author would beg to refer any 
reader seeking information upon those points to his two other papers.! 
We now proceed to describe— 
{a) The manner in which the various experiments were conducted ; 
(b) Some additional refinements which were introduced for securing a still higher 
degree of accuracy in the weighing of counterpoised glass vessels ; and 
(c) The final results obtained by the several methods and their bearing upon 
Landolt’s work and conclusions. 
(2) The Balance.- —The balance used in this research was one which has already 
been fully described.J We may, however, repeat that the chief features of the 
instrument are as follows :—The beam is cantilever in type ; its length is 14 cm., and 
the maximum load for which the balance was built = 200 gr. The beam is cut out 
from solid, hard-rolled phosphor-bronze plate; the terminal knife-edge blocks are 
* A. Heydweiler, ‘Ann. d. Physik.’ (4), vol. V., 1901, p. 394. 
t ‘Phil. Trans.,’ A, vol. 210, pp. 387-415; ‘Roy. Soc. Proc.,’ A, vol. 86, 1912, pp. 591-600. 
I ‘Phil. Trans.,’ A, vol. 210, p. 407. 
