IOWA ACADEMY OP SCIENCE 
181 
EFFECT OP TEMPERATURE INEQUALITIES ON THE BALANCE. 
BY L. D. WELD. 
The topic herein presented had its origin in some work preliminary to 
another investigation, requiring the use of a sensitive balance operated with 
the greatest possible degree of precision. The balance used is by Sartorius, 
with 18 cm. beam and sensibility of about 2.3 scale divisions per mgr., pans 
empty, and so constructed that the sensibility increases slightly with moderate 
increase of load. The writer was studying this balance with a view to getting 
acquainted with its peculiarities, determining its constants and learning what 
degree of precision to expect of it (as should always be done with a new 
balance about to be used in particular work), and had taken a series of con- 
secutive zero readings by the oscillating pointer method with pans empty, when 
a marked progressive change in the readings attracted immediate attention 
and demanded explanation. 
The first readings, separated by a few moments only, were as follows; 
10.56 
10.56 
10.41 
10.45 
10.37 
10.33 
10.32 
10.27 
10.26 
10.09 
Subsequent sets of readings, in which a fair degree of constancy had been 
secured, gave the probable error of a single pointer reading as less than 0.03 
scale division, so that the change observed must have been real and not due 
simply to errors of observation. 
Variations of temperature were naturally looked to as the probable cause. 
The balance was badly located, with the right end of the case adjacent to an 
outside wall and the left away from' it. At this particular time, the wall was 
cold, and the temperature of the room was rising, so that such an effect was to 
be expected. But balances are frequently used in just such locations, and the 
magnitude of the change, amounting in only a few minutes to what would be 
equivalent with this balance to over 0.2 mgr. placed upon the right-hand pan, 
was such as not to be tolerated in precise weighing. It was thought well, 
therefore, to look into the matter somewhat before entering upon further work. 
After experimenting for a time with sensitive thermometers placed inside 
the case, the writer prepared a thermo-electric arrangement of ten pairs of 
elements, which was suspended immediately behind the beam with the two sets 
of junctions in juxtaposition with the two end knife-edges. This apparatus 
was connected with a sensitive D’Arsonval galvanometer of the wall type, which 
gave a deflection of twenty scale divisions per degree centigrade, so that the 
temperature difference between points very near the ends of the beam could be 
fairly estimated to about one two-hundredth of a degree. The object of this 
arrangement was to study the effect of known temperature inequalities in the 
