IOWA ACADEMY OP SCIENCE 
183 
be kept constant throughout the various operations of the double weighing, 
with the opening and closing of the case that are necessary in any balance not 
provided with special reversing apparatus, the interchange method does not 
remedy the difficulty. 
The writer made two double weighings of the same object by the interpola- 
tion method, using the same individual weights in both cases, but under differ- 
ent temperature conditions. The first set was made on a warm day, the room 
being considerably colder than outside; while the second set was made on a 
cool day, the room being steam heated. The balance had not been adjusted as 
to arm ratio since its arrival from the makers; and the right arm was still 
toward the outside wall. The results were as follows: 
FIKST SET. 
Object on left pan 19.23112 grams 
Object on right pan 19.23219 grams 
SECOND SET. 
Object on left pan 19.23148 grams 
Object on right pan 19.23213 grams 
The right arm is obviously the longer in both cases. But if the mean of the 
first set be taken, it will be found that the residual or departure of each 
weighing from it is 0.00052, while for the second set the residual is only 
0.00032. The explanation of this is, that in the first case the right arm was 
warmer than the left, thus adding to the inequality, while in the second, it was 
cooler, subtracting from it. The difference in temperature at the ends of the 
beam could not have been in either case more than one or two-tenths of a 
degree, as indicated by the galvanometer. 
The beam of the balance here used is of phosphor-bronze, with a centigrade 
coefficient of expansion," according to the makers, of 0.0000181. A simple calcu- 
lation, using this figure and the known constants of the balance, shows that 
a uniform temperature gradient of one degree from one end of the beam to 
the other, barring air currents, would produce an anomalous deflection of about 
1:8 scale divisions, corresponding to well toward a milligram. The thermo- 
electric apparatus, when used under the best conditions obtainable in the 
winter with the balance situated as it was, though carefully screened and 
blanketed, showed temperature gradients of nearly one-fourth of one degree, 
and much greater ones when no special precautions were taken. The varia- 
tions observed are therefore not at all surprising. 
In conclusion, it may be suggested that sufficient care is seldom taken in 
conducting balance work to secure freedom from the disturbing influences of 
temperature gradients, direct radiation and accidental air currents. For suc- 
cessful work, the balance should be centrally located in a quiet, thermostatic 
room, on a pier if possible, and carefully screened from radiations from heating 
apparatus, artificial lights and sunlit windows. It is well to blanket the case, 
and the illumination should come from the front, not from the side. Unless 
such precautions are taken, it is doubtful if the great precision often claimed 
for weighing methods can actually be realized. Further, it will bo found a 
source of satisfaction, to anyone endeavoring to get the best results from 
weighings, to install within the case, as described, a thermo-electric series of 
simple design, which will give warning of the presence of temperature inequal- 
