The Weighing.
The balance used was a No. 8 long-armed one, made by Becker and Sons. It was supported by iron brackets fastened to one of the foundation walls of the laboratory.
Here it would be subjected to the least jar and was also well protected from air currents. All weighings were made between the hours of one and five in the morning when the surroundings were as quiet as could be desired. A very slight disturbance was detected by the vibrations on the surface of a cup of mercury placed conveniently between the pans.
That the presence of the operator might not produce any change in the balance during the weighing, he closed the room, placed the light above and behind his head and took his position in front of the balance at least an hour before making a weighing. When his presence no longer affected the balance (which was shown by the zero point remaining constant in a series of determinations) the weighing was begun. The method of weighing by vibrations and upon both pans was employed throughout.
Each zero point was taken as the mean of three closely agreeing zero determinations; each one of the three being the mean of seven readings. The zero of the balance empty was determined just before and after each weighing to detect any change in its position. Usually none was observed. The sensibility of the balance was taken at each weighing with the weights used at that weighing. A displacement of the zero point about six divisions of the ivory scale was effected by the addition of one milligram.
The weights had been especially adjusted and were carefully compared with each other before using.
Weighing by tares was adopted as preferable to any other method. By this means all errors resulting from changes in the moisture of the air were avoided and any errors which might have been introduced by heating or manipulating the crucibles would be counteracted by treating the tare in exactly the same manner.