XC.—The limitations of the balance
Journal of the Chemical Society Transactions
ABOUT three years ago the author had occasion, in the course of a research, to make accurate weighings extending over some months, The balances used were by the first makers, and the discrepancies observed were at first attributed t o the errors of experiment arising from the difficulties of the research. Slowly and reluctantly, however, the conclusion was drawn that the errors lay in the balances themselves, and the following is an account of the nature and extent of these errors. It is
... accepted that a good balance carrying 200 grams in each pan should turn with certainty with 0.1 milligram, an accuracy of one part in two millions. Much higher accuracies have been claimed, but f o r the purpose of the research in question this degree of accuracy was sufficient. It was assumed that it could be obtained without difficulty, and the research was started on this assumption. As has been mentioned above, the cause of the error was not a t first suspected, and after many months of work the balances themselves were critically tested. It may be stated a t once that all were usually accurate over a short period of a few hours or even a few days, but over longer periods of a few weeks or months they showed themselves untrustworthy. I n all, six balances were used, three being provided by the kindnem of Professor Pope a t Cambridge; the other three were a t the laboratory in the author's house.