Isaac Newton was the right man in the right place at the right time to settle an age-old controversy, born of ignorance of statistics: the Pyx Test to monitor the variability of the weight of silver coins at the end of the minting process. But Newton failed miserably, and continued to make the same mistake he had made in the past, a mistake that would persist for another hundred years, benefiting fraud. In 1701, the great English minting of coins ended. Finally, it was decided to mint coins by machine, so as to make them as equal as possible, thus reducing fraud by those who shaved off excess silver from the coin. The law stated that a single coin could differ in weight from the reference value by no more than 1%, and Newton and his predecessors (and those who followed for a century) assumed that this meant that the average weight of a batch of newly minted coins must also differ by no more than 1% from the reference weight. But today we know (a law discovered by Abraham De Moivre in 1718) that the variability of the acceptable mean weight depends on the number of samples, with the square root of the number of samples. (De Moivre’s equation).
