In August 1710, a jury of the Goldsmiths' Company tested the Mint's gold coinage from that year's pyx. Its own assay found the coin fully up to the legal standard. Its formal verdict, built instead on a comparison with that year's reference plate, found the coin a quarter of a grain short. The Mint's master, Sir Isaac Newton, did not dispute the jury's arithmetic. He disputed which of its own two findings to believe.
The jury's authority rests on a practice already centuries old: the Goldsmiths' Company dates its earliest recorded pyx trial to 1248, the Royal Mint to at least 1282. In the practice as it still runs, coins drawn at random from every batch struck are locked away in a box, the pyx, until enough accumulate for a trial, when a jury of the Company's freemen, sworn in by a royal official called the King's Remembrancer, weighs and counts them, and assays their metal against reference pieces called trial plates. The law allows a small tolerance for ordinary error, the remedy; past it, the fault falls on the Mint.
Newton's own memorandum explains why the two findings differed. A trial plate, he wrote, was properly cast by a jury of workmen summoned under a Council order, after repeated assays confirmed it standard. His memorandum reports that the 1707 plate, cast to mark the union of England and Scotland, ran five-twelfths of a grain finer than true standard by many careful assays — wider than the quarter-grain gap the 1710 trial itself showed, and larger than ordinary assaying was expected to miss. The plate it replaced, cast under James II in 1688, was off by only a sixth of a grain; the one before that, cast under Charles II in 1660, matched the standard exactly. So far as Newton could recall, the 1707 plate had been cast without any such order.
What troubled Newton was not that his coin had failed every test. The jury's own assay had found it sound; the formal verdict reported only the comparison with the plate, and said nothing of the disagreement. Newton himself raised the question: when assay and trial-piece disagree, should the jury be bound to report it, and let the Crown and Council judge the difference? If a flawed reference piece could override its own jury's assay without comment, whoever cast the next one could reset the law's own definition of honest coin at will.
Six years later, on 29 August 1716, twenty-four freemen summoned to the jury opened Newton's pyx again: weighed it, counted it, and assayed it against the standard pieces dated 1688, not the plate he had challenged. The gold pix came to fifty-one pounds eight ounces, worth some two thousand three hundred pounds, within a legal remedy of forty grains to the pound. A surviving copy of the jury's verdict records the gold and the accompanying silver agreeable to the Treasury's own standard, in weight, in count, and in alloy, within the remedies allowed.
The plate that failed the Mint in 1710 had been cast too fine, years before. The jury's own assay caught the discrepancy at the time; its formal verdict never said so, and the error stood as the law's own definition of honest money regardless. Every check against a liar runs on some reference nobody thought to question, until someone does. The coin was never the thing in doubt. The standard was.