DD13: Truth from the Flames
How do you prove what a piece of gold is made of? You burn away everything that is not gold, and weigh what is left. That, in one sentence, is the fire assay, the reference method by which assay offices still settle gold purity today. What is remarkable is not that it works. It is how little it has changed. Set Isaac Newton's own description of the assay, written at the Royal Mint more than three hundred years ago, beside a British assay office's account of its method now, and the two read almost line for line.
The oldest test in the trade
The Sheffield Assay Office, founded in 1773 and one of the four UK assay offices working under the Hallmarking Act 1973, calls fire assay by cupellation "the industry standard method for determining the gold and silver content of precious metal alloys". It traces the technique to "ancient civilisations throughout Asia Minor around 3,000 BC" and notes that some of the cupel beads "still survive today in a museum in Ankara".
The office also points to the Bible, and the passage is worth reading as a story, because it is a description of the craft. In the King James text of Jeremiah 6, the prophet is set to "know and try their way", and the image is a refiner's furnace that fails: "The bellows are burned, the lead is consumed of the fire; the founder melteth in vain". The metal that will not come clean is "Reprobate silver". Lead, fire, and a result that tells the truth whether anybody likes it or not: the assay was already a metaphor before it was a science.
Sheffield's old motto said the same thing in Latin: "Ex Flammae Veritas", which the office translates as "Truth from the Flames".
Newton's recipe
When Newton was Master of the Mint he wrote the method down, in a memorandum now in the National Archives and transcribed by the Newton Project at Oxford. "Gold & Silver are assayed by refining them & observing how much they are lightned by refining." For gold, the sample is melted with "twice its weight of silver" and "thrice its weight of led till the lead fumes away & carries away with it all the base metal & leaves the gold & silver alone". The bead is "hammered into a thin plate & put into hot aqua fortis", and "the aqua fortis eats away the silver & leaves the gold alone".
Aqua fortis means strong water. The 1823 Encyclopaedia Britannica defines it as "a name given by artists to nitric acid of a certain strength, from its dissolving power". It dissolves silver and leaves gold untouched, which is the whole trick. The same entry notes that gold yields only to aqua regia, a mixture of nitric and muriatic acid.
Newton was exact about the custom. "Assaymasters & Refiners proceed no further then to two waters." In a second memorandum he added that the second water should be "stronger than the first". It was that convention, two waters and no more, which let him show in 1710 that a trial plate could be made too fine, the quarrel told in Newton in the Dock.
The method today
Now Sheffield's account of its own laboratory. The sample "is then wrapped in a piece of lead foil along with a measured quantity of silver, which should be between 2 and 3 times the expected amount of gold present". This step, inquartation, dilutes the gold "to around 25% (literally 'quartering' the gold)" so that the metals can be "parted completely". The sample goes into a cupel "made of a refractory material such as magnesite or bone ash", in a furnace "at 1100°C". Base metals are "either oxidised or absorbed into the cupel leaving behind just the gold and silver, which forms into a bead when cooled".
The bead is rolled "to a thickness of around 0.25mm", annealed "at a temperature of 700°C", wrapped into a cornet and lowered in a platinum basket into "two different strengths of nitric acid". This is parting. What remains is gold, "weighed and compared to the initial weight of alloy sample". For example, the office adds, "9 carat gold is nominally 37.5%".
Read the two accounts together. Lead to carry off the base metal. Silver to open up the gold. A thin plate, so the acid can reach it. Two baths of nitric acid, the second different from the first. A weighing at each end. The chemistry Newton described is the chemistry still in use.
What three centuries bought
What has changed is the measuring. Newton wrote that assay errors "seldome exceed a quarter of a grain in Gold", and that where two or more assays agree and the middle is taken, fineness "may be determined to lesse then half a quarter of a graine". On his scale of 24 carats of four grains each, our arithmetic makes that about 2.6 parts per thousand for a single assay, and about 1.3 for the careful average.
Sheffield runs "at least two control samples consisting of 99.99% pure gold" with every batch and reports an accuracy of "around 1 part per thousand". Its balances weigh "to within 0.001mg", and samples "as small as 50mg" can be taken. The method, it says, is based on the international standard ISO 11426 and accredited by UKAS. Three hundred years of progress, on these two sets of figures, has improved the best practice of Newton's Mint by something like a quarter, and made it routine. Our arithmetic again: one part per thousand of a one kilogram bar is one gram, about 102 pounds at the 101.96 pounds a gram our markets feed returned at 07:06 UTC on 9 October 2026.
Why the flames still matter
Much of the gold sent for hallmarking is no longer burned to be tested. Sheffield says items submitted for hallmarking "are now routinely tested by XRF", a non-destructive X-ray method, and the Dublin Assay Office describes the same choice between chemical analysis and X-ray fluorescence. But Sheffield is plain that "fire assay remains the reference method used to verify the results". The quick test is trusted because the slow one stands behind it.
That is the deep point about gold. It comes through the fire and the acid that take everything else away, and what is left can be weighed. Paper cannot be assayed. A promise cannot be cupelled. Gold can be tested to destruction, and the test is the oldest one in the trade. Every hallmark is a public record that somebody independent did it, and the Trial of the Pyx is the same idea applied to the coinage itself.
Every piece we sell is hallmarked, or its exemption is evidenced. You can read the marks yourself with our free identifier, see what is on the shelf, or sell with Chapman Gold. Nothing here is financial advice.