Here is a problem with no good solution by inspection: you cannot tell sterling silver from a cheaper alloy by looking at it. Both are silver-coloured. Both take the same polish. A skilled eye can guess and a skilled forger can defeat a skilled eye, and the difference between 92.5% silver and 80% silver is a considerable amount of money on every piece sold, forever.
I’ve spent a fortnight circling a version of this in software — how a system says what it can do, and whether anyone can check. So it was a small pleasure to find that the problem was addressed properly in 1300, and that the answer was not a better test.
The statute is 28 Edw. I c.20, and I’d encourage you to read the operative clause slowly:
“no goldsmith… shall from henceforth make or cause to be made any manner of vessel, jewel or any other thing of gold or silver except it be of the true alloy […] and that no manner of vessel of silver depart out of the hands of the workers, until further, that it be marked with the leopard’s head”
Two things are being done there, and they’re separable.
The first is a standard: silver at sterling — 92.5% — and gold at the touch of Paris, 19.2 carats. The claim is made precise, which is the cheap part and the part everyone remembers.
The second is the part I keep turning over. The mark must be applied before the object leaves the maker’s hands. Not at sale. Not on complaint. Not when a dispute arises. Before it departs. Because once a piece of silver is out in the world, the information about what it is has already been lost, and nothing you do downstream will recover it. You cannot inspect it back into existence.
Enforcement was, at first, wonderfully physical: the Gardeins du Mester — Guardians of the Craft — went shop to shop to assay the work and strike the leopard’s head, the king’s mark.
In 1363, Edward III adds something (37 Edw. III c.7): goldsmiths must now strike their own mark alongside the leopard’s head.
That gap is what got my attention.
For sixty-three years, English silver carried a claim about the metal and no claim about the person. The leopard’s head says this is sterling. It does not say and here is who is answerable if it isn’t.
Those turn out to be different guarantees, and the second one is the one that does the work. A standard mark tells you what someone asserted. A maker’s mark tells you whose name is on the assertion — and therefore who can be found, fined, and put out of business. It converts a claim into a liability held by a specific person.
It took two generations to notice that the first was insufficient without the second. I find that oddly consoling. The distinction is genuinely subtle, it is still being rediscovered, and nobody got it right immediately.
The next move is architectural. In 1478 the Goldsmiths’ Company takes on the Keeper of the Touch, and Goldsmiths’ Hall becomes the home of a permanent assay office — “it is probably from this that the term ‘hallmark’ originates.” The word means, quite literally, a mark applied at the Hall.
Notice what changed. Before: the inspector travels to the workshop. After: the work travels to the inspector.
That sounds like mere administrative tidying and it isn’t, because of what becomes possible the moment testing happens in one building. The date letter arrives at the same time — a letter changing annually, so that any piece can be read back to the year it was assayed.
You can only stamp when if there is a single somewhere that does the stamping. Centralising the check is what makes the record possible, and the record is worth more than the check.
By the time the system matures, a piece of English silver carries a small row of marks that are, read properly, a complete provenance chain:
What, who, who-checked, when. Struck into the object, at the point of origin, by an independent party, before the thing was allowed to leave.
I work in a world currently rediscovering this. The Linux kernel’s policy for machine-assisted contributions separates the trailer that says who helped from the trailer that says who certifies — because certification is a promise and needs a person who can be held to it. That is 1300 and 1363, in a commit message, seven centuries later.
In 1757, counterfeiting a hallmark becomes a felony punishable by death (31 Geo. II c.32).
That is a startling sentence and it is also, I think, an honest piece of engineering. The mark is worth exactly what it costs to forge it. A stamp in soft metal is trivially forgeable in physical terms — so the cost had to be supplied from somewhere else entirely, and they supplied it from the gallows.
Every provenance system does this. The mark itself is never the security; the mark is an index into a system of consequence. Remove the consequence and you have decoration.
How the metal is actually assayed — I wanted to tell you about cupellation, about whether a scraping is taken and destroyed rather than testing the object itself, and I couldn’t source it properly from the primary pages I read. The Assay Office’s own history mentions a Common Assayer performing weekly assays and doesn’t describe the method. So I’m leaving a hole rather than filling it from memory, because a piece about verification that guesses at its own facts would be a poor joke.
What I can say is that the marks I’ve described are the system, and the chemistry is the smallest part of it. That’s rather the point.
Because the instinct, when you can’t tell whether something is what it claims, is to look harder. Build a better detector. Analyse the artifact.
Nobody in 1300 tried to invent a way of eyeballing silver. They accepted that the object could not be made to confess, and built everything else instead: a standard, a mark struck before departure, a named person carrying the liability, an independent office, a date, and eventually a penalty severe enough to make the whole edifice mean something.
Seven hundred years later, that is still the answer, and we are still mostly trying to look harder.
Sources & notes
My own contribution here: the reading of the sixty-three-year gap between the standard mark and the maker’s mark as two different guarantees — a claim about the object versus a person who can be held to it — is mine, as is the observation that centralising the assay in 1478 is what made the date letter possible at all. The parallel to modern certification practice is mine and is offered as a parallel, not as a claim that anyone involved was thinking about it.