No Case to Answer
Twenty dead, and no law broken
Compliance establishes what must be true — and, just as importantly, produces the evidence that it is.
Which means its reach is set by something outside itself: it can only ever examine what somebody has already required. A clean report describes the rules as much as it describes you.
Take any decision that matters and follow it along its chain:
- Governancewho decides
- Riskwhat could go wrong, and what it would cost
- Compliancewhat must be true
- Resiliencewhat all three amount to when the thing actually happens
So work a real one through. This example is the hardest of the four, because the third link did not fail. There was nothing there to fail.
The Green Market, Bradford, 1858
William Hardaker kept a stall in the Green Market in the middle of Bradford and was known to everybody as Humbug Billy. On the last Saturday of October 1858 he was selling peppermint lozenges, bought as usual from a wholesale confectioner in Stone Street.
Sugar was expensive — it would not grow in Britain and had to be shipped in — so confectioners cut it with something cheaper. The trade called it daft: plaster of Paris, powdered limestone, sulphate of lime. A white powder that did nothing, cost almost nothing, and weighed the same as sugar.
There was nothing furtive about this. It was ordinary practice, openly conducted, and perfectly lawful.
The daft came from a druggist. The druggist was unwell that day and his assistant served the order, drawing twelve pounds from a barrel in the store. The barrel was not labelled. Beside the daft, and looking exactly like it, was arsenic trioxide.
Around twenty people died. More than two hundred fell seriously ill. Many of them were children, because the sweets were sold by the ounce to whoever had a penny.
Three men in the dock, and none of them guilty
The confectioner, the druggist and the assistant were all arrested, and the case against all three collapsed. There was no case to answer. It was, the court found, an accident.
And on the law as it stood, that was correct. Consider what would have had to exist for any of it to be a crime:
- A requirement that food contain what it is said to contain. There was none.
- A requirement that a poison be labelled. There was none.
- A requirement that a poison be kept apart from things that are eaten. There was none.
- A requirement that the person dispensing it be qualified. There was none.
Every one of those is now so obvious that it is difficult to read the list without assuming somebody had been negligent. Nobody had. They had all complied, completely, with everything that applied to them.
Compliance asks whether what must be true is true.
It has nothing at all to say about what nobody required.
The distance between a sweet and a poison was a label
It is worth sitting with the specific control that was missing, because it is the least impressive control imaginable. Not an inspection regime. Not a testing laboratory. Not a licensing scheme.
A label on a barrel.
That is the whole distance between a lozenge and a fatal dose: a word written on the side of a cask, by someone who had no reason to write it and no obligation to. The controls that turn out to matter are almost never the expensive ones. They are the ones so cheap and so dull that nobody can be bothered to argue for them, which is also why nobody writes them down.
And then the law arrived, and did not work
Bradford did what disasters do, and produced legislation. The Adulteration of Food or Drink Act followed in 1860.
It was widely criticised almost at once: the drafting was ambiguous and the penalties were too small to deter anyone. Adulteration carried on. It took the Pharmacy Act of 1868, and further legislation after that, before the practice was meaningfully constrained.
So there is a second lesson underneath the first, and it is the one that matters most to anyone who works with a rulebook. Between 1860 and 1868 a compliance question about a Bradford confectioner would have had an answer. There was an Act. There was a duty. A report could be produced, and it could be accurate, and it could say compliant.
A rule that exists but does not bite is indistinguishable from a rule that works, from the inside. The only thing that tells them apart is what happens next, which is the one thing a compliance report never contains.
The keep
- Compliance measures the coverage of the rulebook, not the coverage of the hazard. A perfect score is a fact about the rules.
- “Nobody broke the law” is compatible with twenty deaths. It is a statement about the law.
- Normal and lawful are not the same as safe. Adulteration was both normal and lawful, and it was also the mechanism.
- The controls that turn out to matter are usually too dull to argue for — which is exactly why nobody wrote them down.
- A rule that does not bite looks identical to a rule that does, until the day it is tested.
First move: open your most recent compliance report and look for what it does not mention. A failed control appears as a finding; a hazard that nobody wrote a rule about does not appear at all. Pick one thing the organisation genuinely depends on, and check whether a single obligation in that report actually touches it. The question to hold is not are we compliant — it is what are we not required to tell you?
Compliance can only test what somebody wrote down.
Where nothing is written, everything passes.
Sources
The stall, the nickname, the wholesale confectioner, the composition of “daft”, the unlabelled barrel and the twelve pounds of arsenic trioxide drawn from it: 1858 Bradford sweets poisoning and Historic UK, Dying for a Humbug. The arrests, the collapse of the case and the finding of accident; the Adulteration of Food or Drink Act 1860, the criticism of its drafting and penalties, and the Pharmacy Act 1868: the same two sources. Casualty figures are given as “around twenty” and “more than two hundred” because the accounts differ slightly and the difference is not material to the point. Checked 20 September 2026.