Our positionThis is where we stand, and why. The reasoning is on this page.
The Rat Poison Question
Watch on YouTubeWhat to Say When They Think They Removed the GlutenThe question that fixes the model, and where the analogy honestly stops. AI-generated from this page and reviewed for accuracy.
The move
"It's only a little bit of gluten, it'll be fine."
"It's only a little bit of rat poison, it'll be fine."
Nobody finishes the second sentence.
That's the whole tool. You are not telling anyone they are wrong — you are handing them a sentence and letting them hear it in their own voice. It works because the objection was never really about quantity. It's about who gets to decide, and the swap makes the decision visible.
Two things it is not. It is not a claim that gluten is poisonous; where that breaks is below, and it matters, because the person across from you may know. And it is not a way to win an exchange. It is a way to replace the model somebody is working from, because their model is what will still be in place tomorrow when nobody is watching.
Two roads to "a little is fine"
People arrive at that conclusion by two different routes. Knowing which one you are facing tells you what to say next.
The tell is simple. If they quote you a number, it's the ceiling. If they describe something they did, it is the removal.
The ceiling — "isn't 20 ppm the standard?"
This is the one we have most often, and it does not come from a stranger at a dinner table. It comes from professionals.
Every time we start explaining gluten-free production to a new contract brewery, the same question arrives: why do you want no detectable gluten? Other people accept 20 ppm — why isn't that good enough? It's a fair question, asked in good faith, usually by someone who has read the regulation and is trying to do the job properly.
"20 ppm of gluten in the beer is fine."
"20 ppm of rat poison in the beer is fine."
That is where the light goes on. In our experience it is usually the end of the question, and the rest of the conversation becomes how to hit the spec rather than whether to.
What the swap exposes is not that 20 ppm is dangerous. It's that 20 ppm is a ceiling, not a target — the most that may be present before a beer loses the right to call itself gluten-free, not an amount anyone set out to include. A brewery aiming at the ceiling has quietly decided how much of somebody else's risk budget to spend. A brewery working from grain that never contained gluten never has to make that decision at all, which is the entire reason Truly Gluten-Free is written the way it is.
The removal — "I took it out, so it's gone"
The other road has nothing to do with numbers. It runs on a picture of contamination as a thing, sitting somewhere, which some action lifts back out.
A salad arrives with croutons on it. You send it back. It returns — same plate, croutons picked off in the kitchen. Whoever did that was not being careless. They were being helpful, working from a model in which the gluten is the crouton: remove the crouton, remove the problem. No amount of insisting fixes a model. But "would you pick the rat poison off and bring it back?" fixes it instantly, because nobody holds that model about poison.
A brewer says a rinse between batches is enough. A packager says the filler is fine because the gluten-containing run was yesterday. A taproom says a shared tap line is not a real risk. Same model, bigger vessel:
"A little cross-contact from the previous run is fine."
"A little cross-contact with rat poison from the previous run is fine."
Anyone who has validated a cleaning regime for an allergen knows why the second sentence is absurd. The first one deserves the same treatment. This is why shared equipment gets its own page, and why "we clean between batches" is an answer about intention rather than an answer about verification.
The error on this road is not "a little is fine." It is "I removed it, so it's gone." Nobody measured.
Where the analogy stops
This is a comprehension tool, not a scientific claim, and it is worth knowing exactly where it breaks — because the person you are talking to may know.
Gluten is not a poison. Unlike a true toxin it has a measurable tolerance threshold, the 20 ppm limit is built on real dose-response evidence, and anyone who answers "gluten isn't rat poison, there's a threshold" is right.
Saying so costs you nothing, because the analogy was never about dose. It is about two things a threshold does not settle:
- Removal doesn't work that way. A threshold tells you how much is tolerable. It tells you nothing about whether picking croutons off a salad, or rinsing a line, actually got the amount below it. Almost always nobody measured.
- It isn't their call. A threshold is a population statistic. The person eating the food is the one who carries the consequence, and consent is not something the kitchen gets to give on their behalf.
Used that way, the tool survives contact with someone informed — which the blunt version does not.
The 20 ppm limit rests on real dose-response evidence
The best evidence for the threshold is a double-blind, placebo-controlled trial that fed known daily doses to people with celiac disease: 10 mg/day produced no significant mucosal damage, while 50 mg/day did, over three months.1 A systematic review reached a consistent conclusion — under 10 mg/day is unlikely to cause significant histological change in most people, though individual sensitivity varies and some react below it.2
The regulatory limit exists because of work like this, not in spite of it. That is precisely why the ceiling question is about aiming at the limit, and never about the limit being wrong.
Say it as a question, not an accusation
The move only works if the other person gets to arrive at it themselves. "Would you say that about rat poison?" invites them to finish the thought. "You're trying to poison me" ends the conversation and confirms every stereotype about difficult customers — and, in the trade, gets you written off as the brewery that can't be reasoned with.
The goal is not to win. It is for the next salad, the next batch, or the next spec sheet to be handled by someone whose model has changed.
What the evidence supports
Each source below is named beside the claim it carries on this page; the footnotes hold the full citations.
- Catassi, C. et al. (2007) supports: The best evidence for the threshold is a double-blind, placebo-controlled trial that fed known daily doses to people with celiac disease: 10 mg/day produced no significant mucosal damage, while 50 mg/day did, over three months.1
- Akobeng, A.K. & Thomas, A.G. (2008) supports: The best evidence for the threshold is a double-blind, placebo-controlled trial that fed known daily doses to people with celiac disease: 10 mg/day produced no significant mucosal damage, while 50 mg/day did, over three months. A systematic review reached a consistent conclusion — under 10 mg/day is unlikely to cause significant histological change in most people, though individual sensitivity varies and some react below it.2
Related
- Truly Gluten-Free — the standard this site holds, and why
- Gluten-Reduced Beer Is Not Gluten-Free — the same argument where the industry makes it
- Shared Equipment and Cross-Contact — what actually removes contamination, as opposed to what feels like it should