How to check an ingredient claim yourself

Almost everything written about ingredients asks you to trust the writer. This page asks you not to.

Every source behind this site is public, free and searchable, and the skill of reaching them is smaller than it looks. What follows is the order that gets an answer fastest, with the traps that cost the most time.

Start with the regulation, not with a search engine

If a claim is about whether something is permitted, at what level, or what a label must say, the answer is in a regulation and the regulation is online in full.

eCFR carries the current text of US federal regulations. Title 21 is food, drugs and cosmetics. Its search box works on plain words: type pyrithione zinc and you land on the section that names its permitted concentrations.

Two things about it are worth knowing before you start.

Guessing section numbers wastes hours. The numbering is not intuitive and the same substance can appear in several parts. Search for the substance name rather than reasoning about where it ought to live. This site got that wrong repeatedly before learning it.

A section’s URL path can be wrong and still work. eCFR will serve you the right section from a URL with the wrong subpart in it, so a link that loads is not proof that the citation is accurate.

The other thing a regulation gives you is the shape of the rule. When a monograph lists concentrations for an active, it is telling you the product must print that number. When a rule permits an ingredient only for a stated purpose, as the potassium bromate additive rule does for malting barley, a use outside that purpose is authorised somewhere else or not at all.

Then the Federal Register, for the reasoning

A regulation says what the rule is. It almost never says why, or what the agency considered, or who objected.

The Federal Register does. It is the daily journal where a rule is proposed, commented on, finalised, amended, withdrawn or revoked, and the preamble to a rule is often the single most informative document about an ingredient in existence.

Everything genuinely new on this site in the past month came from there rather than from a search engine:

• Brominated vegetable oil: revoked in 2024 because there is “no longer a reasonable certainty of no harm”, after studies FDA itself commissioned
• Talc: a proposed asbestos testing rule issued in December 2024 and withdrawn in November 2025, which is the kind of fact that simply does not surface any other way
• Bemotrizinol: added to the sunscreen monograph in June 2026
• Saccharin: the EPA rule that records the 1981 listing, the 2000 delisting, and the mechanism that explains both

The trap: a Federal Register document page renders its text from script, so copying from it can fail and the printed view is easier to read. Search by ingredient name, filter by agency, and read the SUMMARY and the section headed with the agency’s reasoning rather than the amendatory instructions.

Then the literature, when the question is about evidence

For anything that is not a rule, the question is what studies found. PubMed Central is a free full-text archive, and a great many papers behind paywalls elsewhere are readable there in full.

Read the abstract’s limitations sentence before anything else. It is usually the most honest line in the paper and it tells you what the finding cannot support.

Three questions settle most claims:

In what? A cell line, a rabbit’s ear, a rodent, or people. Effects in a dish routinely fail to appear in a person, and the comedogenic ratings still quoted in skincare came from rabbit ears.

At what dose? Against what a person is actually exposed to. The erythritol intervention used 30 grams against an estimated dietary intake of 0.023 grams a day, and that ratio is most of the argument.

Presence or cause? Finding a substance in diseased tissue establishes that it got there. The paraben story is the worked example and it is on Parabens.

The traps that cost the most time

A source that answers a machine differently from a person. Several regulators and journals refuse automated access while serving people normally. If a link fails, try it in an ordinary browser tab before concluding it is dead.

A link that loads but does not contain the quote. Test any claim by searching the source page for a distinctive phrase from it. If the phrase is not there, the citation is wrong even if the link works. This is the single most productive check there is, and it is how the errors on this site get found.

Reading a summary of a source instead of the source. Almost every wrong ingredient claim in circulation traces to a chain of people quoting each other. The document at the end of the chain is usually shorter and clearer than any of the retellings.

Confusing status with finding. Banned in Europe covers at least four different situations, only one of which is a regulator finding a problem. The sorting is on the page about ingredients banned somewhere else.

How to check this site

The same way, and we would rather you did.

Every page here lists its sources with a link. Pick a quoted passage, open the source, and search the page for it. If it is not there word for word, we have made a mistake and want to know at hello@pura.to.

That check is also run mechanically against every page before anything ships, and the current count, including what is failing and which sources we cannot read, is published on the verification page. Quotation marks here mean quoted verbatim from a source cited on the page; where a page uses its own phrasing it is set in italics.