Sources and evidence

Most tea writing cites nothing. Some cites a book once and then carries the claim forward for years. TeaHQ records, for every claim that needs one, which document it rests on and how carefully that document was examined — and this page explains what those records mean, so the distinctions are legible rather than internal bookkeeping.

The register currently holds 143 sources. That is a small number, and it is honest: it reflects how much of the catalogue is descriptive rather than evidential, and how much still rests on our own synthesis.

Sources are not interchangeable, so they are recorded by kind.

• Standards and regulatory documents — the strongest kind for anything definitional. A standard defining a grading system settles what the grade means.
• Peer-reviewed research — for claims about mechanism or measurement, and the only class that can support a statement about what is established.
• Institutional and industry publications — research institutes, boards and trade bodies. Useful, and carrying an interest we record rather than ignore.
• Producer documentation — good on how a specific tea is made, weak as independent support for a general claim.
• Ethnographic and historical accounts — where a practice is documented but not tested. These support statements about what people do, not about what is true of the world.
• TeaHQ synthesis — our own account of widely-attested practice, attributed to us. It is a source in the sense that it is identified and accountable, not in the sense that it is independent.

Citing a document and having read the relevant part of it are different acts, and a great deal of bad referencing comes from treating them as the same. TeaHQ records them separately.

Citation identity means we have confirmed the document exists, that it is the one we think it is, and that the reference points to the right place. That is the minimum for any citation.

Passage verification means somebody retrieved the specific passage being relied on and read it. This is a higher bar and it is required whenever a claim asserts what a document says internally — as opposed to merely citing it for context.

Where a claim depends on what a document actually says, and nobody has read that passage, the claim is not published. A citation cannot be upgraded by confidence, and no amount of review substitutes for retrieval.

Every source on this site has been retrieved, recorded and assessed through the TeaHQ evidence process described above. That process establishes what a document is, where it came from, and how much of it was actually read.

No source currently carries a human countersignature. Retrieval and assessment are not the same act as a named person attesting to a document, and this site does not present them as though they were.

Countersignature is a future review layer rather than a claim already completed. The record format has carried a countersignStatus field since the register was built, and every source in it reads pending. When a named reviewer takes accountability for a source, that field changes and this page will say so.

The practical consequence for a reader: where a page rests on an external document, you can check which document and how thoroughly it was read, and you should weigh that as the work of a careful process rather than as a signature.

Where TeaHQ makes a claim about the world rather than describing a tea, the claim carries a tier saying how strongly it is supported.

• Established — supported by qualifying independent evidence. Reserved, and the tier with the strictest source requirements.
• Emerging or limited — some support exists and it is not settled. The page says what is and is not supported rather than rounding to a yes.
• Traditional use — a practice is long-attested in a named culture. This is a statement about the practice, not about its effect.
• Unsupported — a claim circulates widely and evidence does not support it. These are recorded so that the common belief can be addressed directly.

Traditional use and established evidence are different claims and are never merged. “This has been drunk for centuries in Fujian” is a historical statement. “This does something to your body” is a different one, needs different support, and the first is not evidence for the second.

The largest single class of external source on this site is a taxonomic one. One hundred and twenty-seven records from the USDA’s GRIN-Global taxonomy database sit behind 130 of the 150 botanical pages, each cited for one narrow thing: the accepted scientific name, its author citation, the plant family, the recorded common names and the native distribution.

Every one of those sources carries the same sentence bounding its own authority, and it is worth quoting because it is the most important sentence on this page:

A taxonomic and distribution authority. Carries no authority over preparation, flavour, cultural practice or human physiology, and may never be cited for any of them.

That restriction is enforced rather than intended. A taxonomy citation is scoped to the claim it supports rather than to the page it sits on, it may never act as sole support for a functional claim, and a test refuses any botanical record that cites one more broadly. The reason is that an authoritative citation drifts: a national authority named next to a sentence about brewing lends that sentence a weight it has not earned.

What this checking actually found is worth saying plainly, because it is the argument for doing it. Across seven passes it produced three wrong binomials, seven origin statements more precise than any source supports, six hybrids carrying a wild native range they cannot have, and repeated instances of an industry’s geography standing in for a plant’s. Jasmine was credited to Fujian, which scents it. Persimmon leaf to Japan, which drinks it. Sarsaparilla to the Caribbean, which bottled it. None of those was detectable from inside the catalogue and all of them read as confident writing.

Trade and cultural geography survives this checking rather than being corrected away. That chenpi is aged in Xinhui and rooibos grown in the Cederberg are facts about commerce and culture, and they are kept alongside the botanical range rather than replaced by it. The distinction is labelled, which is the whole of the discipline: botanical fact, trade knowledge and cultural attribution are three different things and the page says which is which.

Most of what you read here is description rather than evidence: what a tea is, where it grows, how it is made, what it tastes like, how to brew it. That material is our own account of widely-attested practice, and it is attributed to us rather than dressed in a citation it does not have.

Synthesis is allowed to describe a practice. It is not allowed to certify what some other document says, and it cannot on its own support a claim about health or physiology — which is one of the reasons that surface is unpublished.

Where TeaHQ says the evidence does not support a claim, that is a statement about the evidence rather than about the world. It means the studies do not show what the claim asserts — not that the opposite is true, and not that the question is closed.

The same applies to the gaps. No caffeine figure for a tea means nobody has measured or published one to a standard we can cite. It does not mean the tea has no caffeine, and it is not an invitation to infer one from the category.

You will see it as brewing parameters marked as synthesis rather than recorded practice; as oxidation described in words rather than percentages; as comparison cells that say “not recorded”; and as an entire health section that does not exist yet.

The editorial policy explains how those decisions are made, and the about page explains who makes them.