Method
An atlas is worth what its sources are worth. This page therefore states where the displayed data comes from, who arbitrates it when works contradict each other, and what keeps data from reaching the screen. Nothing below is an intention: every rule cited here is written into the project register, and a program checks that it is held.
Two works, or the field stays empty
A published statement rests on two independent works. Two references only count as two when they come from different works: a plate and its accompanying text form a single lineage, even across two pages.
When a single work carries the matter, the block says so rather than making up the number. And when none carries it, the field stays empty: that emptiness records a search that came back with nothing, which is information — not an oversight that slipped through.
Facts are taken up, never wording. The displayed text is rewritten from the observed fact; it does not copy the sentences of the work consulted — the matter belongs to the reader, the phrasing stays with its author.
Who decides when works disagree
Works are not equal everywhere, and none is authoritative on everything. Each role therefore has its own, named in advance; but an authority of role is not a quota of sources, and it never dispenses with the second work.
A disagreement is never averaged. The discarded position is kept together with the authority that carries it, and it reaches the screen only when it changes the gesture: averaging two locations would designate a place that no source states, and that no one has ever palpated.
- OMS 2008 Codes, names and counts of the 361 points
OMS · WHO Standard Acupuncture Point Locations in the Western Pacific Region (2008) - Focks Point locations and channel pathways — first authority
Focks · Atlas of Acupuncture - Maciocia Secondary pathways the first authority does not carry
Maciocia · The Channels of Acupuncture - Deadman Benchmark of concision for record text
Deadman, Al-Khafaji & Baker · A Manual of Acupuncture - Travell & Simons Myofascial loci and referred pain territories
Travell & Simons · The Trigger Point Manual
Two branches, never merged
The Chinese tradition and contemporary medicine are not read under the same regime, and confusing them distorts both. In the tradition, classical consensus prevails and recent work comments on it. In contemporary medicine it is the reverse: recent work corrects the older.
Anatomy and cautions belong to neither branch. They appear in both positions of the toggle and are always reasoned in the contemporary regime, without exception.
Where a point and a myofascial locus meet, the screen states the measured distance between them, and stops there. Calling them identical would be a conclusion; measuring them is a fact.
Uncertainty is drawn, not recounted
The position of a point varies from one work to another, and that variation has been measured rather than guessed. It lives in the shape of the mark: the wider the patch, the less constrained the position, and the eye ends up reading that scale without being told.
Three tiers are told apart this way — bone contact, nearby landmark, proportional construction. No radius and no tolerance is written into a record: what the shape already says, the text has no need to repeat.
Naming an uncertainty is not enough: it is then settled. Laying out a controversy would look like added honesty; mostly it would hand the reader a burden the product is meant to carry.
What this atlas does not do
No therapeutic indications. “For this disorder, these points” is out of scope by design — not while awaiting something better, but because it would be a different trade than that of a map.
No text written by a language model is displayed, and no image of a body, a point or a pathway is generated. All geometry is built by program from sourced coordinates: what appears on screen can therefore be rebuilt, and recounted.
A machine does not assign clinical validation. The highest status an automated process may write is “technically validated”, which says nothing about clinical soundness; beyond that, a person answers for it.
And nothing displayed here replaces the examination of the person present.
The gates fail closed
A control that rests on a writer's vigilance does not exist: it holds as long as attention holds, which is to say not very long. The project's controls are therefore programs. They fail closed, and they run on every code push.
Two entry refusals are not negotiable. A depth without the critical structure it crosses is refused, because a number alone warns of nothing. And a fact whose source key is absent from the register cannot be published, however plausible it looks.
This page goes through the same regime. A program re-reads its text before every build and refuses promises, superlatives and professional titles; it also checks that every rule stated here still points to the project text that carries it, so that rewritten doctrine turns the page summarising it red.
The source register
The register lists the works, articles and datasets consulted, each with its rights regime: what may be read and cross-checked, what may be cited, what may enter the product. A work can feed the whole undertaking without a single one of its sentences coming out.
No text from a copyrighted work is reproduced here. What comes out of the register are restated facts and references — enough to find the page again, never enough to spare anyone the book.
157 Entries in the register, each with its rights regime verified at its date