TF4 Is Not a Point

Truth is the daughter of time, not of authority. Francis Bacon

For most of recorded commerce, a foot was a local fact. The foot of Amsterdam ran shorter than the foot of Rhineland; the bushel of one market town would not fill the bushel of the next. None of these measures were wrong. Each was internally consistent, carefully kept, and perfectly adequate for the town that maintained it. They failed only at the boundary, where two systems met and each assumed the other meant what it meant. Trade did not wait for a world government to impose the metre. It ran, for centuries, on conversion tables.

Acupuncture is at the conversion-table stage, and it does not yet know it.

The orbiter problem

In 1999 the Mars Climate Orbiter arrived at its destination and flew into the atmosphere instead of the intended orbit. The cause was not a bad number. One system produced impulse in pound-force-seconds, another consumed it as newton-seconds. Both units were valid, both internally consistent, and both were used correctly by the engineers who chose them. The spacecraft was lost because the number crossed a boundary without its unit attached, and arithmetic proceeded without complaint.

This is the failure mode that matters, and it is worth being exact about why. A code that cannot be resolved raises an error. Somebody sees the error. A code that resolves to the wrong thing raises nothing at all. It produces a clean, plausible, entirely incorrect result, and it does so at machine speed across an entire corpus.

So whose TF4?

AcuiQ maintains a catalogue of 773 acupuncture points and reads treatment protocols out of the published literature at scale. Both the catalogue and the literature label auricular points with a region prefix and a number. Both use the prefixes AH, AT, TF and HX. They do not agree on what the numbers mean.

The ISO and WFAS standards number auricular points anatomically, walking the ear region by region. Our catalogue numbers them alphabetically by English name within each region. Neither approach is wrong. They are simply different conventions, and they collide:

Read the first row again. Shenmen is the most frequently used auricular point in the entire clinical literature. In our catalogue its number belongs to a hypotensive point. Every paper that wrote TF4 and meant the calming point at the top of the triangular fossa was being read as prescribing something for blood pressure, and nothing anywhere reported a problem.

What the corpus says

In July 2026 we audited every point reference AcuiQ had extracted. The scale of the error is small, bounded, and precisely the shape you would predict:

Ninety-seven references out of 48,579. As an error rate this is trivial. As a class of error it is not, because the sixty dropped references announced themselves and the thirty-seven wrong ones did not. Had we only fixed what failed loudly, we would have finished the job convinced it was done, and left every one of the silent errors in place.

Two further findings came out of reading the source papers, and both matter more than the counts. Of the ten open-access auricular studies we pulled in full, exactly one stated which nomenclature it followed. And two of those ten departed from the standard they were otherwise using, numbering the antitragus in their own way, which means that naming your standard is necessary but not sufficient.

Every town had its own foot

The obvious conclusion is that acupuncture needs a global standard for point nomenclature. The obvious conclusion is wrong.

Acupuncture already has excellent standards. The World Health Organization published standard point locations in 2008. China maintains GB/T 12346 for channel points and GB/T 13734 for auricular points. The World Federation of Acupuncture-Moxibustion Societies publishes its own. ISO Technical Committee 249 works the same ground, through the SC 1 subcommittee it moved that work into in 2025, when it widened from Chinese medicine to traditional medicine generally. These documents are careful, they are internally rigorous, and they were produced by people who understood the problem far better than any single database vendor does.

Adding a further standard to that list would not reduce the confusion by one reference. It would add a sixth foot to a world that already has five, and it would deserve to be ignored. The gap is not in the standards. It is in the space between them, and in the citation habit that lets a code travel without the scheme that gives it meaning.

Two things that would fix it

The first costs one line and no committee. Prefix the namespace. Write WFAS:TF4 or GBT13734:TF4 rather than TF4. A journal can adopt this in a style guide tomorrow, a reviewer can ask for it in a single comment, and it converts a locally-scoped label into something a machine can resolve without guessing. Units travel with numbers in every other quantitative discipline. Point codes should travel with their scheme.

That is not a new idea, and it is worth knowing who had it first. The Lyon group agreed the ear region codes this article has been arguing about, and it agreed them prefixed: MA-TF, MA-AT, MA-AH, MA-HX, where MA stands for micro-system auricular point. The prefix was the namespace. The literature kept the two letters and the number and dropped the MA, which is exactly the character that would have told a reader whose scheme they were in.

The second is the conversion table, and we built one. It is published below and as a JSON file under CC BY 4.0. It maps ISO and WFAS auricular codes onto the AcuiQ catalogue, and every row carries the article it was read from and the Chinese name it was matched on. We built it the slow way on purpose: not by reconstructing a standard from memory, but by pulling the full text of each source paper and reading the author’s own pairing of name to code. The first summary we consulted claimed TF4 was the sympathetic point. It is not. That error would have propagated into a clinical dataset had we trusted it, which is the whole argument for provenance on every row.

Twenty-three mappings came out of that, twenty-one traced to a printed name-and-code pairing in an open-access paper, two inferred from the standard and marked as such. Where we could not resolve a code honestly we left it alone and said so.

* Holds only for the paper it was read from, which numbers the point differently from the standard it otherwise follows. Version 1.0, 23 rows.

One category turned out to be our own gap rather than the field’s. We had been dropping every reference to the scalp acupuncture lines, MS1 through MS14, on the grounds that no catalogue we held contained them. They are in fact standardised, and have been since 1991, when the revised WHO nomenclature fixed the 361 classical points in part 1 and the eight extra meridians, 48 extra points and 14 scalp lines in part 2. The lines were not missing from the literature. They were missing from us, and they have now been imported.

Which sharpens the point about the ear. That nomenclature is the closest thing to a unified acupuncture standard, and auricular points are the one category it does not contain. A WHO working group met in Lyon in November 1990 and adopted 39 ear points, listed a further thirty-odd as not yet considered, and closed by recommending that WHO convene a subcommittee to finish the job. No subcommittee came, and neither did ratification. Standardisation continued nationally instead, through GB/T 13734 in China and the WFAS standard in 2013, and disagreement between the Chinese and European schemes persists to this day. The fragmentation we hit was not evenly distributed across acupuncture. It sat precisely where the international process stopped.

The boundary is where the work is

The metre eventually won, but it won late, and it won after conversion tables had already done the practical work of holding international trade together for generations. Standardisation is the slow achievement of institutions. Interoperability is the immediate obligation of everyone operating between them.

Acupuncture is being read by machines now, ours among them, and the volume of that reading will only grow. Every protocol aggregated into a database, every systematic review counting point frequencies, every model trained on the clinical corpus inherits whatever ambiguity the citation habit leaves behind. The codes were never the problem. What is missing is the namespace that makes a code mean one thing, and the tables that carry meaning across the boundary between one careful system and another.

Corrections are welcome and will be credited. If a mapping here is wrong, we would rather know.