Qi:
what can actually be measured

Two families of words. The first names things in the body. The second names the tools we use to decide whether a claim about them is true — and those are the ones most readers are missing. A person who cannot tell an effect from a mechanism cannot follow this argument, in any language.

Start with the character, because it settles more of the argument than any laboratory has. 氣 is built from 气 — a picture of vapour drifting — over 米, grain. Steam coming off cooking rice. That is the root image: not a force, not a field. Something you can see rising off hot food on a cold morning.

And it never really left. If you learn Mandarin today, qi is one of the first words you meet, and it is entirely ordinary. The weather is 天氣 — sky-qi. Air is 空氣 — empty-qi. To get angry is 生氣 — to grow qi. A balloon is 氣球, a qi-ball. None of this is mystical to a Chinese speaker any more than "spirit" is mystical in "team spirit."

The classical uses, which are not one idea

By the Warring States period the word is doing philosophical work, but not consistently. Mencius, in the fourth century BCE, writes of 浩然之氣 — a "flood-like qi" that a person accumulates by acting rightly over and over. That is a moral quality, closer to what English means by integrity or nerve than to anything electrical. Zhuangzi, in the same era, uses it cosmologically: a human life is qi gathering, and death is the same qi dispersing again — no more remarkable, he says, than the turn of the seasons.

A thousand years later the Neo-Confucians made it systematic. For Zhu Xi in the twelfth century, everything that exists is 理 li, its pattern or principle, together with 氣 qi, the stuff-and-activity that the pattern is expressed in. Qi there is nearer to "matter in motion" than to "energy" — it is what there is, rather than a quantity that flows through what there is.

What the translation did

"Energy" is a hard-edged word in English. Since the middle of the nineteenth century it has meant a conserved scalar quantity measured in joules, and the whole reason it is useful is that you can count it and find it has not changed. When qi is rendered as "energy," that promise comes along for free — and the moment a reader accepts it, they expect a meter.

Everything below is a working instrument, not an illustration. The numbers move because they are computed, and where a figure is showing a published result the citation is on the panel. Where something is not established, the instrument says so in the same voice it uses for the things that are — which is the habit worth taking away.

Four things that are true

Start with the strongest, which is also the least exotic. In the Cochrane review of exercise for preventing falls — 108 trials and more than 23,000 people — tai chi reduced the number of older people who fell by about a fifth, and that particular outcome was rated high-certainty. Read the neighbouring row honestly too: the reduction in the overall rate of falls was rated low-certainty, with an interval running to 0.99. And ordinary balance and functional exercise did better than tai chi in the same review.

Second, the needle really does something mechanical. Sixty people, a robotic needling instrument, subjects blinded: rotating the needle one way raised the force needed to pull it out by 167%. That is not subtle and it is not disputed. But the same paper reports that needling at acupuncture points beat control points by only 18% — so the twist matters about nine times more than the location — and its author states plainly that the effect "is not unique to acupuncture points."

Third, the chemistry. In mice, thirty minutes of needling at a hind-leg point raised extracellular adenosine roughly twenty-four-fold, and the pain relief disappeared entirely in animals bred without the adenosine A1 receptor. That is a proper mechanism, of the kind this field is usually accused of never producing. It is also a mouse, there was no sham needling arm to test whether the point mattered, the human follow-up managed only a 1.8-fold rise while measuring no pain outcome at all, and nobody outside that laboratory has reproduced the chain.

Fourth, the breathing, which is bench two. There is nothing controversial about it: the pressure-regulating loop between heart, vessels and brainstem has a resonance near a tenth of a cycle per second, and breathing at that rate — about six breaths a minute — produces the largest heart-rate swing the system can make. Qigong, tai chi, pranayama and the recited rosary all land in that neighbourhood without having consulted each other.

And the acupuncture number everyone quotes

The largest analysis pooled the original records of individual patients from 29 trials and 17,922 people. Real acupuncture beat sham acupuncture — genuinely, statistically. It beat it by 0.23 standard deviations for back and neck pain, 0.16 for osteoarthritis and 0.15 for chronic headache. Against no acupuncture the same comparisons were 0.55, 0.57 and 0.42.

Those two rows are the whole argument in six numbers. The treatment works. Most of what it delivers is also delivered by fake needles at wrong places — which means the needling specifics contribute the smaller share, and the consultation, the attention, the ritual and the expectation contribute the larger one.

Set the whole case out in one ledger and the argument stops being about mysticism versus science. It becomes about where on the ladder the claim stopped climbing — and the answer is not where most people expect.

Read the column downwards, because the order is the finding. Everything at the bottom of the ladder — the unit, the instrument, the mechanism — is a No. Everything near the top, where a treatment either helps a person or does not, is a Yes. That is upside down compared with how most claims fail, and it is why qi is worth spending a whole page on.

Benefit without mechanism, and what to do about it

Something works, and the explanation attached to it is wrong or unverifiable. This is an ordinary situation in medicine rather than a scandal: aspirin was in use for about seventy years before anyone could say why it worked, and it did not stop working while we waited. The correct response is to keep the practice, keep testing it, and stop defending the explanation.

What makes it awkward is that the practice and the explanation arrive as a single package, taught together for two thousand years. Pulling them apart is unfamiliar work. It is much easier to accept both or reject both, and both of those shortcuts get the answer wrong — one by keeping a mechanism the evidence never supported, the other by throwing away balance training that demonstrably keeps people off the floor.

You will not remember the adenosine study. You might remember these.

1 · What is the unit?

Not "is it real" — what would you write on the axis. Millimetres of mercury is a unit. Beats per minute is a unit. "Vibrational frequency" as used in a shop is not a unit, even though frequency genuinely is one, because nobody will tell you of what. If the answer is a shrug, everything above that point on the ladder is decoration.

2 · Compared to what?

The single most useful question in this whole subject, and the one bench three was built around. Acupuncture beats nothing by a lot and beats sham by a little, and those two sentences describe the same treatment. A study with no comparison group is a story about people who got better, which is a story you can tell about almost any group of people who waited.

3 · How big?

"Statistically significant" answers a different question than the one you are asking. With enough participants a difference far too small to feel becomes certain. Insist on the size, in units you understand — a fifth of a standard deviation, a fifth fewer falls — and a lot of confident headlines quietly deflate.

4 · What would prove it wrong?

The sharpest of the four. "Qi flows through meridians" cannot answer it as usually stated. "A practitioner can tell which hand is near theirs, better than chance" answers it completely — and so it could be tested, and it was, and it failed. Notice that the second version is the friendlier one: it gave the claim a way to win.

Qi was chosen for this page because it fails in the direction people do not expect. The usual worry is an explanation with nothing under it, and that worry is correct here — but the practice underneath the explanation kept passing its tests anyway. Once you have seen a claim come apart into a part that survives and a part that does not, you stop grading things with a single mark.