Right now one of your nostrils is probably carrying more air than the other. In a couple of hours it will likely have swapped. This is the nasal cycle, and it is the physiological fact that the whole of swara yoga is built on.

I keep citing it in passing. This article is the version with the numbers in it: what has actually been published, how big the studies were, and how much of the familiar description survives contact with the measurements.

The short answer is that the cycle is real and its mechanism is well understood, but the tidy alternating rhythm everyone draws is a minority finding.

What it is, and who found it

The turbinates — bony shelves inside each nasal passage — are covered in erectile tissue: venous sinusoids that fill and empty like any other erectile tissue in the body. When they fill, the airway narrows. When they drain, it opens. The two sides do not do this in step.

Richard Kayser, a German rhinologist, published the first systematic description in 1895, in a paper on measuring nasal airflow. Clinicians had noticed the phenomenon piecemeal for decades; Kayser measured it.

What he described has been confirmed many times since. What has not held up as well is the shape he and his successors gave it.

The periods people report

Here the literature is a mess, and it is worth seeing why.

The 10-hour axis is compressed after the 6-hour mark. Kahana-Zweig and colleagues' lower bound of 15 minutes was the floor of their analysis window, so the true minimum may be shorter still.

Pendolino and colleagues, reviewing the field for Rhinology Online in 2018, give the range as roughly 30 minutes to 6 hours, citing Gungor and colleagues from 1999.

The most careful single measurement I have found is Kahana-Zweig and colleagues at the Weizmann Institute, who logged airflow in both nostrils of 33 healthy adults for about 24 hours each and published in PLOS ONE in 2016. Their average dominance spell was 2.15 hours — but individual spells ran from 15 minutes to 10 hours 20 minutes. Spells during sleep averaged 4.5 hours, more than twice the waking figure.

So “two to three hours” is not wrong, exactly. It is an average being quoted as though it were a rule, in a distribution wide enough that the average describes rather few of the individual events inside it.

How many people actually have a regular cycle

This is the finding that changed how I write about swara yoga, and it is not obscure — it sits in the standard reviews.

The classical picture is periodicity and reciprocity: a steady rhythm, with the two sides as mirror images, one closing as the other opens. Early observational work reported something like it in up to 80 per cent of people.

Then the quantitative studies arrived.

Hanif, Jawad and Eccles, reviewing the literature for Clinical Otolaryngology in 2000, report that the more careful measurements put true periodicity and reciprocity at between 21 and 39 per cent. Pendolino and colleagues’ 2018 review reports the same band. Gilbert and Rosenwasser had already put it bluntly in 1987: an ideal cycle is barely found.

Kahana-Zweig’s recording study agrees from a different direction. All 33 of their subjects switched dominant side at least once. But only 21 showed a clear rhythm in even one nostril, and only 5 showed one in both. Sometimes congestion rose on both sides at once rather than trading off. The authors concluded that each nostril may oscillate somewhat independently — which, if right, means “the nasal cycle” is a slightly misleading name for the thing.

The neat diagram on the left is the one in almost every article, including some of mine. It describes a minority of people.

The autonomic control is the solid part

If the rhythm is shakier than advertised, the mechanism is firmer. The nasal sinusoids are under autonomic control, and the human evidence for that is unusually direct.

Sympathetic outflow releases noradrenaline, which constricts the sinusoids, drains the tissue and opens the airway. Cut or block that supply and the same side congests. Pendolino’s review catalogues the classic demonstrations: cervical sympathetic transection and stellate ganglion block both alter the cycle on the same side, work going back to Stoksted in the 1950s.

The cleanest modern demonstration comes from spinal cord injury. Gainche and colleagues, in the Journal of Clinical Sleep Medicine in 2016, compared 10 people with tetraplegia against 9 able-bodied controls. Nasal resistance in the tetraplegia group was more than twice as high — 7.0 against 3.0 cm H₂O/L/s. Then they applied a topical sympathomimetic, and resistance fell to 2.3, essentially the control level.

That is about as close to a causal demonstration as human physiology gets without an experiment nobody would approve. Remove the sympathetic signal and the nose congests. Supply the missing chemical and it opens.

Note the sample sizes, though: ten and nine.

Where the whole field is thin

Three things stand out once you start chasing the references, and they are the reason for this article’s subtitle.

The studies are small. The largest careful recording study I found is 33 people. The sympathetic demonstration is ten against nine. For a phenomenon present in every human being and trivially cheap to measure, that is a remarkably thin evidence base.

The studies are old. A great deal of what gets repeated traces back to work from the 1920s to the 1950s — Heetderks, Stoksted, Hasegawa and Kern — in an era with different instruments and different reporting standards. Those findings may well be right. They have mostly not been re-run.

The reviews often omit sample sizes entirely. When I went looking for how many people sat behind each prevalence figure, the reviews frequently did not say. A reader cannot weigh a number whose denominator is missing. Pendolino and colleagues are candid about the state of things: understanding of the nasal cycle, they write, is still very poor, and the reported frequencies for each pattern are often discordant.

What I take from this

The cycle is real. It is autonomically driven, centrally controlled, and readable from outside the body in about five seconds. Nothing here undermines that, and it is a genuinely useful fact.

What does not survive is the picture of a metronome. Your nostrils are not trading off on a reliable two-hour schedule. On the best available measurements, a clear reciprocal rhythm is a minority pattern, the spread of individual spells is about twentyfold, and sleep roughly doubles the period.

For anyone reading the traditional literature, that has a specific consequence. A system that assigns actions to the dominant nostril assumes dominance is a stable, well-behaved state variable. For a minority of people, it may be. For most, the variable is noisier than the system assumes.

That is worth knowing before you build anything on top of it — which, in my case, I already had.

If you want the other side of this, I have written about what the Śiva Svarodaya claims about timing and about which of its claims survive checking.