Resonant breathing · 0.1 Hz · five-second lag

Breathe slow enough that blood pressure has time to fall.

JM Thomas · 17 August 2026

Blood pressure answers a fast inhale about five seconds late. Six breaths a minute is the tempo that meets that lag, so heart rate and pressure start drawing the same sine.

You already bought the calm-breath app. Then you sat there tapping 4-7-8 and box and two extra sips. The pulse never got a full five seconds to fall.

The delay is five seconds

Shaffer and Meehan wrote the lag in 2020. The vascular tree is slow. Volume takes time to move. Five seconds late. Vaschillo treated that delay as the reason a human adult rings near 0.1 Hz, which is six cycles a minute.

Picture a swing. Push when the seat comes back. Push on the way out and you fight yourself. Ten seconds a cycle. Inhale four. Exhale six. That 4:6 split is what Suminska used at home. The longer out-breath is the half of the loop that lets pressure drop.

Lehrer and Gevirtz mapped the phase in 2014. Near 0.1 Hz, heart rate and breathing go together. Zero degrees. Inhale, heart speeds. Exhale, heart slows. Respiratory sinus arrhythmia, stretched until it matches the baroreflex. Heart-rate swings at that tempo can grow four to ten times the resting range. That is the ring. A mood app does not name it.

Resonance is a tempo

Lehrer, Vaschillo, and Vaschillo wrote the training manual in 2000. The job is a bigger RSA wave, not a prettier thought. You breathe at the rate that makes the wave tall. Kids ring faster. Smaller trees of blood. Shorter delay.

Personal RF is a sweep. About 7.0 down to 4.5, half a breath at a time, two minutes each. Clinics still argue about whether that sweep earns the extra week. Steffen put 95 students in a single 15-minute sit. RF beat RF-plus-one on positive mood, and beat quiet sitting on the LF/HF ratio and on systolic pressure during a lab stressor. One session. One breath too fast already cost them. That is a photograph of the swing being pushed a little early.

A four-week file tells a quieter story. Six even. Start there if you have no clinic. The Polish trial used that as a whole arm.

A Polish RCT ran two pacers

Suminska, Rynkiewicz, and Szulczewski published in Scientific Reports in 2026. DOI 10.1038/s41598-026-53333-6. Eighty-eight randomized. Eighty finished, 51 women, mean age 30.86. High stress on the PSS-10. Three arms: personal RF, fixed 0.1 Hz, waitlist.

Four weekly lab sits, about 40 minutes. Twenty minutes at home daily with a visual 4:6 pacer. Mean 21.44 home sits, SD 6.22. RF got remeasured each lab week. The 0.1 Hz arm still sat through that sweep, then practiced at six anyway.

Session by group: stress F(2,77)=8.506, p less than .001, eta-squared 0.181. Anxiety F(2,77)=5.5, p=.006, eta-squared 0.125. Depression F(2,77)=3.336, p=.041, eta-squared 0.08. Both trained arms dropped. Waitlist stayed flat. Cohen's d inside the RF arm versus the 0.1 Hz arm: stress 1.28 vs 0.89, anxiety 0.83 vs 0.59, depression 1.04 vs 0.99. The between-group tests on those change scores did not split them. Bayesian BF01 sat in the anecdotal-null zone. Quote that. Do not shop a winner. You can sit in a chair with a cheap timer, inhale four, exhale six, and give blood pressure the five seconds it already wanted before you ever owned a clinic box.

Resting RMSSD went down at the final session in every arm, including waitlist. HF did the same. LF did not move. Four weeks of this drill did not raise the overnight number. The feeling scores moved. The resting trace did not. That is the lock I came to write. A calmer week can still print a flatter morning screenshot.

Innsbruck ran ten sessions

Canazei, Hufner, Sperner-Unterweger, Lampe, Weninger, Staggl, Dresen, and Weiss. General Psychiatry 2025, 38(6):e102357. PMID 41324027. Ninety-eight allocated. Twenty quit in week one. Seventy-eight finished, mean age 47.2, range 19 to 70, 67% women. Seventy-one percent somatic symptom disorder. Crossover. Fifteen minutes, twice a day, five weekdays. Equal inhale and exhale. Started mean day 10.8 after admission. DRKS00027323.

Sixty percent rang at six breaths a minute. Twenty-four percent at eight. Sixteen percent at 10 or 12. Their sweep went 12, 10, 8, 6. A lot of untrained chests still sat at six.

Primary outcomes were somatic well-being after the morning sit and presleep arousal after the evening sit. Primary missed. Body perception rose across the week in both arms. That is the ward doing ward work. Secondary: anxiety interaction eta-squared-p 0.09, 95% CI 0.01 to 0.22. Insomnia eta-squared-p 0.10, 95% CI 0.01 to 0.23. Acute HRV during the sit, eta-squared-p 0.23 to 0.51. No breathing-related adverse events. Eighty-three to ninety percent said they would reuse it.

Ten sessions on a ward is a different dose than four weeks at home. Keep the files apart. The steal from Innsbruck is the honesty on the primary. The chest can wait. The chest can still feel loud while anxiety and sleep scores ease, and that is still a real week if the 1-to-10 after twenty minutes dropped a point you actually wrote down.

The double sip lasts seconds

I already mapped the sip on the physiological sigh page. Li named the mouse circuit. Balban ran five minutes of cyclic sighing. This page is the long loop that meets the five-second lag. Use the sip when you need a brake in a hallway. Use six a minute when you have twenty minutes and a chair.

Do not stack them as one brand. The sip pops alveoli. Six a minute trains the baroreflex tempo. Write which sit you ran. The log hates a mash.

Sit twenty minutes at four-in, six-out

Nose in if you can. Soft belly. If you get lightheaded, you are shoving air. Back off. Write the number. Twenty minutes. Seven mornings. Yes or no in a notebook. A pretty pacer with no log is another app.

Skip the RF sweep in week one. Six even is the start the fixed arm used. If you later want the clinic version, sweep 7.0 down to 4.5 in half-breath steps, two minutes each, and keep the rate that made the pulse wave tallest. Most adults still land near six.

If twenty minutes still leaves the chest loud, sit until that 1-to-10 drops a point. Then go back to six. That first sit is Protocol 01. Free. The twenty minutes still needs a nerve quiet enough to hold a slow count.

Locks