← Longevity & Supplement Guides
How Heart Rate Variability Went From Noise to Your Morning Number
In 1963, the American Journal of Obstetrics and Gynecology printed two papers by Edward Hon and S. T. Lee one after the other: pages 804 to 813, then 814 to 826. The first was about the electrocardiogram of the dying fetus. The second was about the heart-rate patterns that came before fetal death. Hon was a Yale obstetrician who had trained as a radio engineer before he trained as a doctor. He had built one of the first systems that could record a baby’s heartbeat continuously during labour.
Later reviews of this work all point to one observation. Before the fetus showed distress, the small changes in time from one beat to the next altered first. The average heart rate had not yet moved in any useful way. The trouble showed up in the gaps between beats before it showed up in the count.
For more than two hundred years, that small wobble had been a footnote. Physiologists saw it, measured it, and gave it a name that made it sound like a fault: sinus arrhythmia. Today the same wobble is a number that many people check before breakfast. This is the story of how noise became a signal. It is also the story of what got lost on the way from a hospital tape to your wrist.
A horse, a dog and a footnote
The first record is older than the stethoscope. In 1733 the English clergyman and experimenter Stephen Hales published Haemastaticks, his work on blood pressure in animals. A historical review by George Billman credits Hales as the first to report that the time between beats and the pressure in the arteries rose and fell with breathing. His subject was a conscious horse.
In 1847 the German physiologist Carl Ludwig built the kymograph, a smoked drum that recorded a pressure trace on paper as it turned. In dogs, he recorded that the pulse sped up as the animal breathed in and slowed as it breathed out. Billman calls this the first documented record of respiratory sinus arrhythmia. It is the same rhythm that a modern sensor detects every night while you sleep.
Then, for about a century, very little happened. The finding was not wrong. It was simply not interesting. Breathing changed the pulse a little. A healthy heart did not keep perfect time. Physiologists saw this as a fact about the machinery, much like a car engine that hums at a slightly different pitch on a hill.
Why the wobble looked like noise
Medicine had good reasons to ignore it. For most of history, the pulse was a rate: beats counted against a watch for fifteen seconds, then multiplied by four. A rate is an average, and an average is designed to hide the differences between the things it averages. Sixty beats a minute can mean sixty identical gaps of one second each. It can also mean gaps that stretch and shrink by a tenth of a second with every breath. Your fingers on a wrist cannot tell the difference.
Willem Einthoven’s string-galvanometer electrocardiogram, developed around the turn of the twentieth century, made each beat visible as a sharp spike, the peak that Einthoven labelled R. For the first time, the gap between two beats could be measured with a ruler instead of guessed. But early ECG machines were large, fixed in place, and used for short recordings of a patient lying still. A few seconds of paper strip shows a heart at rest in a clinic. It cannot show a heart living through a day.
Two things had to happen before the wobble could become a measurement. Someone had to record the heart for hours, outside the clinic. And someone had to show that the wobble changed when something important was happening.
A physicist with an 85-pound backpack
The first of those came from Helena, Montana, and from a man who was not a doctor. Norman Jefferis Holter held master’s degrees in physics from UCLA and in chemistry from the University of Southern California. According to the cardiologist Harold Kennedy’s history of the technology, Holter started out working on ways to stimulate nerves without direct contact. Then a visit from the Boston cardiologist Paul Dudley White convinced him that the heart was the more important problem.
Holter returned to Helena in 1947. In the late 1940s he and his collaborator Joseph Gengerelli made the first broadcast of an electrocardiogram, sent by radio from a backpack transmitter that weighed about 85 pounds. The subject wore it while exercising. It was an absurd object, and it proved the point: the heart’s electrical signal could leave the room with the person.
The transistor made the backpack shrink. In 1961 Holter described the result in Science: a portable tape recorder that could capture a continuous ECG from an active person over long periods, with a playback method designed to let a clinician review the long tape quickly. He gave the invention to Bruce Del Mar, whose company developed the first clinical system. The ‘Holter monitor’ is still the name for this kind of recording today.
Holter’s 1961 paper is about recording the hearts of active people, not about variability. But his recorder created the raw material that variability research needed: tens of thousands of beat-to-beat intervals from one ordinary day.
The obstetrician who listened between the beats
Edward Hon’s life looks like it was arranged for this problem. According to a biographical profile on Neonatology on the Web, he was born in 1917 in a village near Guangzhou to an Australian-Chinese family, and he grew up in Tenterfield, Australia. In 1942 he earned a diploma from the Marconi School of Wireless in Sydney. He then studied medicine in California and went to Yale for his residency in obstetrics. There, he proposed an electronics laboratory, and Yale funded it through a Markle Scholar award.
In 1958 he described a system that recorded the fetal ECG continuously during labour. The 1963 papers with Lee followed. Their observation, as the 1996 international standards and Billman’s review both report it, was that changes in the intervals between beats came before any clear change in the heart rate itself. The record does not tell us whether Hon saw this as a general principle of physiology or only as a practical warning sign in the labour ward. His published work was practical. He classified heart-rate patterns during contractions, and that classification still underlies fetal monitoring today.
His work, together with that of Roberto Caldeyro-Barcia and Konrad Hammacher, led to the first commercial fetal monitor in 1968. Continuous monitoring then became routine in many hospitals. Later trials questioned how much that routine use improved outcomes. A good signal does not always lead to a good decision about what to do next. HRV keeps relearning that lesson.
From footnote to morning number
Two and a half centuries of the beat-to-beat interval
From a bedside test to a power spectrum
Through the 1970s, the wobble began to work in clinics. David Ewing and his colleagues in Edinburgh developed simple bedside tests for people with diabetes. A patient breathed deeply, stood up, or strained against a closed airway, and the clinician measured how much the beat-to-beat interval changed. A small change suggested that the nerves controlling the heart were not responding normally. The tests were cheap and quick, and they changed a curiosity into a clinical tool.
In 1978 an Australian group led by M. M. Wolf looked at 176 patients admitted to a coronary care unit. Seventy-three of them showed sinus arrhythmia, measured from the variation in their beat-to-beat intervals on admission. That group had fewer deaths in hospital. It was a small study. It was also the first time that the old footnote had been tied to an outcome that clinicians could not ignore.
Then the engineers arrived. In 1981 Solange Akselrod and colleagues at MIT published a short paper in Science that applied power spectrum analysis to heart-rate fluctuations. A power spectrum takes a messy signal and sorts it into slow, middle and fast rhythms, much as a prism splits light. Fast fluctuations, which follow breathing, mostly reflect the vagus nerve, the main ‘brake’ of the parasympathetic system. Slower fluctuations reflect a mix of influences. For the first time, the wobble looked like a reading from the nervous system itself.
That last sentence is the one that later needed the most correction. We will come back to it.
The number that made it famous
In 1987 Robert Kleiger, J. Philip Miller, J. Thomas Bigger and Arthur Moss published the paper that made heart rate variability a famous term in cardiology. They studied about 800 survivors of acute heart attacks. Each had a 24-hour Holter recording. From it, the researchers calculated the standard deviation of all normal beat-to-beat intervals over the day. That measure is now called SDNN.
As later reviews summarise it, patients whose SDNN was below 50 milliseconds had about 5.3 times the mortality, over an average follow-up of around 31 months, of those above 100 milliseconds. The finding was strong, it was easy to state, and it turned Holter’s tapes into a new kind of data.
Look at what that number was. It came from a whole day of recording, not a five-minute reading. It came from people who had just had a heart attack, not from healthy people checking an app. And it was an association: low variability marked a group at higher risk. The study did not test whether raising someone’s SDNN would change anything. In their 2017 review, Fred Shaffer and J. P. Ginsberg ask that exact question about training people into a higher SDNN category, and they leave it open. It is still open.
1996: the rulebook
By the early 1990s, laboratories were measuring “HRV” in many incompatible ways. In 1996 the European Society of Cardiology and the North American Society of Pacing and Electrophysiology published a joint Task Force report in Circulation: standards of measurement, physiological interpretation and clinical use. It defined the measures, set the frequency bands, and described two kinds of recording: short ones of around five minutes, and long ones of 24 hours.
The rulebook did its job. Results from different laboratories could finally be compared. But a standard always does two things at once. It makes a measurement portable, and it makes the version that it describes look like the only version. The Task Force wrote its rules mostly for clinical recordings on ECG machines. Thirty years later, those rules still shape the HRV numbers that consumer devices report.
What got simplified away
The most important correction came from the physiologists. Billman’s 2011 review sets out the problems plainly. HRV is an indirect measure. It reflects the nervous system’s effect on the heart’s pacemaker, but it does not measure nerve activity directly. The idea that the slower ‘low-frequency’ band measures the sympathetic ‘accelerator’ has not held up, and the popular LF/HF ‘balance’ ratio is now strongly disputed. Even the stretch of the heart’s own tissue during breathing can change the intervals without any nerve input.
Breathing is the largest variable in the room. Slower, deeper breaths increase the breathing rhythm in the heartbeat, sometimes by a lot. Billman notes that many human studies did not control how people breathed during their recordings. Shaffer and Ginsberg add that changing breathing rate inside the normal range can move one common HRV measure without changing average heart rate at all.
Recording length matters as much. Shaffer and Ginsberg state that 24-hour, short-term and ultra-short-term values are not interchangeable, even when the same formula is used. A long recording includes a whole day of work, meals, stairs and sleep. A short one captures one quiet moment. Short resting values correlate poorly with 24-hour values. So the number that made HRV famous, a day-long SDNN in heart-attack survivors, is not the number that most people now see.
How it reached your wrist
The number on most wearables is a cousin of Kleiger’s, not a copy. It is usually RMSSD: the root mean square of the successive differences between beats. RMSSD looks only at how much each gap differs from the one before it. That makes it sensitive to the fast, breath-linked changes that mostly reflect the vagus nerve. It is also less sensitive to the slow drifts that dominate a 24-hour SDNN.
A worked example
Same heart rate, different heart: what RMSSD actually counts
Sport brought RMSSD into daily life. In 2013 Daniel Plews, Paul Laursen and colleagues wrote about elite endurance athletes in Sports Medicine. Their group used the log of RMSSD, measured in the morning, to follow how athletes adapted to training. Across that paper and Plews’s later thesis, the practical finding was modest and useful: weekly averages told them much more than any single morning. A few valid readings a week were enough to see the trend, and one low day by itself meant little. We have written about how that idea grew into HRV-guided training.
The last step was the sensor. Most wrist devices do not read the heart’s electrical R wave. They read pulse waves optically, through the skin, and estimate the intervals from those waves. That works best when you are still, which is why most wearables report HRV from the night. Each step from 1963 made the measurement easier to get. Each step also moved it a little further from the conditions in which its meaning was first established.
What the story is actually evidence for
Read honestly, the history supports three claims and refuses a fourth.
First, the wobble is real physiology, not instrument noise. Hales saw it in a horse and Ludwig saw it in a dog. A modern sensor sees it in you. The fast part of it mostly reflects the vagal brake on your heart, and it follows your breathing.
Second, the wobble can change before the average does. That was Hon’s observation in the labour ward, and it is why your HRV can drop after a short night, a late heavy meal or a few drinks while your resting heart rate barely moves. The signal is sensitive. It is not specific. A low reading tells you that something changed, not what changed.
Third, the meaning depends on the method. Twenty-four hours, five minutes and one night are different measurements. Your number is best compared with your own number, taken the same way, at the same time of day. Age lowers HRV, and healthy people differ widely from one another. A friend’s higher score says almost nothing about you.
The claim the history does not support is the one that is easiest to sell: that a higher HRV is a goal in itself, and that raising it will make you live longer. The famous mortality finding came from heart-attack survivors, and it was an association. No one has shown that pushing a healthy person’s HRV upward changes their long-term health. Be careful with any product or plan that promises to ‘boost’ the number. HRV is a thermometer for your recovery, not a thermostat.
That is the approach behind the Agen Band. Its overnight HRV is a derived value, calculated from an optical sensor, and it is shown as a trend against your own baseline. It is a wellness signal to notice, not a medical measurement and not a diagnosis. Our guide to what readiness scores measure goes into the details. If you have a heart condition, or a reading that worries you, talk to your doctor. A wrist estimate is not a clinical test.
The bottom line
For two hundred years, the small variation between heartbeats was a footnote with the wrong name. It took an 85-pound backpack, an obstetrician who had trained as a radio engineer, and a few spectral analyses to show that the gaps carried information that the average did not. That history gives one rule: a number is only as good as the question it answers. This week, ignore your single-day HRV. Look at a seven-day average instead. When a low day appears, write down what happened the day before: sleep, alcohol, a hard session, travel, a late meal. Within a month you will learn more about your own pattern than a population chart can tell you. As with the 220-minus-age formula and the 10,000 steps goal, the useful number is the one you compare with yourself.
