Epsilon Waves: The Slowest Claimed Brain Rhythm

Epsilon waves are brain activity slower than 0.5 Hz, below delta, as meditation and brainwave-audio writing describes them. The clinical EEG glossary does not list them. It counts activity down to 0.1 Hz as delta and calls anything slower infraslow activity, which is real but needs special amplifiers to record. Harmonance plays no epsilon track; the player plays a delta session falling from 3.5 to 1.5 Hz.

Ready to play Falling Delta Session 3.5 to 1.5 Hz

Open the 3.5 to 1.5 Hz page

Nine blue ribbons of light in dark space, each slower than the one above, from quick ripples at the top to one long, gentle swell at the foot — Nine ribbons in the delta band's blue, each slower than the one above, down to one long swell. Harmonance plays no epsilon track; the player plays delta.

What are epsilon waves?

Epsilon waves are described, mostly in meditation and brainwave-audio writing, as brain activity slower than 0.5 Hz: less than one cycle every 2 seconds, below the delta band that dominates deep sleep. The name carries on down the Greek alphabet that already labels the EEG bands, alpha, beta, gamma, delta and theta, and writing that uses it usually ties it to very deep meditation or rest.

What the name does not have is a definition. The International Federation of Clinical Neurophysiology keeps the glossary that clinical EEG reports are written in, and its 2017 revision defines the frequency bands clinical EEG works with. It has no entry for epsilon, and the word does not appear in it anywhere (IFCN glossary). The activity the name points at is real, though. Science has other names for it, and a plain reason the ordinary EEG rarely shows it.

Are epsilon waves a real EEG band?

Not as a band. The glossary sets the bands like this: delta from 0.1 to under 4 Hz, theta from 4 to under 8 Hz, alpha from 8 to 13 Hz, and gamma from over 30 to 80 Hz (IFCN glossary). A later correction fixed beta at over 13 to 30 Hz (glossary correction). Anything slower than 0.1 Hz is "infraslow activity", and the glossary gives it one synonym: subdelta.

So by the glossary's own numbers, most of what epsilon is said to cover, 0.1 to 0.5 Hz, already belongs to delta, and the part below 0.1 Hz already has a name. The glossary adds a practical note to delta that matters more than any name: "for practical purposes lower frequency limit is 0.5 Hz, as DC potential differences are not monitored in conventional EEGs". In other words, 0.5 Hz is not where the brain changes. It is where the usual recording stops looking.

EEG bands on one log scale from 0.01 to 100 Hz: epsilon is a dashed outline below 0.5 Hz, in no glossary; delta and the session's beat are blue — The EEG bands as the clinical glossary defines them, on one scale of frequency. The range the name epsilon is given is a dashed outline, because no glossary defines it. Delta, and the beat the delta session plays as measured from its master, are in blue. The dashed line marks 0.5 Hz, where a routine EEG stops looking.

The numbers behind the chart
Activity	Range	From
Gamma band	over 30 to 80 Hz	IFCN glossary
Beta band	over 13 to 30 Hz	glossary correction
Alpha band	8 to 13 Hz	IFCN glossary
Theta band	4 to under 8 Hz	IFCN glossary
Delta band	0.1 to under 4 Hz	IFCN glossary
Routine EEG floor	0.5 Hz	IFCN glossary
Infraslow activity	below 0.1 Hz	IFCN glossary
Infraslow oscillations	0.02 to 0.2 Hz	DC EEG study
Epsilon, as the name is used	below 0.5 Hz	no glossary entry
Delta session, measured	3.4996 falling to 1.4993 Hz	Harmonance master

Why does a normal EEG miss activity below 0.5 Hz?

Because it is built to. A routine clinical EEG records roughly 0.5 to 50 Hz, a bandwidth that, in the words of one review, was "originally set by trivial technical limitations" and has stayed "practically unaltered for over 50 years" (Vanhatalo, Voipio and Kaila). The low edge comes from a filter. EEG amplifiers pass the signal through a low frequency filter, which the glossary describes as a circuit that "reduces the sensitivity of the EEG signal to relatively low frequencies (for example below 0.5 Hz)". It also gives the arithmetic: with a time constant of 0.3 s, a 0.5 Hz wave loses 30 per cent of its size, and anything slower loses more (IFCN glossary).

The diagram builds one illustrative signal from three waves, a slow 0.05 Hz swell, a 0.3 Hz wave in the range epsilon is said to cover, and a 2 Hz delta wave, and passes it through exactly that filter. The swell keeps 9 per cent of its size, the 0.3 Hz wave keeps 49 per cent, and the delta wave keeps 97 per cent. On a routine trace, the slowest activity has been filtered out before anyone looks.

A blue wave built from a slow swell, a 0.3 Hz wave and a 2 Hz wave, and below it the same signal after a routine EEG filter, its slow parts flattened — One illustrative signal, drawn in code. Above, as a full band amplifier records it; below, after a filter with a 0.3 s time constant, the setting the glossary ties to the 0.5 Hz floor. The bars show how much of each wave gets through.

The same review argues for "full-band" EEG, recorded with direct-current (DC) coupled amplifiers, because the brain's activity "ranges at least from 0.01 Hz to several hundred Hz" and a routine recording throws the slow end away (Vanhatalo, Voipio and Kaila).

What does the brain do below 0.5 Hz?

Quite a lot, and none of it is called epsilon. With direct-current EEG, Vanhatalo and colleagues recorded "large-scale infraslow oscillations" of 0.02 to 0.2 Hz across wide areas of the human cortex during sleep, "strongly synchronized with faster activities". These oscillations, they wrote, are "not detectable in conventional electroencephalography because of its limited recording bandwidth (typical lower limit 0.5 Hz)" (Vanhatalo and colleagues). Later work matched infra-slow scalp potentials of 0.01 to 0.1 Hz to the brain's resting-state networks as fMRI sees them (Hiltunen and colleagues).

Just above that, sleep has its own slow rhythm. The slow oscillation, under 1 Hz, was first described in anaesthetised cats, whose cortical neurons cycled most often at 0.3 to 0.4 Hz (Steriade, Nuñez and Amzica). In human sleep, all-night recordings of eight healthy young men, with the amplifier's low-frequency loss corrected, found it peaking at 0.7 to 0.8 Hz, within a range of 0.55 to 0.95 Hz, and found slow waves coming and going every 20 to 30 seconds (Achermann and Borbély). By the glossary, deep sleep itself is dominated by delta waves of 0.5 to 2 Hz.

So the slowest rhythms are real, measured and studied. They belong to sleep, to the waxing and waning of the brain's networks, and to the slow tides beneath faster activity. The research names them infraslow activity and the slow oscillation, and none of these studies calls them epsilon.

Do epsilon waves appear in meditation?

Not in the research so far. A systematic review gathered 56 EEG studies of mindfulness meditation and found the commonest change was more alpha and theta power than in eyes-closed rest, "although such outcomes were not uniformly reported". It found "no consistent patterns" in beta, delta or gamma, and its summary names no band below delta at all (Lomas, Ivtzan and Fu).

That does not show that nothing slow happens in a meditating brain. A recording filtered from about 0.5 Hz is simply not set up to see it. But the idea that epsilon marks a special state of meditation has, for now, no measurement behind it.

What would an epsilon binaural beat sound like?

A binaural beat plays one tone to each ear, a few hertz apart, and the brain hears the difference as a pulse or a movement. At low rates, the authors of one study write, the result is "a single sound, which is moving across the head between the left and right ears"; as the rate rises, that turns into a loudness fluctuation and then roughness (Ross and colleagues).

An epsilon beat would be slower than one cycle every 2 seconds. The arithmetic is simple, because the beat rate is the difference between the two tones. A 108 Hz tone in one ear against 108.5 Hz in the other makes one cycle every 2 seconds; against 108.1 Hz, one every 10 seconds; against 108.05 Hz, one every 20 seconds. By Ross's description of slow beats, a sound like that would drift slowly across the head rather than pulse.

The slowest beat Harmonance plays is the end of the delta session. Its master, measured in 6,481 windows across all nine hours, starts at 108.0006 Hz in the left ear against 111.5002 Hz in the right, a beat of 3.4996 Hz, and ends at 108.0000 against 109.4993 Hz, a beat of 1.4993 Hz. That is one cycle about every 0.67 seconds, three times faster than the 0.5 Hz line.

A beat below 0.5 Hz has been tested once, as far as a PubMed search shows. In a study in Scientific Reports, 12 healthy adults took four 90-minute afternoon naps: once with no sound, and once each with beats of 0, 0.25 and 1 Hz on a 250 Hz carrier. With the 0.25 Hz beat, the sleep stages known as N2 and N3 began sooner than with no sound. But the EEG showed "no significant results regarding neural entrainment" at 0.25 or 1 Hz: the brain did not measurably follow the beat. The study was supported by the electronics company KYOCERA, two of its authors worked there, and a patent application was filed on the results (Fan, Zhu and colleagues).

For faster beats the picture is mixed too. A review of 14 EEG studies of binaural beats found five in line with the idea that the brain falls into step with the beat, eight against it, and one mixed (Ingendoh, Posny and Heine). Timing seems to matter. In sleeping people, short sounds played in step with the brain's own slow oscillation strengthened that rhythm, while the same sounds played out of step "remained ineffective" (Ngo and colleagues). A steady beat, unlike those timed sounds, is not matched to the listener's own rhythm.

Is it the same as the epsilon wave on an ECG?

No. Search for epsilon waves and some results come from cardiology. On an electrocardiogram, the epsilon wave is a small signal "between the end of the QRS complex and the beginning of the T wave", seen in arrhythmogenic cardiomyopathy, which affects the heart muscle. Willis Hurst, writing in the journal Circulation, recorded why it has that name: Fontaine, who described it, "chose the ε because it follows delta (δ) in the Greek alphabet and is the mathematical symbol for smallness" (Pérez-Riera and colleagues).

It belongs to the heart, not the brain, and has nothing to do with meditation or sound. No source checked for this guide records where the brainwave name came from.

What plays here instead?

Harmonance plays no epsilon track and will not invent one: a sound sold on a claim like this is exactly what the site refuses to make. The player at the top plays the Falling Delta Session, two pure tones whose beat falls slowly from 3.5 to 1.5 Hz over nine hours, the slowest rhythm on the site. Use headphones, so each ear gets its own tone, and keep the volume low. Listen if you are curious, notice what you notice, and stop whenever you like.

The delta session page plays it on its own, the binaural beats collection holds the sessions for the other bands, and the guide to delta waves covers the band epsilon is said to sit beneath. For the rest of the map, the guides to brainwave frequencies and binaural beats set out the bands and how a beat is made, and the lambda wave guide checks another name sometimes added to the map.

Sources and limits

This guide sets the way the name epsilon is used beside the clinical glossary's definitions and the research on activity below 0.5 Hz. Its sources record no origin for the brainwave name, so none is given. The chart draws the bands to the glossary's limits; a range written "below" has no lower edge and is drawn from 0.01 Hz, the chart's left margin. The diagram's signal is illustrative, built in code from three sine waves; its filter is the glossary's 0.3 s time constant. The delta session's beat is read from its master's own measurement, not from its recipe. The 0.25 Hz study is one small, industry-supported study of daytime naps, not yet repeated in the sources here.