Beats

Pluck two guitar strings that are almost in tune with each other and you hear something neither string is making: a single note that swells and fades, swells and fades — wah… wah… wah… Slow it down and the mystery sharpens. Each string vibrates at its own steady frequency with its own steady amplitude; nothing physical connects them; neither knows the other exists. Yet the sound reaching your ear pulses like a slow heartbeat. Where does the new, slow rhythm come from? The answer is the prettiest two-line calculation in oscillation physics. Add two cosines of nearly equal frequency and the sum factors: a fast oscillation at the average frequency — the note you hear — multiplied by a slow envelope at half the difference. The pulsing is called beats, and its rate carries sharp information: you hear a loudness maximum exactly $|f_1 - f_2|$ times per second. That is why a piano tuner needs no electronics — mistune a string by half a hertz against a reference and the error announces itself as one slow beat every two seconds, a timing your ear can judge to exquisite precision. Beats are the time-domain face of interference: the same crest-meets-crest, crest-meets-trough story as a two-slit pattern, playing out in time instead of space. And engineered beats run the modern world under the name heterodyne — every radio mixer, every laser locked to an optical clock, works by beating a signal against a reference and reading the slow difference note. Master the one identity at this article's heart and you own the tuner's trick, the factor-of-two subtlety everyone trips on, and the front end of every receiver on Earth.

Key takeaways

Superposing two equal-amplitude tones at nearby frequencies gives $\cos\omega_1 t + \cos\omega_2 t = 2\cos(\bar\omega t)\cos(\tfrac{\Delta\omega}{2}t)$ : a fast carrier at the average frequency $\bar\omega$ , multiplied by a slow envelope at half the difference $\Delta\omega/2$ .
The perceived beat frequency is the full difference $f_{\text{beat}} = |f_1 - f_2|$ , not half of it: loudness follows the magnitude of the envelope, which peaks twice per envelope cycle.
Beats require no coupling, no resonance, no nonlinearity — the two oscillators are completely independent; linear superposition does everything.
Tuning by beats: adjust one oscillator until the beats slow to zero; the residual beat rate is the frequency error in hertz, which is why the method resolves mistunings far too small to hear as pitch.
Unequal amplitudes give partial beats: the envelope swings between $A_1+A_2$ and $|A_1-A_2|$ but never reaches zero — the modulation weakens, the rate stays $|f_1-f_2|$ .
Heterodyne detection is beats engineered on purpose: mix a signal with a local oscillator and read the difference frequency, the working principle of radio receivers and laser beat-note locks.
Key idea: two nearby tones add to a carrier at the mean frequency inside an envelope at half the difference — and because the ear and every detector respond to the envelope's magnitude, the audible beat rate is the full difference $|f_1-f_2|$ .

The core insight

Two tones of nearly equal frequency superpose into a fast carrier at the average frequency whose amplitude breathes at half the difference frequency — and since loudness tracks the magnitude of that envelope, the perceived beat rate is the full difference $|f_1 - f_2|$ .

Where this lives today

The most precise measurements humans make are read out as beat notes. An optical atomic clock runs on a laser locked to a strontium transition at $429\,228\,004\,229\,873$ Hz — about $429$ THz, far too fast for any electronic counter. To compare two such clocks, physicists overlap the two laser beams on a photodiode and let them beat: the diode, a square-law (intensity) detector, outputs exactly the $1 + \cos(\Delta\omega\,t)$ of equation (4), a gentle radio-frequency oscillation at the difference of two optical frequencies. A drift of one hertz in the beat note — trivially countable — corresponds to a fractional frequency shift of $1/(4.29\times10^{14}) \approx 2.3\times10^{-15}$ ; today's best clock comparisons resolve beat-note changes corresponding to parts in $10^{18}$ , sharp enough to detect the gravitational time dilation from raising one clock a few centimetres. The frequency comb that mediates such comparisons is itself a beat machine: each measurement is a beat note between the clock laser and the nearest comb tooth. The same trick at humbler frequencies is in your pocket and on your bench: every phone and radio front-end mixes the incoming carrier against a local oscillator to heterodyne it down to a fixed intermediate frequency ( $10.7$ MHz in classic FM receivers), and laser labs "offset-lock" one laser a precise few megahertz from another by feeding their measured beat note into a feedback loop. Doppler radar guns beat the returned echo against the transmitted wave; the beat frequency is the speed reading. Whenever a frequency is too high, too optical, or too inconvenient to count directly, the working engineer's move is the one this article taught: make it beat against something you trust, and count the slow difference.

The physical picture

Picture the two sound waves arriving at your ear as two runners on a circular track, moving at almost the same speed. Right now they are shoulder to shoulder: their pressure crests arrive together, reinforce, and the sound is loud. But the slightly faster runner creeps ahead — a little more each lap — until she is exactly half a lap ahead. Now her crests arrive with the other's troughs: cancellation, near silence. Keep going and she gains the full lap; they are shoulder to shoulder again and the loudness returns. The swell-and-fade period is the time to gain one full lap — one full cycle of relative phase — so the beat rate equals the difference of the two frequencies. Nothing couples the runners. The beat lives entirely in how their combined effect adds up.

Figure

Two nearby tones and their sum — a fast carrier at the average frequency breathing inside a slow envelope

Read the bottom trace slowly, because the whole factor-of-two subtlety of beats is drawn in it. The dashed envelope completes exactly one full cycle across the frame — down through zero, to its most negative, back through zero, up to its starting value. But your ear does not care about the envelope's sign; a large negative envelope means large swings of air pressure just as surely as a large positive one. So the sound is loud at the left edge, silent a quarter of the way in, loud again in the middle — where the envelope is at its most negative — silent at three quarters, and loud at the right edge. Two loudness peaks per envelope cycle. That is why the beat you count is at the full difference frequency $\Delta\omega$ , not the envelope's $\Delta\omega/2$ . The runner picture becomes precise with phasors — each tone drawn as a rotating arrow in the complex plane, the physical signal being the sum:

Figure

The phasor picture — two arrows rotating at slightly different rates drift from aligned to opposed and back; their sum is the breathing envelope

The phasor picture also settles a question the algebra can obscure: the beat period is simply the lap-gaining time. The relative angle between the arrows grows as $\Delta\omega\,t$ , the sum arrow is longest whenever that angle passes a multiple of $2\pi$ , and successive maxima are therefore spaced by $T_{\text{beat}} = 2\pi/\Delta\omega$ — once per lap, in perfect agreement with the two-peaks-per-envelope-cycle count. Two independent clocks, one emergent rhythm.

The identity that does all the work

Set up the superposition. Two sources — tuning forks, strings, lasers — produce equal-amplitude oscillations at nearby angular frequencies, and a linear medium (air, in the concert-hall case) simply adds them at your detector. There is no interaction term to model, which is exactly what makes beats a kinematic phenomenon:

x(t) = A\cos\omega_1 t + A\cos\omega_2 t \qquad (1)

where: $A$ — the common amplitude; $\omega_1, \omega_2$ — the two angular frequencies, taken close together ( $|\omega_1 - \omega_2| \ll \omega_1$ ); the two phases are chosen zero at $t = 0$ , which just sets the clock so a loud moment happens at the start. Each term separately is boring. The magic is entirely in the sum.

Derivation AThe beat identity from plain superposition

We derive the factored form $x = 2A\cos(\bar\omega t)\cos(\Delta\omega\,t/2)$ from equation (1), and — the part that matters — learn to read it as a fast carrier at the average frequency inside a slow envelope at half the difference. The route is to rewrite each frequency relative to their mean, expand with the angle-addition formula, and watch half the terms cancel. This is deliberately the pedestrian route rather than quoting the sum-to-product identity from a table: seeing which terms cancel shows why equal amplitudes matter, and the same maneuver with unequal amplitudes (later) shows exactly what survives when they don't. Watch the definitions: $\bar\omega$ is the average of the two frequencies but $\Delta\omega/2$ is half their difference — mixing those up is the source of every factor-of-two error in this topic.

Step 1. Define the mean and the difference, and write each frequency as the mean plus or minus half the difference: $\omega_1 = \bar\omega + \tfrac{\Delta\omega}{2}$ and $\omega_2 = \bar\omega - \tfrac{\Delta\omega}{2}$ , with $\bar\omega = \tfrac12(\omega_1 + \omega_2)$ and $\Delta\omega = \omega_1 - \omega_2$ . Why recenter: the two tones are nearly identical, so the natural variables are "what they share" (the mean) and "how they differ" (the split) — the physics of beats is the physics of that split.

Step 2. Expand both cosines with $\cos(\alpha \pm \beta) = \cos\alpha\cos\beta \mp \sin\alpha\sin\beta$ , taking $\alpha = \bar\omega t$ and $\beta = \tfrac{\Delta\omega}{2}t$ :

x = A\big[\cos\bar\omega t\cos\tfrac{\Delta\omega}{2}t - \sin\bar\omega t\sin\tfrac{\Delta\omega}{2}t\big] + A\big[\cos\bar\omega t\cos\tfrac{\Delta\omega}{2}t + \sin\bar\omega t\sin\tfrac{\Delta\omega}{2}t\big]

Why the signs matter: the two tones sit symmetrically on either side of the mean, so the cross terms come in with opposite signs — equal amplitudes guarantee they cancel identically.

Step 3. Add. The $\sin\bar\omega t$ terms kill each other; the $\cos\bar\omega t$ terms double:

x(t) = 2A\,\cos\!\left(\tfrac{\Delta\omega}{2}\,t\right)\cos\big(\bar\omega t\big) \qquad (2)

where: $\cos\bar\omega t$ — the carrier, oscillating at the average of the two frequencies (this is the pitch you hear); $2A\cos(\Delta\omega\,t/2)$ — the envelope, a slowly varying effective amplitude, completing one cycle in $4\pi/\Delta\omega$ . For two tones at $440$ and $442$ Hz the carrier runs at $441$ Hz while the envelope takes a leisurely full second per cycle — a factor of nearly $500$ between the time scales, which is what makes the ear parse the product as "one note, breathing" rather than as two notes.

Step 4. Read the structure. Nothing oscillates at $\omega_1$ or $\omega_2$ anymore — yet nothing has been lost, since the identity is exact and reversible. The same signal has two honest descriptions: two steady tones (the frequency-domain view), or one breathing tone (the time-domain view). Beats are what interference looks like when you watch it in time. Two quick checks confirm the identity: set $\omega_1 = \omega_2$ and equation (2) collapses to $x = 2A\cos\omega t$ — constant envelope, no beats, infinite beat period, as expected; and expanding (2) back with the product-to-sum identity regenerates equation (1) exactly.

Equation (2) is exact for any pair of frequencies — the algebra never assumed they were close. What closeness buys is the separation of time scales that makes the factored reading meaningful: with $\Delta\omega \ll \bar\omega$ the envelope holds nearly still over hundreds of carrier cycles, and "a tone at $\bar\omega$ with slowly varying amplitude" becomes the physically right description. Push the frequencies apart until $\Delta\omega \sim \bar\omega$ and the identity still holds, but the "envelope" wiggles nearly as fast as the "carrier" and the decomposition stops earning its keep — you just hear two different notes.

Why you hear the difference, not half of it

Here is the trap that catches nearly everyone once. The envelope in equation (2) oscillates at angular frequency $\Delta\omega/2$ — half the difference. Yet every textbook, every tuner, and your own ears insist the beat frequency is the full difference $|f_1 - f_2|$ . Both statements are correct, and the reconciliation deserves to be done carefully, with numbers.

Go deeper · Derivation B · 2 minThe perceived beat rate is the full difference frequency

Derivation BThe perceived beat rate is the full difference frequency

We show that the number of loudness maxima per second is $f_{\text{beat}} = |f_1 - f_2|$ , twice the envelope's own frequency $\Delta\omega/2$ , by asking what a real detector — an ear, a photodiode, a power meter — actually responds to. The strategy is to compute the intensity, the mean-square signal averaged over a few fast carrier cycles (which is what any slow detector does automatically), and find its oscillation frequency. The point of principle: detectors are insensitive to the overall sign of the waveform, and discarding a sign doubles a frequency.

Step 1. Identify what loudness follows. The ear responds to the power in the sound wave — proportional to the square of the pressure amplitude — not to the instantaneous pressure, whose $\sim\!441$ Hz wiggles are far too fast to register as loudness changes. So loudness tracks $|2A\cos(\Delta\omega\,t/2)|$ , the envelope's magnitude. Why the absolute value: an envelope of $-2A$ produces exactly as violent a pressure swing as $+2A$ ; the sign records an unobservable overall phase flip of the carrier.

Step 2. Count the maxima of the magnitude. The function $|\cos\theta|$ has period $\pi$ , not $2\pi$ — folding the negative lobes up halves the period. With $\theta = \tfrac{\Delta\omega}{2}t$ , maxima of $|{\cos\theta}|$ recur when $\theta$ advances by $\pi$ , i.e. every

T_{\text{beat}} = \frac{\pi}{\Delta\omega/2} = \frac{2\pi}{\Delta\omega}, \qquad f_{\text{beat}} = \frac{\Delta\omega}{2\pi} = |f_1 - f_2| \qquad (3)

where: $T_{\text{beat}}$ — the time between successive loudness peaks; the half from the envelope and the half from the folding cancel, leaving the bare difference. Why this is also the phasor result: one loudness peak per full lap of relative phase, and relative phase advances at $\Delta\omega$ .

Step 3. Confirm with the intensity computed honestly. Square equation (2) and average over a few carrier periods, using $\langle\cos^2\bar\omega t\rangle = \tfrac12$ and then the double-angle identity $2\cos^2\theta = 1 + \cos 2\theta$ :

\big\langle x^2\big\rangle_{\text{fast}} = 2A^2\cos^2\!\left(\tfrac{\Delta\omega}{2}t\right) = A^2\big[1 + \cos(\Delta\omega\,t)\big] \qquad (4)

where: the intensity oscillates at exactly $\Delta\omega$ — the doubling is now automatic, performed by the squaring itself. No absolute values, no hand-waving: the measurable power waxes and wanes at the full difference frequency.

Step 4. Put in the standard numbers. Forks at $f_1 = 442$ Hz and $f_2 = 440$ Hz: the envelope frequency is $(f_1-f_2)/2 = 1$ Hz — one envelope cycle per second — but you hear loudness maxima at $t = 0,\ 0.5,\ 1.0,\ 1.5\ \mathrm{s},\dots$ : two beats per second, at the full $2$ Hz difference. Count beats for ten seconds, get twenty, and you have measured a $0.45\%$ frequency mismatch with no instrument but your ear. Two checks on equation (4): it ranges from $2A^2$ (constructive, twice the mean intensity — not four times, since the mean of the cosine term is zero) down to $0$ (destructive), and its average $A^2$ equals the sum of the two tones' individual mean-square intensities $\tfrac12A^2 + \tfrac12A^2$ — energy is redistributed in time, never created or destroyed, the signature of all interference.

The last check in that derivation deserves its own sentence, because it is the interference story in miniature: beating does not amplify anything on average. The loud moments ( $2A^2$ , four times one tone's intensity peak-to-mean) are paid for, joule for joule, by the silent ones. A two-slit pattern makes the same trade in space — bright fringes funded by dark ones — and beats are literally the same physics with $t$ in place of position on the screen. The ear, sweeping through the pattern at a fixed point in space, hears the fringes go by in time.

Partial beats: when the amplitudes don't match

Real pairs of sources rarely match in loudness — one guitar string rings louder than the other, one laser outshines its reference. The beats survive, but the silences don't. The phasor picture makes the result almost obvious before any algebra: an arrow of length $A_1$ and an arrow of length $A_2$ , drifting through all relative angles, add to a maximum of $A_1 + A_2$ when aligned and a minimum of $|A_1 - A_2|$ when opposed — which is zero only if the lengths match.

Go deeper · Derivation C · 2 minUnequal amplitudes and the depth of the beats

Derivation CUnequal amplitudes and the depth of the beats

We derive the envelope of $x = A_1\cos\omega_1 t + A_2\cos\omega_2 t$ for unequal amplitudes and quantify how "deep" the beating is. The strategy is the phasor sum: the two arrows' relative angle grows as $\Delta\omega\,t$ , and the resultant's length follows from the law of cosines applied to the parallelogram of the two arrows. This route replaces the trigonometric grind with one line of geometry, and it exposes the structure: the beat rate is untouched by the amplitude mismatch; only the depth of modulation suffers. The watch-point is that with unequal amplitudes the resultant's phase also wobbles slowly — the carrier is no longer exactly $\cos\bar\omega t$ — but for loudness purposes only the length matters.

Step 1. Draw the two phasors at time $t$ : lengths $A_1$ and $A_2$ , relative angle $\phi(t) = \Delta\omega\,t$ . The resultant's squared length is given by the law of cosines (with the interior angle $\pi - \phi$ , hence the plus sign):

A_{\text{eff}}^2(t) = A_1^2 + A_2^2 + 2A_1A_2\cos(\Delta\omega\,t) \qquad (5)

where: $A_{\text{eff}}$ — the slowly varying envelope amplitude; the cosine term is the interference term, oscillating at the full difference frequency $\Delta\omega$ (no factor of two ambiguity survives here — the squared envelope beats at $\Delta\omega$ directly, consistent with equation (4), which is this formula's $A_1 = A_2$ special case). Why the intensity language is cleaner for unequal amplitudes: the squared form never goes negative, so there is no folding to track.

Step 2. Read off the extremes. Aligned phasors ( $\Delta\omega\,t = 2\pi n$ ): $A_{\text{eff}} = A_1 + A_2$ . Opposed ( $\Delta\omega\,t = (2n+1)\pi$ ): $A_{\text{eff}} = |A_1 - A_2|$ . The beats swing between these; total silence requires $A_1 = A_2$ exactly. Why the survivor: at the destructive moment the smaller tone cancels its own size out of the larger one, and the excess $|A_1 - A_2|$ keeps sounding at the carrier pitch.

Step 3. Quantify the depth with the modulation depth, the standard figure of merit borrowed from radio engineering:

m = \frac{A_{\text{eff}}^{\max} - A_{\text{eff}}^{\min}}{A_{\text{eff}}^{\max} + A_{\text{eff}}^{\min}} = \frac{(A_1 + A_2) - |A_1 - A_2|}{(A_1 + A_2) + |A_1 - A_2|} = \frac{\min(A_1,A_2)}{\max(A_1,A_2)} \qquad (6)

where: $m$ — modulation depth, running from $1$ (equal amplitudes, full beats with true silences) toward $0$ (one tone dominant, barely a flutter). The simplification to a simple ratio is a pleasant surprise: the audibility of the beats is set purely by how close the two loudnesses are, on a scale where $m = A_2/A_1$ for $A_2 < A_1$ . Two limiting cases confirm equation (5): for $A_1 = A_2 = A$ , it gives $A_{\text{eff}}^2 = 2A^2(1 + \cos\Delta\omega t) = 4A^2\cos^2(\Delta\omega t/2)$ , whose square root is the $|2A\cos(\Delta\omega t/2)|$ envelope of equation (2), and $m = 1$ ; for $A_2 = 0$ , $A_{\text{eff}} = A_1$ constant and $m = 0$ — both limits behave as expected.

Figure

Partial beats — with amplitudes 2 to 1 the envelope pinches but never closes; the beat rate is unchanged

Beats are not amplitude modulation — and the difference is measurable

The bottom trace of the hero figure looks like an amplitude-modulated (AM) signal — a carrier whose loudness is deliberately varied, as in AM radio. The resemblance is real but the anatomy differs, and a spectrum analyzer tells them apart instantly. Beats are the sum of exactly two spectral lines, at $f_1$ and $f_2$ ; there is no energy whatsoever at the beat frequency $\Delta f$ or at the carrier $\bar f$ — those frequencies are perceptual, not spectral. True AM, $x = (1 + m\cos\Omega t)\cos\bar\omega t$ , multiplies rather than adds, and multiplication manufactures three lines: the carrier at $\bar\omega$ plus two sidebands at $\bar\omega \pm \Omega$ . Expand the beats formula (2) backwards and you find it is precisely AM with the carrier line removed — what radio engineers call suppressed-carrier modulation. The practical consequence: to extract a real, countable oscillation at $\Delta\omega$ from two beating tones, some device must perform a multiplication — squaring, rectifying, any nonlinearity. Your ear's intensity response (equation 4) is exactly such a squarer, which is why you hear the beat even though no instrument tuned to $2$ Hz would detect anything in the air. That one sentence is the engine of heterodyne detection, beats built on purpose. Mix an incoming signal at $\omega_1$ with a stable local oscillator at $\omega_2$ in a nonlinear element, and the output contains a genuine spectral component at the difference — shifted down from, say, hundreds of megahertz to a convenient intermediate frequency where amplifiers and filters are cheap and precise:

Figure

Heterodyne detection — engineered beats shift a signal to a low difference frequency that is easy to amplify and count

Where beats appear elsewhere in this cluster

Two siblings of this article are beats wearing dynamical clothes, and it is worth naming them precisely so you recognize the pattern. First: switch on a sinusoidal drive near (but not at) a lightly damped oscillator's natural frequency, and during the settling window the decaying free oscillation at $\omega_0$ coexists with the steady response at $\omega$ — two nearby frequencies in one signal, and the amplitude visibly beats at $|\omega - \omega_0|$ until the transient dies. That ringing-through-onset story belongs to the Transient and Steady-State Response article. Second: couple two identical pendulums with a weak spring and start one swinging; the energy drains completely into the other and back, with a slosh period set by the difference of the two normal-mode frequencies. What oscillates there is energy between bodies, not just a summed displacement — but the mathematics is equation (2) applied to the two mode coordinates, and the Coupled Oscillators article owns the mechanism. If you can read the beat identity fluently here, both phenomena become one-line recognitions there.

Limiting cases

Case	Result	Meaning
identical frequencies, $\Delta\omega \to 0$	$x \to 2A\cos\omega t$ , $\ T_{\text{beat}} \to \infty$	beats slow to a stop — the tuner's endpoint; only relative phase (fixed forever) sets the loudness
nearby frequencies, $\Delta\omega \ll \bar\omega$	$x = 2A\cos(\tfrac{\Delta\omega}{2}t)\cos\bar\omega t$	clean separation of time scales: one audible pitch at $\bar f$ , loudness beating at $\vert f_1 - f_2\vert$
well-separated frequencies, $\Delta\omega \sim \bar\omega$	identity still exact, reading fails	envelope and carrier change comparably fast — perceived as two distinct notes, not one beating note
unequal amplitudes, $A_1 \ne A_2$	envelope between $A_1{+}A_2$ and $\vert A_1{-}A_2\vert$	partial beats: same tempo, shallower modulation, no true silences; depth $m = \min/\max$
one amplitude zero, $A_2 = 0$	$x = A_1\cos\omega_1 t$	no interference partner, no beats — a reminder that beating needs two coherent tones

Common trapA 440 Hz fork and a 442 Hz fork beat once per second

They beat twice per second. The tempting-but-wrong chain of reasoning: the envelope is $\cos(\tfrac{\Delta\omega}{2}t)$ , half the difference of the frequencies is $1$ Hz, so one beat per second. The error is forgetting that loudness follows the envelope's magnitude: the moment of most-negative envelope is exactly as loud as the moment of most-positive envelope, so each envelope cycle delivers two loudness maxima, and the audible beat rate is the full $|f_1 - f_2| = 2$ Hz. The safest habit is to skip the envelope entirely and use the intensity, $\langle x^2\rangle \propto 1 + \cos(\Delta\omega\,t)$ , which oscillates at the full difference with no signs to fold — or the phasor picture, one beat per full lap of relative phase. In an exam, if your beat frequency came out as $(f_1-f_2)/2$ , you have almost certainly reported the envelope frequency by mistake.

Worked examples

1 · Tuning a guitar string against a fork (confidence builder)

A guitarist tunes the A-string against a $440.0$ Hz tuning fork. Sounding them together, she counts $12$ beats in $4.0$ seconds. She tightens the string slightly and the beating slows to $3$ beats in $5.0$ seconds. Find the string's frequency at each stage (given that tightening raises pitch), the perceived pitch during the first comparison, and how long she must listen to confirm the final tuning is within $0.1$ Hz.

Step 1. First beat rate: $f_{\text{beat}} = 12/4.0 = 3.0$ Hz, so the string is off by exactly $3.0$ Hz: either $437.0$ or $443.0$ Hz. Beats give the magnitude of the error, never its sign.

Step 2. Resolve the sign with the experiment she actually performed. Tightening the string raises its frequency, and the beats slowed — the string moved toward the fork. That only makes sense if it started below: $f_{\text{string}} = \boxed{437.0\ \mathrm{Hz}}$ . (Had it started at $443$ Hz, tightening would have sped the beats up to $4$ Hz or more.)

Step 3. After tightening: $f_{\text{beat}} = 3/5.0 = 0.60$ Hz, and still below the fork (she moved up by $2.4$ Hz, not past $3.0$ ), so $f_{\text{string}} = 440.0 - 0.6 = \boxed{439.4\ \mathrm{Hz}}$ .

Step 4. Perceived pitch during the first comparison: the carrier at the mean, $\bar f = \tfrac12(440.0 + 437.0) = 438.5$ Hz — a hair flat of concert pitch, pulsing three times a second. Nobody hears "two notes"; the $3$ Hz separation is far inside the ear's fusion range.

Step 5. Confirming $0.1$ Hz: a mistuning of $0.1$ Hz produces one beat every $10$ seconds. To be confident she is within that, she must hear no complete swell-and-fade over roughly one beat period — so listen for about $\boxed{10\ \mathrm{s}}$ of steady loudness. This is the quiet superpower of the method: a $0.1$ Hz error at $440$ Hz is $0.023\%$ , hopelessly beyond pitch judgment, yet trivially countable as a slow beat.

2 · Partial beats from mismatched forks (exam level)

Two tuning forks at $f_1 = 440$ Hz and $f_2 = 443$ Hz produce pressure amplitudes at the listener's ear of $A_1 = 1.0$ (in some unit) and $A_2 = 0.6$ . Find the beat frequency, the loudest and quietest envelope amplitudes, the modulation depth, the intensity swing in decibels, and the perceived pitch.

Step 1. Beat frequency first, because nothing about the amplitudes affects it: $f_{\text{beat}} = |f_1 - f_2| = \boxed{3.0\ \mathrm{Hz}}$ , a loudness maximum every $0.33$ s.

Step 2. Envelope extremes from the phasor geometry: $A_{\max} = A_1 + A_2 = 1.6$ and $A_{\min} = |A_1 - A_2| = 0.4$ . The sound never dies: at the quietest moment, the unmatched $0.4$ of the louder fork keeps sounding.

Step 3. Modulation depth via equation (6): $m = \min/\max = 0.6/1.0 = \boxed{0.6}$ — deep enough to be blazingly obvious to the ear, though short of full silence.

Step 4. Intensity swing. Intensity goes as amplitude squared, so the loud-to-quiet ratio is $(A_{\max}/A_{\min})^2 = (1.6/0.4)^2 = 16$ , which in decibels is $10\log_{10}16 = \boxed{12\ \mathrm{dB}}$ . Cross-check with equation (5): $A_{\text{eff}}^2$ swings between $1.0 + 0.36 + 1.2 = 2.56$ and $1.36 - 1.2 = 0.16$ , ratio $16$ . ✓

Step 5. Perceived pitch: the mean $\tfrac12(440 + 443) = 441.5$ Hz — with a subtlety worth a mark: because the amplitudes are unequal, the resultant's phase wobbles slightly, dragging the instantaneous frequency periodically toward the stronger fork's $440$ Hz. The dominant perception remains a $441.5$ Hz-ish tone pulsing at $3$ Hz, $12$ dB deep. The exam habits: beat rate from the difference alone; depth from the amplitude ratio alone; never let the two calculations contaminate each other.

Origins · 1 minHermann von Helmholtz and how this was discovered

Hermann von Helmholtz

1821–1894 · Heidelberg and Berlin

Helmholtz — physician, physiologist, and physicist in one career — made beats the cornerstone of the first physical theory of musical harmony. In On the Sensations of Tone (1863), written while he was professor of physiology at Heidelberg, he traced consonance and dissonance to countable physics: when two tones (or their overtones) lie a few hertz apart they beat slowly and pleasantly; pushed to some tens of beats per second the pulsing fuses into a grating roughness, and that roughness, he argued, is what we call dissonance. An octave is smooth because every overtone of the upper note lands exactly on an overtone of the lower — nothing beats. He built tuned resonators to dissect complex sounds ear-by-frequency, effectively performing Fourier analysis with brass spheres, and his beat-based reasoning still underlies how piano tuners lay temperaments today. That the same man also formulated the conservation of energy in full generality (1847) and mentored Heinrich Hertz makes him one of the century's great bridges between physics and perception.

History

1636

— Mersenne's Harmonie universelle records what organ builders already knew: two pipes slightly out of tune pulse audibly, and the pulsing slows as they approach unison.

1863

— Helmholtz's On the Sensations of Tone explains beats quantitatively and builds from them a physical theory of consonance, dissonance, and musical roughness.

1901

— Fessenden conceives the heterodyne principle: beat a radio signal against a local oscillator and read the audible difference tone — engineered beats enter technology.

1918

— Armstrong's superheterodyne receiver perfects the idea; mixing every station down to one fixed intermediate frequency still runs virtually all radio receivers.

2005

— Hall and Hänsch share the Nobel Prize for the optical frequency comb, which measures optical frequencies by counting beat notes between laser light and comb teeth.

Going further

References. Morin, Introduction to Classical Mechanics, Ch. 4 (Oscillations) · Goldstein, Classical Mechanics, 3rd ed., Ch. 6 (Oscillations) · Landau & Lifshitz, Mechanics, 3rd ed., §22-23 · Taylor, Classical Mechanics, Ch. 5 (Oscillations) · French, Vibrations and Waves, Ch. 1 The sum-to-product identity gives $x=2A\cos(\Delta\omega t/2)\cos(\bar\omega t)$ ; the factor-of-two relation $f_{\text{beat}}=|f_1-f_2|$ follows from the $|\cos|$ period; the A440/442 example gives $T_{\text{beat}}=1/2$ s, and 256/260 Hz gives 4 beats/s. No empirical constants are used; all numbers are standard tuning-fork/concert-pitch values. The additional worked numbers on this page — 3.0 Hz and 0.60 Hz beat rates, $m = 0.6$ , 12 dB swing, 441.5 Hz mean pitch; 10.7 MHz IF; Sr clock frequency 429,228,004,229,873 Hz and the $2.3\times10^{-15}$ per-hertz fraction; 6.9 kHz radar Doppler shift at 34.7 GHz for 30 m/s — are illustrative arithmetic computed from $f_{\text{beat}} = |f_1-f_2|$ , equations (5)–(6), and the standard Doppler formula $\Delta f = 2fv/c$ .

Builds on: Simple Harmonic Motion · Amplitude, Phase, and Angular Frequency · Complex Exponential Representation of Oscillations Leads to: Driven Oscillations · Resonance · Small Oscillations About Equilibrium

How can two forks beat if nothing connects them? Doesn't interference require the waves to interact?

No interaction ever happens — and that is the point the article's phrase "kinematic phenomenon" is making. Each fork broadcasts its own steady wave through the air, and air, being a linear medium at these amplitudes, simply carries the sum of the two pressure patterns. The waves pass through each other unchanged, like ripples crossing on a pond. The beat exists only in the summed quantity at a given point — your eardrum, which can only be at one pressure at a time. "Interference" is a slightly misleading historical word: nothing interferes with anything; amplitudes add, and the added total happens to swell and fade. That is why the two sources need no coupling, no resonance, and no knowledge of each other — and why beating two lasers on a photodiode works across completely independent machines.

Is the beat physically present in the air, or is my ear inventing it?

Split the question in two. The envelope is physically present: the air pressure at your ear really does swing through large excursions, then small ones, at the difference rhythm — a pressure recording shows exactly the pinched waveform of the hero figure. What is not present in the air is any oscillation at the beat frequency: Fourier-analyze the pressure and you find lines at $f_1$ and $f_2$ only, nothing at $\Delta f$ . Your ear supplies the final step because it responds to intensity — a square-law operation — and squaring genuinely creates a component at $\Delta\omega$ (equation 4). So the ear is not hallucinating; it is performing a mathematically honest nonlinear operation on a real envelope. Every heterodyne receiver does the identical thing with a diode instead of a cochlea.

Beats tell me two frequencies differ by 3 Hz, but not which one is higher. How do tuners get the sign?

By perturbing and listening again — the trick in worked example 1. Beats are blind to sign because $|f_1 - f_2|$ is symmetric, so a string 3 Hz flat and a string 3 Hz sharp sound identical against the reference. But change breaks the symmetry: tighten the string slightly (raising its pitch) and if the beats slow down you were flat and are converging; if they speed up you were sharp and are diverging. Musicians internalize this as "tune through the beat": approach from below, hear the beats decelerate to a stop, and if you overshoot they reappear and accelerate. Electronic tuners cheat by measuring the actual frequency, but the perturbation method is self-calibrating and needs no hardware — which is why it survived four centuries of technology.

Why do beats become inaudible when the two frequencies are far apart? The identity still holds.

The identity is exact at any separation, but perception has time scales. At a few hertz of difference the loudness swells slowly enough for the ear to track — countable beats. Push $\Delta f$ to a few tens of hertz and the swelling is too fast to follow as loudness but too slow to fuse; the ear reports a harsh roughness, which Helmholtz identified as the physical basis of dissonance. Beyond roughly the ear's critical bandwidth (very crudely, $10$ – $15\%$ of the center frequency, so tens of hertz at concert pitch), the cochlea resolves the two tones into separate channels that no longer share a detector — and with no common square-law element, the difference frequency is never generated. You then simply hear two notes. The progression beats → roughness → two tones, swept in seconds on a synthesizer, is one of the nicest demonstrations that hearing is physics plus physiology.

The coupled-pendulum demonstration looks exactly like beats. Is energy also sloshing between my two tuning forks?

No — and the comparison is worth making precise because the waveforms genuinely look alike. Two independent forks each keep their own energy forever (minus damping); what beats is only the summed signal at your ear, while each source hums along unaffected. Two coupled pendulums are different machinery: the coupling spring splits the system into two normal modes with slightly different frequencies, and a single struck pendulum starts in a superposition of both modes. Each pendulum's displacement is then a two-frequency sum — mathematically the beat identity again — but now the envelope tracks a physical transfer: when pendulum one's envelope pinches to zero, its energy has genuinely moved into pendulum two. Same equation, different referent: summed signal versus individual body. The Coupled Oscillators article derives where the two mode frequencies come from.

If I play two notes an octave apart, the difference frequency equals the lower note. Why don't I hear beats then?

Because beats need nearly equal frequencies for the carrier–envelope reading to mean anything. At an octave, $f_2 = 2f_1$ , the "envelope" of the identity would oscillate at $f_1/2$ — comparable to the "carrier" at $3f_1/2$ — so there is no separation of time scales and no slow loudness variation; the waveform is simply periodic at $f_1$ , which the ear hears as a stable, consonant dyad. Helmholtz's point survives in a subtler form, though: if the octave is slightly mistuned, the second harmonic of the lower note (at $2f_1$ , physically present in any real instrument's tone) sits a few hertz from the upper note and beats with it audibly. Piano tuners tune octaves exactly this way — not by hearing the fundamental difference, but by silencing the beat between harmonic and note.

Does the beat identity work for sound from moving sources, like two passing sirens?

Yes, and it is a working measurement principle. If motion Doppler-shifts one source's received frequency, the beat between the shifted and unshifted tones counts the shift directly: a radar gun beats the echo returned from a car against its own transmitted frequency, and the resulting beat note is proportional to the car's speed. At $34.7$ GHz (a common radar band), a car at $30$ m/s shifts the echo by about $2f v/c \approx 6.9$ kHz — a trivially countable beat riding on an uncountable carrier. The one caveat is that the identity assumes both tones are steady during a beat period; a rapidly accelerating source smears the beat note, which is itself information (radar processors read acceleration from exactly that smear).

Beats

Where this lives today

The physical picture

The identity that does all the work

Why you hear the difference, not half of it

Partial beats: when the amplitudes don't match

Beats are not amplitude modulation — and the difference is measurable

Where beats appear elsewhere in this cluster

Limiting cases

Worked examples

1 · Tuning a guitar string against a fork (confidence builder)

2 · Partial beats from mismatched forks (exam level)

Going further

Before this

Where this leads