Angular Resolution, Seeing and Diffraction

How finely can a telescope see? Not, it turns out, as finely as you might wish by simply building it larger. Two hard limits stand in the way. The first is set by the wave nature of light itself: a beam passing through an aperture of finite diameter $D$ cannot be confined to a perfect point but spreads into a diffraction pattern of angular scale $\sim\lambda/D$ , so even a flawless mirror images a star as a small disk. For a circular pupil that disk is the Airy pattern, and two stars closer than $\theta_{\min}\approx1.22\,\lambda/D$ merge into one. This diffraction limit is the best any instrument of that size can do, and it sharpens as the aperture grows or the wavelength shrinks. The second limit is the sky itself. Above every ground-based observatory the atmosphere churns, mixing parcels of air at slightly different temperatures and hence different refractive indices, so the incoming wavefront — flat when it left the star — arrives crumpled. The largest patch over which it stays coherent, the Fried parameter $r_0$ , is only $10$ – $20\,\mathrm{cm}$ in visible light, so a long exposure smears the star into a 'seeing disk' of width $\sim\lambda/r_0\approx0.5''$ to $2''$ . A ten-metre telescope thus has far more diffraction-limited sharpness than the air will allow: it collects vastly more light, but without help delivers no finer detail than a backyard scope. That gap between what diffraction permits and what the atmosphere concedes is the engine of modern high-resolution astronomy: it drives telescopes into space, spawns adaptive optics that reshape a mirror a thousand times a second, and links continents into interferometers whose baseline $B$ replaces $D$ . Resolution, in the end, governs which binaries yield masses, how sharply galaxies and lenses are mapped, and how far up the distance ladder we can climb.

Key takeaways

A telescope of aperture $D$ observing at wavelength $\lambda$ cannot resolve angular detail finer than the diffraction limit $\theta_{\min}\approx1.22\,\lambda/D$ , set by the Airy pattern of a circular pupil.
The Rayleigh criterion fixes the number $1.22$ : two point sources are just resolved when the peak of one Airy disk lands on the first dark ring of the other, at which point their summed profile dips to about 74% of the peak.
Earth's atmosphere blurs long exposures through turbulence measured by the Fried parameter $r_0\sim10$ – $20\,\mathrm{cm}$ ; the resulting seeing disk $\sim\lambda/r_0\approx0.5''$ – $2''$ usually swamps diffraction for any large telescope.
Below the seeing limit, extra aperture buys light-gathering area $\propto D^2$ (sensitivity) but not resolution — the two must not be confused.
Space telescopes, adaptive optics, speckle imaging and interferometry recover or exceed the diffraction limit; an interferometer reaches $\theta\sim\lambda/B$ with baseline $B\gg D$ .
Key idea: a finite aperture $D$ diffracts light into an Airy pattern of angular scale $\sim\lambda/D$ , setting the ultimate resolution, while atmospheric turbulence imposes a coarser practical floor $\sim\lambda/r_0$ that must be beaten by observing from space, correcting the wavefront, or synthesizing a larger baseline.

The core insight

A finite aperture $D$ diffracts light into an Airy pattern of angular scale $\sim\lambda/D$ , setting the ultimate resolution; atmospheric turbulence imposes a coarser practical limit $\sim\lambda/r_0$ that must be overcome by space observation, adaptive optics, or interferometry.

Where this lives today

Resolution is where much of frontier astronomy is decided, and every trick in the toolbox is in active use. Adaptive optics is now standard on every 8–10 m telescope: a wavefront sensor reads the distortion off a natural or laser guide star and a deformable mirror reshapes itself thousands of times a second, restoring near-diffraction-limited images in the near-infrared. That let the Keck and VLT teams track individual stars orbiting the $4\times10^6\,M_\odot$ black hole Sgr A* at the Galactic Centre — work recognized with the 2020 Nobel Prize. At the long-baseline extreme, the Event Horizon Telescope pushed radio VLBI to its planetary limit: an Earth-sized baseline at $\lambda\approx1.3\,\mathrm{mm}$ delivered $\sim20\,\mu\mathrm{as}$ resolution and, in 2019 and 2022, the first images of the shadows of the black holes in M87 and our own Galaxy. From space, Hubble's $0.05''$ optical images and JWST's diffraction-limited infrared views anchor the distance scale, while the coming $30$ – $40\,\mathrm{m}$ ground giants — the ELT, TMT and GMT — are being built around adaptive optics precisely because their enormous apertures are otherwise worthless for resolution under the open sky.

The physical picture

A wave that squeezes through an opening of width $D$ loses the information about exactly where within that opening it passed, and by the uncertainty relation between position and transverse momentum that ignorance reappears as an angular spread $\Delta\theta\sim\lambda/D$ . A small pupil samples the wavefront over a short baseline and pins the arrival direction only crudely; a large pupil averages over a wide baseline and localizes the source far more precisely. Diffraction, in this reading, is not a defect of the optics but the price of a finite aperture — and the reason a bigger mirror is intrinsically sharper. For a circular pupil the spread is a structured bullseye, the Airy pattern: a bright central disk ringed by faint concentric maxima, with the first dark ring at the angle that sets the resolution.

Figure

Two equal point sources at the Rayleigh separation θ = 1.22 λ/D: each images to an Airy disk (dashed), and their sum (solid ink) dips to 74% of the peak between them — the textbook threshold for calling two stars just resolved. For Hubble with D = 2.4 m at λ = 550 nm this separation is 0.058 arcsec.

Now let the same starlight fall on a ground-based mirror. The atmosphere behaves like a drifting mosaic of weak lenses, each roughly $r_0\sim15\,\mathrm{cm}$ across, each imposing a random tilt and delay on the piece of wavefront it covers. A small telescope peers through essentially one such cell and stays nearly diffraction-limited — though it gathers little light. A large telescope straddles thousands of cells whose scrambled phases add up so that the star's light, instead of collapsing into a $\lambda/D$ Airy disk, spreads into a boiling $\lambda/r_0$ blob that a long exposure averages into the smooth seeing disk. The resolution is then set by the atmosphere, not the mirror, and stays stubbornly near $1''$ however big $D$ becomes.

Figure

Angular resolution versus aperture at λ = 550 nm: the diffraction limit θ = 0.14 arcsec ÷ (D in metres) falls as 1/D (blue), but uncorrected ground-based seeing pins images to a floor near 1 arcsec (amber dashed) whatever the aperture. Hubble in space rides the diffraction line to 0.058 arcsec; an 8 m telescope gains light but no sharpness until adaptive optics restores it to 0.017 arcsec.

The core equations

The starting point is the intensity pattern a uniformly illuminated circular aperture casts in the far field. Fraunhofer diffraction makes this pattern the squared magnitude of the pupil's Fourier transform, and for a circle that transform is a first-order Bessel function. The result is the Airy pattern, a central disk surrounded by ever-fainter rings.

I(\theta) = I_0\left[\frac{2\,J_1(x)}{x}\right]^2,\qquad x = \frac{\pi D \sin\theta}{\lambda} \qquad (1)

where: $I(\theta)$ — intensity at angle $\theta$ from the optical axis; $I_0$ — peak intensity; $J_1$ — Bessel function of the first kind, order 1; $D$ — aperture diameter; $\lambda$ — wavelength. Reading: the bracket falls to zero at the first root of $J_1$ , carving out the dark ring that bounds the bright Airy disk. Locating that first null and converting the small angle to a resolvable separation gives the single most-used formula in observational optics — the Rayleigh criterion. It says the finest angular detail an aperture can register scales as the wavelength divided by the diameter, with a numerical coefficient fixed by the Bessel root.

\theta_{\min} = 1.22\,\frac{\lambda}{D} \qquad (2)

where: $\theta_{\min}$ — minimum resolvable angular separation, in radians; $1.22$ — coefficient from the first zero of $J_1$ , equal to $3.8317/\pi$ ; $\lambda$ — observing wavelength; $D$ — aperture diameter. Reading: halve $\lambda$ or double $D$ and you halve the smallest gap you can split. Astronomers rarely quote radians. Multiplying by the number of arcseconds in a radian turns the criterion into a working formula in the units of the sky.

\theta_{\min}('') \approx 1.22\,\frac{\lambda}{D}\times 206265 \qquad (3)

where: $206265$ — arcseconds per radian, equal to $180\times3600/\pi$ ; $\lambda/D$ — the dimensionless ratio (same units). Reading: at $\lambda=550\,\mathrm{nm}$ this collapses to $\theta_{\min}\approx0.14''/(D/\mathrm{m})$ . Against this ideal stands the atmosphere. Long-exposure seeing is set not by the mirror but by the largest coherent patch of turbulent air, the Fried parameter, so the blur width is the wavelength divided by $r_0$ rather than by $D$ .

\theta_{\mathrm{seeing}} \approx \frac{\lambda}{r_0} \qquad (4)

where: $\theta_{\mathrm{seeing}}$ — full width of the long-exposure seeing disk (radians); $r_0$ — Fried parameter, the atmospheric coherence length; $\lambda$ — wavelength. Reading: with $r_0\approx0.1$ – $0.2\,\mathrm{m}$ in the visible, $\theta_{\mathrm{seeing}}\approx0.5''$ – $2''$ at a good site. The Fried parameter is not fixed: it grows toward longer wavelengths and shrinks as you look through more air. This scaling is why the near-infrared sky is so much steadier than the visible one, and why targets low on the horizon look worse.

r_0 \propto \lambda^{6/5}\,(\cos z)^{3/5} \qquad (5)

where: $r_0$ — coherence diameter of the wavefront; $z$ — zenith angle (larger $z$ means more air-path and worse seeing); $\lambda$ — wavelength. Reading: doubling $\lambda$ enlarges $r_0$ by $2^{6/5}\approx2.3$ , so seeing and adaptive correction both ease dramatically toward the infrared. How much of the pupil the turbulence chops up is captured by the number of independent coherence cells across the aperture — the same count that fixes how many actuators an adaptive-optics mirror needs.

N \sim \left(\frac{D}{r_0}\right)^2 \qquad (6)

where: $N$ — number of independent atmospheric patches over the pupil (roughly the number of deformable-mirror actuators required); $D$ — aperture diameter; $r_0$ — Fried parameter. Reading: an $8\,\mathrm{m}$ mirror with $r_0=0.15\,\mathrm{m}$ sees $N\approx2800$ cells, so its corrector must reshape thousands of zones in real time. Finally, resolution need not come from a single dish. Two apertures separated by a baseline $B$ interfere, and the fringe spacing they measure resolves structure as if the light had passed through one mirror of diameter $B$ .

\theta \approx \frac{\lambda}{B} \qquad (7)

where: $\theta$ — angular resolution of the array; $B$ — baseline between apertures; $\lambda$ — observing wavelength. Reading: an Earth-sized baseline at millimetre wavelengths reaches tens of microarcseconds, enough to resolve a black-hole shadow.

Derivation AThe Rayleigh criterion from the Airy pattern

The goal is to turn the diffraction pattern of a circular aperture into a hard number for resolution. We take as given that a uniformly illuminated circular pupil, in the far field, produces the Airy intensity $I(\theta)\propto[2J_1(x)/x]^2$ — the squared Fourier transform of a disk. From there we locate the first dark ring, translate its position into an angle, and invoke Rayleigh's convention for when two overlapping disks count as separated. Every link in that short chain is where the famous factor $1.22$ actually comes from.

Step 1. Write the Airy argument $x=\pi D\sin\theta/\lambda$ and ask where the pattern first vanishes. The bracket $2J_1(x)/x$ is zero wherever $J_1(x)=0$ for $x\neq0$ , and the first positive root of the first-order Bessel function is $x=3.8317$ . Why: the bright central disk is bounded by this first null, so its angular radius is exactly the scale on which the image structure lives.

Step 2. Set $x=3.8317$ and solve for the angle: $\pi D\sin\theta/\lambda=3.8317$ gives $\sin\theta=(3.8317/\pi)\,\lambda/D=1.22\,\lambda/D$ . Why: the pure number $3.8317/\pi=1.2197$ is the entire origin of the coefficient $1.22$ — it is a property of Bessel functions, not of any telescope.

Step 3. Use the small-angle approximation. Astronomical separations are minute, so $\sin\theta\approx\theta$ and the angular radius of the Airy disk is $\theta\approx1.22\,\lambda/D$ radians. Why: at $\theta\sim10^{-6}$ rad the difference between $\sin\theta$ and $\theta$ is a part in $10^{12}$ , utterly negligible.

Step 4. Apply Rayleigh's convention. Two equal point sources are declared just resolved when the central peak of one Airy disk falls precisely on the first dark ring of the other — a separation equal to the disk radius, $\theta_{\min}=1.22\,\lambda/D$ . Why: at exactly this spacing the combined profile shows a central dip to about 74% of the peaks (as the first figure draws), a saddle the eye and detector can register; any closer and the dip fills in and the pair looks single. *The key numbers: the first Bessel root is $3.8317$ , giving coefficient $3.8317/\pi=1.2197\approx1.22$ ; the combined-profile dip at the Rayleigh separation works out to $0.735$ of the peak.*

Go deeper · Derivation B · 1 minThe diffraction limit in convenient units

Derivation BThe diffraction limit in convenient units

The goal is to convert the abstract $1.22\,\lambda/D$ into a number an observer can carry in their head and apply to any telescope on sight. We fix a fiducial visible wavelength, insert the radians-to-arcseconds factor, and reduce the whole expression to a one-line rule of thumb — then test it against the seeing floor to see why the diffraction limit is so rarely the operative one on the ground.

Step 1. Insert a reference wavelength. Take $\lambda=550\,\mathrm{nm}$ , near the peak of the visual band, so $\theta_{\min}=1.22\times(5.5\times10^{-7}\,\mathrm{m})/D$ with $D$ in metres and the answer in radians. Why: pinning $\lambda$ removes one variable and lets resolution be read off from aperture alone across the visible.

Step 2. Convert radians to arcseconds by multiplying by $206265$ : $\theta_{\min}('')=1.22\times5.5\times10^{-7}\times206265/(D/\mathrm{m})$ . Why: $206265=180\times3600/\pi$ is the number of arcseconds in a radian, the standard bridge between physicists' radians and astronomers' arcseconds.

Step 3. Evaluate the constant: $1.22\times5.5\times10^{-7}\times206265=0.138$ , so $\theta_{\min}\approx0.14''/(D/\mathrm{m})$ . A $1\,\mathrm{m}$ mirror reaches $0.14''$ , a $2.4\,\mathrm{m}$ Hubble $0.058''$ , and a $10\,\mathrm{m}$ giant $0.014''=14\,\mathrm{mas}$ . Why: this compact rule lets you judge any visible-light telescope's ideal sharpness in one step.

Step 4. Compare with seeing. Typical ground seeing is $\sim1''$ , which already exceeds the diffraction limit of any aperture larger than $\approx0.14\,\mathrm{m}$ . Why: this single inequality explains the entire enterprise of adaptive optics and space telescopes — for any serious aperture, the atmosphere, not diffraction, is the wall to be broken. *The arithmetic: $1.22\times5.5\times10^{-7}\times206265=0.1384$ ; for HST, $0.138/2.4=0.058''$ ; a $10\,\mathrm{m}$ aperture gives $0.0138''$ .*

Go deeper · Derivation C · 1 minWhy seeing beats diffraction on a big ground telescope

Derivation CWhy seeing beats diffraction on a big ground telescope

The goal is to show quantitatively that a large ground-based mirror in the visible is limited by air, not by its own diameter, and to expose the mechanism — a pupil paved with thousands of independently tilted patches — that makes it so. We set the diffraction scale $\lambda/D$ beside the turbulence scale $\lambda/r_0$ , form their ratio, count the coherence cells, and read off the consequence for how one must fight back.

Step 1. Name the two competing angular scales. Diffraction offers $\theta_{\mathrm{diff}}\sim\lambda/D$ ; turbulence delivers $\theta_{\mathrm{seeing}}\sim\lambda/r_0$ , with $r_0\approx0.1$ – $0.2\,\mathrm{m}$ in the visible. Why: whichever scale is larger wins, because it is the coarser blur that dominates the final image width.

Step 2. Form the ratio: $\theta_{\mathrm{seeing}}/\theta_{\mathrm{diff}}=(\lambda/r_0)/(\lambda/D)=D/r_0$ . For $D=8\,\mathrm{m}$ and $r_0=0.15\,\mathrm{m}$ this is $\approx53$ — the raw image is some fifty times broader than the diffraction limit. Why: the wavelength cancels, leaving a purely geometric penalty that grows linearly with aperture, so bigger mirrors suffer a larger relative loss.

Step 3. Count the coherence cells. The pupil holds $N\sim(D/r_0)^2\approx(53)^2\approx2800$ independent patches, each imaging the star with its own random tilt. Why: the seeing disk is the incoherent sum of $\sim2800$ jittering sub-images, which is why a long exposure smears smoothly to $\lambda/r_0$ rather than $\lambda/D$ .

Step 4. Read off the remedy. To recover $\theta\sim\lambda/D$ one must flatten the wavefront with adaptive optics using of order $N$ actuators, freeze it with millisecond speckle or lucky exposures, or leave the atmosphere entirely for space. Why: each technique attacks the same $N$ cells — correcting them, sampling faster than they change, or removing them — which is why all three appear together in the high-resolution toolkit. *The numbers check out: $D/r_0=8/0.15\approx53$ , giving $N=(53)^2\approx2800$ cells — in line with the actuator counts of real 8 m adaptive-optics systems.*

Limiting cases

The operative resolution depends on which effect dominates and on the wavelength regime:

Case	Result	Meaning
Diffraction-limited ( $D \lesssim r_0$ , or space, or AO-corrected)	$\theta \approx 1.22\,\lambda/D$	Small ground telescopes at long exposure, all space telescopes, and adaptive-optics-corrected apertures. Resolution improves as the aperture grows.
Seeing-limited ( $D \gg r_0$ , uncorrected)	$\theta \approx \lambda/r_0\sim 0.5''\!-\!2''$	Typical large ground-based optical telescopes. Resolution set by the atmosphere, independent of $D$ ; only the collected light grows with aperture.
Interferometric ( $B \gg D$ )	$\theta \approx \lambda/B$	Two or more apertures separated by baseline $B$ ; the array resolves like a single dish of diameter $B$ , as in the VLTI or radio VLBI.
Wavelength scaling of seeing	$r_0 \propto \lambda^{6/5}$	The sky is far steadier in the infrared than the visible, so diffraction-limited imaging and adaptive optics are much easier at longer wavelength.

Common trapA bigger telescope always gives sharper images.

It does not — not from the ground, and not automatically. Below the seeing limit, doubling the aperture does nothing for resolution: it quadruples the collecting area (fainter objects, faster) but the atmosphere still smears every star to the same $\sim1''$ seeing disk, because that width is set by $r_0$ , not by $D$ . Confusing sensitivity with resolution is the single most common error here. Two traps ride alongside it. First, the coefficient $1.22$ is specific to a uniformly illuminated, unobstructed circular pupil judged by the Rayleigh convention; a slit, a central obstruction, or a different criterion (Sparrow, Dawes) all shift the number. Second, the criterion is a convention, not a law — with high signal-to-noise and a known point-spread function, careful fitting can separate sources somewhat below $\theta_{\min}$ , exactly what astrometry and deconvolution exploit.

Worked examples

1 · Diffraction limit of Hubble versus a backyard telescope (fully worked)

Compare the ideal visible-light resolution of the Hubble Space Telescope ( $D=2.4\,\mathrm{m}$ ) with a $1\,\mathrm{m}$ ground telescope, both at $\lambda=550\,\mathrm{nm}$ . Which one actually delivers its diffraction limit?

Step 1. Apply the arcsecond form of the Rayleigh criterion, $\theta_{\min}\approx0.14''/(D/\mathrm{m})$ . For Hubble, $\theta_{\min}=0.138/2.4=\mathbf{0.058''}$ . For the $1\,\mathrm{m}$ scope, $\theta_{\min}=0.138/1.0=\mathbf{0.14''}$ .

Step 2. Check reality. Hubble sits above the atmosphere, so it truly reaches $0.058''$ and resolves individual Cepheids in galaxies tens of Mpc away. The $1\,\mathrm{m}$ scope looks through $\sim1''$ seeing, so its images blur to about $1''$ — roughly seven times coarser than its own $0.14''$ diffraction limit.

Step 3. Draw the lesson: the ground telescope's aperture is wasted on resolution — it gathers light well but is atmosphere-limited. Hubble at 0.058 arcsec beats a same-band 1 m ground telescope not because 2.4 m is much bigger, but because it escaped the seeing floor entirely.

2 · The near-infrared resolution of JWST (fully worked)

Estimate the diffraction-limited resolution of the James Webb Space Telescope, aperture $D=6.5\,\mathrm{m}$ , observing at $\lambda=2\,\mu\mathrm{m}$ in the near-infrared.

Step 1. Work in radians first: $\theta_{\min}=1.22\,\lambda/D=1.22\times(2\times10^{-6}\,\mathrm{m})/(6.5\,\mathrm{m})=3.75\times10^{-7}\,\mathrm{rad}$ .

Step 2. Convert to arcseconds: $\theta_{\min}=3.75\times10^{-7}\times206265=\mathbf{0.077''}\approx\mathbf{77\,mas}$ .

Step 3. Sanity-check against Hubble. Webb's larger, colder mirror and its infrared bandpass give resolution comparable to Hubble's optical figure while reaching far cooler, more distant objects. JWST is diffraction-limited near 0.07 arcsec at 2 μm — the price of the long wavelength is that a 6.5 m mirror resolves no finer than a 2.4 m one would in the visible.

3 · Turbulence cells and the reach of an interferometer

An $8\,\mathrm{m}$ telescope observes in $0.15\,\mathrm{m}$ seeing at $\lambda=550\,\mathrm{nm}$ . How many adaptive-optics actuators does it need, and how does linking two such telescopes on a $130\,\mathrm{m}$ baseline change the resolution?

Step 1. Count coherence cells: $N\sim(D/r_0)^2=(8/0.15)^2\approx\mathbf{2800}$ . The AO system must sense and correct of order this many zones, which is why real 8 m systems carry roughly $10^3$ actuators running at kilohertz rates, aiming to restore the single-dish limit $\theta=0.138/8=\mathbf{0.017''}$ from the raw $\sim1''$ seeing.

Step 2. Now interfere the pair. With baseline $B=130\,\mathrm{m}$ , $\theta\approx\lambda/B=(5.5\times10^{-7})/130\times206265\approx\mathbf{0.9\,mas}$ . The interferometer resolves nearly twenty times finer than either 8 m dish alone — the baseline, not the mirror, now sets the sharpness.

Physics of the portrait

Angular resolution has two founding calculations and one founding phrase. George Biddell Airy did the mathematics in 1835, working out that a circular aperture spreads a star into the ringed disk that bears his name. But the idea that fixed the number astronomers still quote — the moment two stars stop being two — belongs to John William Strutt, the third Baron Rayleigh, who in his 1879 study of the resolving power of gratings and telescopes set down the convention that one Airy peak must sit on the neighbour's first dark ring. That single stipulation turned a diffraction pattern into an engineering specification, and $1.22$ has been on every telescope datasheet since.

Origins · 1 minLord Rayleigh and how this was discovered

Lord Rayleigh

1842–1919 · diffraction theory and the resolving power of optical instruments

John William Strutt, third Baron Rayleigh, was the last of the great classical physicists to hold nearly the whole of wave physics in one mind. His Theory of Sound (1877–78) is still read, and the scattering law that carries his name — intensity $\propto\lambda^{-4}$ — is why the sky is blue. In an 1879 paper on the resolving power of diffraction gratings and telescopes he introduced the criterion that made resolution quantitative: two point sources are just separable when the principal maximum of one diffraction pattern coincides with the first minimum of the other. It was a deliberately conservative, eye-calibrated convention, and Rayleigh knew it was one — he remarked that a keen observer with a clean pattern could do somewhat better, anticipating the modern practice of fitting below the nominal limit. In 1904 he won the Nobel Prize for the discovery of argon, made after noticing a stubborn discrepancy in the density of nitrogen — the same habit of taking small numerical residuals seriously that underlies the science of pushing telescopes to their limit.

History

1835

— Airy derives the diffraction pattern of a circular aperture, the Airy disk

1879

— Rayleigh states the resolving-power criterion for telescopes and gratings

1953

— Babcock proposes adaptive optics to undo atmospheric distortion in real time

1966

— Fried defines the atmospheric coherence length r₀ that governs seeing

2019

— the Event Horizon Telescope resolves a black-hole shadow at about 20 μas

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 6 (Telescopes) · Hecht, Optics, 5th ed., Ch. 10 (Diffraction) · Binney and Tremaine, Galactic Dynamics, 2nd ed., observational appendix · Ryden, Introduction to Cosmology, 2nd ed. · Roddier, Adaptive Optics in Astronomy (Fried statistics and AO design). Key numerical anchors: the first zero of $J_1$ is $3.8317$ , giving coefficient $3.8317/\pi=1.22$ and a just-resolved dip to $0.735$ of the peak; $1.22\times5.5\times10^{-7}\times206265=0.138''$ . The Fried scalings $r_0\propto\lambda^{6/5}(\cos z)^{3/5}$ and $N\sim(D/r_0)^2$ originate with Fried (1966), J. Opt. Soc. Am. 56, 1372, with $r_0\sim10$ – $20\,\mathrm{cm}$ and seeing $0.5''$ – $2''$ representative of good sites. Instrument figures — HST $0.05''$ , JWST $77\,\mathrm{mas}$ at $2\,\mu\mathrm{m}$ , VLTI $130\,\mathrm{m}$ baselines, EHT $\sim20\,\mu\mathrm{as}$ at $1.3\,\mathrm{mm}$ — are published mission and array values.

Prerequisites. Luminosity and Radiant Flux · The Celestial Sphere · The Magnitude System. Leads to. Stellar Parallax and the Parsec · The Cosmic Distance Ladder · Photometric Systems and Filters.

Ask the tutor: Ask me to work through the diffraction limit of any telescope you name, or to show exactly how adaptive optics, interferometry, or a space mission would beat the atmosphere for your target.

Is angular resolution the same thing as light-gathering power?

No, and conflating them is the classic beginner's error. Light-gathering power scales with collecting area, $\propto D^2$ , and decides how faint an object you can detect and how fast. Angular resolution is about separating close features and scales as $\lambda/D$ . A bigger mirror always gathers more light, but resolves finer detail only if the wavefront stays coherent across the aperture. On the ground in the visible the atmosphere breaks that coherence, so a large telescope's extra aperture buys sensitivity while seeing pins its resolution near $1''$ . The two improve together only when you are truly diffraction-limited — in space, or under adaptive-optics correction.

Where does the specific number 1.22 come from?

It is $3.8317/\pi$ , and $3.8317$ is the first positive zero of the Bessel function $J_1$ — the function that describes the diffraction pattern of a circular aperture. Solving for the angle at which the Airy pattern first goes dark, the geometry contributes a $\pi$ and the Bessel root contributes $3.8317$ , and their ratio is $1.2197$ . The factor is thus a pure consequence of the pupil being a circle: a single slit of width $D$ gives $1.00\,\lambda/D$ to its first null instead, and a central obstruction or non-uniform illumination gives yet another coefficient. So $1.22$ is specific to a clear, uniformly lit circular aperture and the Rayleigh convention, not a universal constant.

What exactly is the Fried parameter, and why does it matter more than my telescope?

The Fried parameter $r_0$ is the diameter of the largest patch of the incoming wavefront that the atmosphere leaves essentially flat — the coherence length of the turbulence, only $10$ – $20\,\mathrm{cm}$ in visible light at a good site. Because long-exposure seeing scales as $\lambda/r_0$ rather than $\lambda/D$ , once your aperture exceeds $r_0$ the atmosphere, not your mirror, sets the image sharpness — which is why an $8\,\mathrm{m}$ telescope and a $20\,\mathrm{cm}$ one deliver similar raw resolution from the same site. It also fixes how hard adaptive correction is: the pupil holds $(D/r_0)^2$ independent cells, each a zone the deformable mirror must sense and flatten.

Why is it so much easier to reach the diffraction limit in the infrared?

Two effects push the same way. First, the diffraction limit itself, $1.22\,\lambda/D$ , is coarser at longer wavelength, so there is less sharpness to recover. Second, and more importantly, the Fried parameter grows as $r_0\propto\lambda^{6/5}$ , so the atmosphere presents far fewer and larger coherence cells in the infrared. Going from $0.5\,\mu\mathrm{m}$ to $2\,\mu\mathrm{m}$ (a factor of 4 in $\lambda$ ) enlarges $r_0$ by $4^{6/5}\approx5.3$ , cutting the number of cells $(D/r_0)^2$ — and the required actuators and correction speed — by roughly thirty. This is why the first adaptive-optics systems, and most routine diffraction-limited ground imaging, work in the near-infrared rather than the visible.

How can an interferometer resolve finer than either of its telescopes?

Because resolution is set by the largest separation over which the instrument samples the wavefront, not by how much light it collects. Two apertures a distance $B$ apart measure the interference fringes of the incoming light, and the fringe spacing encodes structure on the angular scale $\lambda/B$ — exactly as a single mirror of diameter $B$ would. What the interferometer sacrifices is filling: it samples only certain baselines, so arrays rotate with the Earth and combine many baselines to synthesize an image, achieving the resolution of an enormous virtual aperture while collecting only the light of the real dishes.

Does the Rayleigh criterion mean you can never separate closer sources?

No — it is a convention, not an impassable wall. Rayleigh chose the peak-on-first-null spacing because it produces a visible dip in the combined profile for a human eye on a noisy pattern, but it is deliberately conservative. With high signal-to-noise, a well-characterized point-spread function, and model fitting, one can locate and separate two sources somewhat below $\theta_{\min}$ ; astrometry routinely centroids a single star to a small fraction of the Airy disk, and the Sparrow and Dawes limits push closer. What you cannot do is create information the aperture never captured: structure far below $\lambda/D$ is genuinely lost, because those high spatial frequencies never entered the pupil at all. This is also why space telescopes still matter despite adaptive optics — AO leaves a halo of residual uncorrected light (a Strehl ratio below one) over a small field, whereas space delivers stable, wide-field, diffraction-limited imaging with no turbulence.