Reflection & Refraction · Guide · glidepathscience.org
GuideReflection & Refraction
Contents
This Guide covers the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round. Read Part 1 before the Intro track and Part 2 before the Advanced track, about twelve minutes each. Indices here are relative to air, with air at 1.000, as index tables give them.
Part 1 · The essentials
Light as rays, and n = c/v
In one uniform material light travels in straight lines, so we can follow it as rays, arrows along its path. It travels slower in matter than in vacuum, and the index of refraction says how much slower:
$n = c/v$
with $c$ = 2.998 × 10⁸ m/s. Water’s index is 1.333, crown glass’s 1.517, diamond’s 2.417. At a boundary the frequency cannot change: it is set by the source, and the color you see goes with it. So when the speed drops by $n$, the wavelength drops by $n$ too: $\lambda_n = \lambda/n$.
“Optically denser” means a higher index, not a greater mass density: cooking oil floats on water but has the higher index.
Reflection and mirror images
Every angle in optics is measured from the normal, the line perpendicular to the surface where the ray meets it. The law of reflection:
$\theta_r = \theta_i$
A protractor laid along the mirror reads the complement, 90° minus the angle from the normal; that is the most common slip.
Light from each point of an object spreads out, and after a plane mirror the reflected rays spread as if from a point behind it. Traced backward, they meet as far behind the mirror as the object is in front, on the line through the object perpendicular to the mirror. That point is the image. It is virtual: no light is there. Every observer sees it in the same place.
To see yourself head to toe, the ray from your toes must reflect halfway up to your eyes, and the ray from the top of your head halfway between head and eyes. The mirror needs to be half your height, whatever the distance.
Snell’s law
Light that meets a boundary at an angle changes direction: it refracts. Think of the wavefronts, the crests, crossing the boundary. The end of a front that reaches the slower medium first slows first, and the front swings round, like a marching band stepping onto mud at an angle.
$n_1\sin\theta_1 = n_2\sin\theta_2$
Here $n_1$ is the index on the side the light comes from. Into a higher index the ray bends toward the normal; into a lower one, away from it. Along the normal it does not bend at all. Light paths are reversible, so the same law runs backward. That is why a fish looks higher than it is: the ray leaving the water bends away from the normal, and your eye traces it straight back. Not all the light goes on: every boundary also reflects some, about 4 % for glass in air head-on and more at grazing angles.
Total internal reflection and fibers
Heading into a lower index, light bends away from the normal, and at one angle the refracted ray skims along the surface, $\theta_2$ = 90°. That is the critical angle:
$\theta_c = \sin^{-1}(n_2/n_1)$
Beyond it no light gets out: all of it reflects, more completely than from any metal mirror. This only happens for $n_1 > n_2$. Light entering a denser medium always gets in, even at grazing incidence.
A light pipe or fiber guides light by bouncing it off its walls beyond $\theta_c$. A ray at angle α to the axis meets the wall at 90° − α from the wall’s normal, so the steepest guided ray makes $90^\circ - \theta_c$ with the axis. A tight bend tips the wall toward the rays until they meet it inside $\theta_c$, and light leaks. A drop of water on the rod raises $\theta_c$ where it touches; a cladding of lower index keeps the guiding surface clean.
Dispersion and the rainbow
The index depends on the wavelength: in glass and water it is higher for violet than for red. Crown glass reads 1.514 for red (656 nm) and 1.530 for violet (405 nm). So a prism bends violet most and fans white light into a spectrum: dispersion. Newton passed one color through a second prism and it stayed one color: prisms separate the colors already in white light.
A rainbow is sunlight refracted into raindrops, reflected once inside and refracted out. The light piles up at one angle from the antisolar point, the shadow of your head: 42.5° for red and 40.6° for violet. So the red drops sit higher in the sky, and each drop sends your eye one color. The antisolar point is as far below the horizon as the Sun is above it, so the top of the bow is 42.5° minus the Sun’s elevation above the horizon. A second reflection inside makes the fainter secondary bow, near 51°, with its colors reversed.
Part 2 · Going further
Fermat and the vector law
Fermat’s principle: light travels between two points along the path whose travel time, $t = (n_1L_1 + n_2L_2)/c$, is stationary, here a minimum. Setting $dt/dx = 0$ for the crossing point gives $n_1\sin\theta_1 = n_2\sin\theta_2$, and the law of reflection follows the same way. Near the best path the time hardly changes, which is what “stationary” means.
Ray tracers never compute angles. With unit vectors $\hat{d}$ along the ray and $\hat{n}$ the normal facing it, and $\eta = n_1/n_2$:
$\hat{t} = \eta\hat{d} + (\eta\cos\theta_1 - \cos\theta_2)\hat{n}$, with $\cos\theta_1 = -\hat{n}\cdot\hat{d}$
How much reflects, and the evanescent wave
Fresnel’s equations divide the light at a surface between the reflected and transmitted rays, $R + T = 1$. For unpolarized light at normal incidence:
$R = \left((n_1-n_2)/(n_1+n_2)\right)^2$
About 4 % for glass in air, far less between similar indices. $R$ rises to 100 % at grazing incidence; inside a denser medium it climbs smoothly and steeply to 100 % at $\theta_c$, with no jump. How the reflection splits into s and p light, and Brewster’s angle, where p light does not reflect at all, are in the Polarization Lab.
Beyond $\theta_c$ the field still reaches past the surface, decaying as $e^{-z/d_p}$:
$d_p = \lambda/\left(2\pi\sqrt{n_1^2\sin^2\theta_1 - n_2^2}\right)$
This evanescent wave carries no light across on its own; a second surface within a few $d_p$ lets some tunnel through (frustrated total internal reflection, wave optics).
Dispersion formulas and the spectrometer
Away from absorption, Cauchy’s formula describes dispersion: $n = A + B/\lambda^2$. From two indices,
$B = (n_F - n_C)/(1/\lambda_F^2 - 1/\lambda_C^2)$
with λ in µm (F at 486.1 nm, C at 656.3 nm). It fails near absorption bands. The Abbe number, $V_d = (n_d - 1)/(n_F - n_C)$, sums up dispersion in one number: crown about 64, flint about 36.
A prism’s deviation is least when the light crosses it symmetrically, parallel to the base. Then
$n = \sin\left((A + D_{min})/2\right)/\sin(A/2)$, $D_{min} = 2\sin^{-1}\left(n\sin(A/2)\right) - A$
which is why the spectrometer measures indices so well: a 0.1° error in $D_{min}$ moves $n$ by only about 0.001. Each color has its own $D_{min}$; their spread is the prism’s angular dispersion.
Descartes’s rainbow
A ray entering a drop at angle $i$, refracted to $r$ and reflected $k$ times inside, is deviated by $D = 180^\circ k + 2i - 2(k+1)r$. Rays from a band of impact parameters near the stationary point, $dD/di = 0$, leave together, a caustic, and that is the bow:
$\cos^2 i = (n^2-1)/(k^2+2k)$
The bow lies $|180^\circ - D|$ from the antisolar point. Once-reflected light fills the sky inside the primary and stops at it; twice-reflected light starts at the secondary and fills the sky outside. Neither reaches the band between, Alexander’s dark band. The secondary’s colors are reversed because its deviation passes 180°. Supernumerary bows, just inside the primary, are interference, beyond the ray model.
Fibers and graded index
In a step-index fiber the steepest guided ray travels farther than the axial one. Over a length $L$ the two arrive
$\Delta t = (L n_1/c)(n_1/n_2 - 1)$
apart: modal dispersion, which blurs pulses and limits the data rate. The acceptance half-angle in air is $\sin^{-1}\mathrm{NA}$, with $\mathrm{NA} = \sqrt{n_1^2 - n_2^2}$. A graded-index core, its index falling from the axis, curves rays back toward it, and the steep rays’ longer paths run through faster glass: the spread drops more than a hundredfold. Material dispersion, from the index’s change with wavelength, adds its own spread; single-mode fibers have no modal spread at all.
Where $n$ changes smoothly, a ray curves toward higher index, and in layers $n\cos\alpha$ stays constant along it. Air’s $n - 1 \propto 1/T$, so air over a hot road has a lower index, and a shallow ray turns back up before reaching the road if $\cos\alpha = n_{road}/n_{eye}$ allows. The eye sees sky there and reads it as water: a mirage, gradual bending, not a reflection.
Reference
Common mistakes
- Measuring angles from the surface instead of the normal.
- Putting the image on the mirror’s surface, or having it move when you move.
- Thinking that stepping back from a mirror shows more of you.
- Bending light away from the normal as it enters glass or water.
- Expecting total internal reflection going into the denser medium, or using sin⁻¹(1/n₁) when the outside is not air.
- Giving the critical angle when the question asks for the angle to a fiber’s axis.
- Thinking the color or frequency changes in water; only the speed and wavelength do.
- Red bending most, or prisms adding colors.
- Calling a rainbow a reflection, or expecting it at 42° from the Sun.
- Mixing $n_1$ and $n_2$ in Fresnel’s and Snell’s formulas: $n_1$ is always the side the light comes from.
Equation sheet
| Part 1 | |
|---|---|
| reflection | $\theta_r = \theta_i$ |
| index | $n = c/v$, $\lambda_n = \lambda/n$ |
| Snell | $n_1\sin\theta_1 = n_2\sin\theta_2$ |
| critical | $\theta_c = \sin^{-1}(n_2/n_1)$, for $n_1 > n_2$ |
| fiber | steepest ray $90^\circ - \theta_c$ to the axis |
| mirror | image as far behind as the object in front; full length: half your height |
| bow top | 42.5° − the Sun’s elevation |
| Part 2 | |
|---|---|
| Fermat | $t = (n_1L_1 + n_2L_2)/c$ stationary |
| vectors | $\hat{t} = \eta\hat{d} + (\eta\cos\theta_1 - \cos\theta_2)\hat{n}$ |
| head-on | $R = \left((n_1-n_2)/(n_1+n_2)\right)^2$ |
| depth | $d_p = \lambda/\left(2\pi\sqrt{n_1^2\sin^2\theta_1 - n_2^2}\right)$ |
| Cauchy | $n = A + B/\lambda^2$ |
| $B = (n_F - n_C)/(1/\lambda_F^2 - 1/\lambda_C^2)$ | |
| prism | $n = \sin\left((A + D_{min})/2\right)/\sin(A/2)$ |
| $D_{min} = 2\sin^{-1}\left(n\sin(A/2)\right) - A$ | |
| bows | $\cos^2 i = (n^2-1)/(k^2+2k)$ |
| $D = 180^\circ k + 2i - 2(k+1)r$ | |
| modal | $\Delta t = (L n_1/c)(n_1/n_2 - 1)$ |
| mirage | $\cos\alpha = n_{road}/n_{eye}$, $n - 1 \propto 1/T$ |
Symbols: $n$ index of refraction · $c$ speed of light in vacuum · $v$ speed in the medium · λ wavelength · $\theta_1$, $\theta_2$ angles of incidence and refraction, $\theta_i$, $\theta_r$ of incidence and reflection, $\theta_c$ critical angle, all from the normal · R reflected fraction · $d_p$ penetration depth · A, B Cauchy’s coefficients, or A a prism’s apex angle · D deviation · $i$, $r$ angles in a drop, $k$ reflections inside · NA numerical aperture · L length · α angle from the horizontal · T absolute temperature.
Further reading
Samuel J. Ling, Jeff Sanny and William Moebs, University Physics Volume 3 (OpenStax, 2016), free online under CC BY-NC-SA 4.0: Chapter 1, The Nature of Light, and Section 2.1, Images Formed by Plane Mirrors; the Intro track follows them. E. Hecht, Optics, 5th ed. (Pearson, 2017). H. M. Nussenzveig, “The theory of the rainbow,” Scientific American 236(4), 116 (1977). And this lab’s Field Manual (H): how the model works, and what it leaves out.
Reflection & Refraction v1.2 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org