Refractive Index and Optical Path Length

The idea that optics measures length in wavelengths is the working principle of the most precise instruments ever built. Insert a gas cell, a thin film, or a tissue sample into one arm of a Michelson or Mach–Zehnder interferometer and it shifts the fringe pattern by $\Delta\phi = 2\pi(n-1)t/\lambda_0$; counting fringes yields the index or the thickness to parts in $10^{6}$ or better — the basis of gas refractometers and film-thickness monitors. Optical coherence tomography images the retina by ranging each layer by its optical thickness $n\,t$, and phase-contrast microscopy (Zernike, Nobel 1953) makes transparent cells visible by turning the OPL differences they impose into brightness. Push the idea to its limit and you reach the LIGO gravitational-wave detectors, whose 4 km arms are read as optical paths so stable that a length change of $10^{-18}$ m shows up as a phase shift. The same slab bookkeeping runs in reverse in a wave-plate or spatial light modulator, where a controlled $(n-1)t$ is added on purpose to sculpt a beam's phase front. And the frequency-dependence quietly held constant here — $n(\nu)$ — is the doorway to the next article: dispersion, the reason a prism has colours at all.

Picture a rank of soldiers marching in step onto a muddy field: the boots slow, and if you watch just the spacing of the ranks, they bunch up. The cadence — how often a boot strikes — is set by the drummer and never changes, so when the ranks slow, they cannot spread out; they pack tighter. That bunching is the entire content of $\lambda = \lambda_0/n$. The drummer is the source frequency $\nu$, which doesn't change at the boundary; the mud is the equivalent of a higher index; the crowded ranks are the shortened wavelength.

The picture leaves three habits worth keeping. First, colour lives in the frequency, not the wavelength. Inside glass the wavelength is smaller, yet the light is the same red or green it always was, because the eye and every atom respond to $\nu$, which the boundary conserves. Second, the medium slows the wave by re-radiating it — the passing field shakes the bound electrons, they re-emit a delayed wavelet, and the superposition crawls, which is exactly why $n$ ends up depending on frequency (dispersion). Third, and most useful, distance in optics is measured in wavelengths: a millimetre of glass is not one millimetre of "light-journey" but $n$ millimetres' worth of crests, and that discrepancy is a quantity you can put on a graph. Put it on a graph, and the punchline appears. Take two rays of identical geometric length — same start, same end, same ruler distance — but send one straight through air and route the other through a glass slab of index $n$ and thickness $t$. They arrive having counted different numbers of wave crests: the slab ray is ahead by exactly $(n-1)t$ of optical path, or $(n-1)t/\lambda_0$ extra wavelengths. That single expression, linear in the thickness, is the whole working principle of a phase plate and of every interferometer that measures an index by counting fringes.

The blue line is the whole story of the optical path in one straight stroke: double the glass and you double the extra crests, because $(n-1)t$ is linear in $t$. The red dot is the number a real experiment sees — a single millimetre of crown glass shifts the wave by nearly a thousand full cycles, a shift no ruler could ever read but that an interferometer resolves to a fraction of one fringe. That is why optics measures length in wavelengths: the glass is optically "longer" than it is physically, and the difference is where all the useful information hides.

Everything begins with a definition that turns a speed into a pure number. The wave travels at a phase velocity $v$ inside the medium, always less than the vacuum speed $c$ for a transparent dielectric, and the ratio of the two is the index:

where: $n$ — refractive index (dimensionless); $c = 2.998\times10^{8}\ \mathrm{m/s}$ — vacuum speed of light; $v$ — phase velocity of the wave in the medium. Read it as a slowdown factor: $n = 1.5$ means the crests advance at two-thirds of $c$. At the microscopic level $n = \sqrt{\varepsilon_r\,\mu_r}\approx\sqrt{\varepsilon_r}$ ties the index to the material's response to the optical field, but for geometrical optics the one number $n$ is all we need. Let us consider light crossing a boundary. The incident and transmitted waves must keep their phases matched at the interface at all times, and matching the factor $e^{-i2\pi\nu t}$ on both sides forces the same $\nu$ in each medium. With the frequency locked and the speed cut to $c/n$, the relation $v = \nu\lambda$ has no choice but to shrink the wavelength:

where: $\nu$ — temporal frequency (Hz), fixed by the source; $\lambda_0$ — vacuum wavelength; $\lambda$ — wavelength inside the medium. The physical reading is blunt: colour is frequency and it is conserved, while wavelength and speed both fall by the factor $n$. The spatial oscillation rate rises to $k = 2\pi/\lambda = n k_0$, where $k_0 = 2\pi/\lambda_0$ is the vacuum wavenumber — the wave packs in $n$ times as many crests per metre. Because phase accrues at $k = n k_0$ per metre, the phase gathered along a ray weights each geometric step by the local index. That weighting defines the optical path length — the quantity that optics actually cares about and not the ruler distance:

where: $n(s)$ — local index along the ray; $\mathrm{d}s$ — element of geometric arc length. The OPL is the distance light would have covered in vacuum during the same time that the beam actually took, so $\mathrm{OPL} = c\,t$ with $t = \int \mathrm{d}s/v$ the true travel time. Since one wavelength of vacuum path is $\lambda_0$, the OPL is exactly $\lambda_0$ times the number of crests laid down along the path — metres converted into a wavelength count. That count is the phase, and the conversion is a single multiplication by the vacuum wavenumber:

where: $\phi$ — total phase accumulated along the ray; $k_0 = 2\pi/\lambda_0$. Two rays with equal OPL arrive with equal phase and interfere constructively, no matter how different their geometric routes — the fact that makes interferometry possible. Reduce the whole of ray optics to a statement about this quantity and you have Fermat's principle: the physical ray is the one whose OPL is stationary, $\delta\!\int n\,\mathrm{d}s = 0$, from which reflection, Snell's law, and total internal reflection all follow (their own articles). Here we take $(3)$–$(4)$ and push on the simplest case: a flat slab.

1 · Extra wavelengths through a glass slab (confidence builder)

A 1.0 mm plate of crown glass ($n = 1.52$) is placed in the path of a $\lambda_0 = 550\ \mathrm{nm}$ green beam. How many extra wavelengths of optical path does it add relative to the same 1.0 mm of air, and what phase shift is that?

Step 1. The excess optical path of a slab over vacuum is $\Delta = (n-1)t$. Numerically $\Delta = (1.52-1)(1.0\times10^{-3}\,\mathrm{m}) = 0.52\times10^{-3}\,\mathrm{m} = 0.52\ \mathrm{mm}$ of vacuum-equivalent path — the glass is optically half a millimetre "longer" than it is thick.

Step 2. Convert that excess to a wave count: $N = \Delta/\lambda_0 = (n-1)t/\lambda_0 = (0.52\times10^{-3})/(550\times10^{-9}) = 945.45$, so the slab inserts about 945 extra wavelengths. (The exact value is $945.45$, which rounds to 945, not 946 — a distinction that matters when an interferometer resolves fractions of a crest.)

Step 3. The phase shift is $2\pi$ per extra wavelength: $\Delta\phi = 2\pi N = 2\pi(945.45) = \boxed{5.94\times10^{3}\ \mathrm{rad}}$. In a Michelson interferometer this single slab sweeps the pattern through about 945 fringes on one pass — count them and you have measured $n$ from a single millimetre.

Step 4. Sanity check by counting crests directly: inside the glass $\lambda = \lambda_0/n = 362\ \mathrm{nm}$, so the 1.0 mm slab holds $1000/0.362 \approx 2762$ crests versus $1000/0.550 \approx 1818$ in the same length of air; the difference $\approx 944$ matches $N$ to rounding. The extra crests are real and countable.

2 · Foucault's water test and a laser under water (exam level)

In 1850, Foucault compared the speed of light in air and in water. By what factor is light slower in water ($n = 1.333$) than in air, and what is the wavelength of a 633 nm HeNe laser once it is inside the water?

Step 1. Speeds go as $v = c/n$, so the ratio is set purely by the indices: $v_{\text{air}}/v_{\text{water}} = (c/n_{\text{air}})/(c/n_{\text{water}}) = n_{\text{water}}/n_{\text{air}}$. With $n_{\text{air}} = 1.0003$, this is $1.333/1.0003 = \boxed{1.33}$ — light in water runs at about three-quarters of its air speed.

Step 2. Put in the absolute number: $v_{\text{water}} = c/n_{\text{water}} = (2.998\times10^{8})/1.333 = \boxed{2.25\times10^{8}\ \mathrm{m/s}}$. This is the result that settled a two-century long argument: Newton's corpuscular theory required light to speed up in the denser medium, the wave theory required it to slow down, and Foucault observed it slowing.

Step 3. The frequency is fixed, so the wavelength compresses by $n$: $\lambda_{\text{water}} = \lambda_0/n = 633/1.333 = \boxed{475\ \mathrm{nm}}$. The beam that is red in air is still red under water — its frequency $\nu = c/\lambda_0 = 4.74\times10^{14}\ \mathrm{Hz}$ is unchanged — even though its in-water wavelength (475 nm) is one a diver would associate with blue-green in vacuum.

Step 4. The two examples together carry the whole lesson: a medium acts on light through one number $n$, which slows the wave ($v = c/n$, Example 2), shortens its wavelength ($\lambda = \lambda_0/n$, both), and thereby loads extra optical path onto any slab it passes through ($(n-1)t$, Example 1) — while never touching the frequency that is the light's true colour.

The refractive index is defined as $n = c/v$, but that definition is only as meaningful as our ability to measure $v$ inside matter — and for two centuries the sign of the effect, whether light speeds up or slows down in glass and water, was the single most consequential unmeasured number in optics.

Texts. Hecht, Optics, 5th ed., ch. 4 (The Propagation of Light) · Saleh & Teich, Fundamentals of Photonics, 3rd ed., ch. 1 (Ray Optics) and ch. 5 (Resonant and Dispersive Media) · Foot, Atomic Physics, ch. 7 (index and atomic polarizability) · Loudon, The Quantum Theory of Light (light–matter response) · Demtröder, Laser Spectroscopy (dispersion and the index near resonance). Key results to keep in mind: $n = c/v$ and $\lambda = \lambda_0/n$ follow from the fixed source frequency; the optical path length is $\mathrm{OPL} = \int n\,\mathrm{d}s$, the accumulated phase is $\phi = k_0\,\mathrm{OPL}$, and a slab inserts an excess $(n-1)t$ with phase $2\pi(n-1)t/\lambda_0$. The crown-glass worked number is $(0.52\times10^{-3})/(550\times10^{-9}) = 945.45 \approx 945$ waves, a phase shift $\approx 5.9\times10^{3}$ rad. Foucault's speed ratio is $v_{\text{air}}/v_{\text{water}} = n_{\text{water}} \approx 1.33$, and a 633 nm HeNe laser has a wavelength of 475 nm in water. Standard index values near 589 nm: air 1.0003, water 1.333, fused silica 1.458, crown glass $\approx1.52$, diamond 2.417. Note: the fused-silica fringe count is 778 ($777.6$ for a 1.0 mm silica slab at 589 nm), not 777.

Prerequisites. Ray Model of Light · Wave frequency and wavelength. Leads to. Fermat's Principle of Least Time · Snell's Law of Refraction · Total Internal Reflection and Critical Angle · Dispersion and Prisms.