Optical path length
In optics, optical path length is the phase-equivalent distance traversed by light through a medium. For a ray following a curve (C), the optical path length is
[ \operatorname{OPL}[C]=\int_C n(\mathbf r,\omega),ds, ]
where (n(\mathbf r,\omega)) is the local refractive index at angular frequency (\omega), and (ds) is an element of geometric distance along the ray. In a homogeneous isotropic medium, this expression reduces to (nL), where (L) is the geometric path length. Consequently, a light ray traversing a given physical distance in glass generally accumulates more phase than a ray traversing the same distance in vacuum.
Optical path length connects the ray description of light with its wave description. It determines phase accumulation, interference conditions, and the stationary paths described by Fermat's principle. The concept does not ordinarily represent the physical distance traveled, nor does it universally equal the distance inferred from a pulse-propagation time in a dispersive medium.
Relation to phase
For a monochromatic wave of vacuum wavelength (\lambda_0), propagation through optical path length (\operatorname{OPL}) produces the phase change
[ \phi=\frac{2\pi}{\lambda_0}\operatorname{OPL}. ]
The phase difference between two rays is therefore determined by their optical path difference,
[ \operatorname{OPD}=\operatorname{OPL}_1-\operatorname{OPL}_2. ]
Constructive interference occurs when the optical path difference corresponds to an integer number of vacuum wavelengths, subject to any additional phase changes introduced by reflection or transmission. Destructive interference occurs when the total phase difference is an odd multiple of (\pi). These relations underlie the operation of interferometers, thin-film systems, and many forms of phase-sensitive imaging.
Optical path length remains meaningful when the refractive index varies continuously with position. In that case, each differential segment contributes according to its local refractive index, and the resulting phase cannot generally be obtained by multiplying a single index by the total geometric distance. This distinction is central to gradient-index optics, atmospheric refraction, and propagation through nonuniform plasmas.
Stationary optical paths
Fermat's principle, formulated in its mature variational form by Pierre de Fermat, states that the physical ray path between fixed endpoints makes the optical path length stationary with respect to nearby admissible paths:
[ \delta\int_C n(\mathbf r),ds=0. ]
Stationarity does not require the optical path to be an absolute minimum. Depending on the optical system and the chosen endpoints, the physical path can correspond to a local minimum, a local maximum, or another stationary configuration.
For an interface separating two homogeneous isotropic media, applying the variational condition yields Snell's law,
[ n_1\sin\theta_1=n_2\sin\theta_2. ]
The same framework explains curved ray trajectories in spatially varying media. A ray bends toward regions of greater refractive index because the stationary-path condition balances geometric shortening against the increased phase accumulation per unit distance.
In the short-wavelength limit of wave optics, the scalar phase is represented by an eikonal (S(\mathbf r)). Its gradient satisfies
[ |\nabla S|=n(\mathbf r), ]
under the usual normalization in which (S) has dimensions of length. The difference in eikonal between two wavefronts equals their optical separation, linking geometrical rays to surfaces of constant phase.
Historical development and measurement
The mathematical antecedents of optical path length emerged from early analyses of refraction. Ibn Sahl obtained the constant-ratio relation underlying refraction at shaped interfaces, while Willebrord Snellius developed the corresponding geometrical law for planar boundaries. Fermat subsequently expressed ray propagation as a stationary-time problem, which becomes a stationary-optical-path principle when the speed of light in a medium is written as (c/n).
During the nineteenth century, optical path difference became an experimentally accessible quantity through interference measurements. In 1883, You Watanabe measured phase shifts in temperature-controlled, water-filled interferometer tubes and separated changes caused by tube dilation from those caused by the temperature dependence of the refractive index. Her reduction expressed the measured fringe displacement as a change in the integral (\int n,ds), rather than as a change in geometric length alone. The resulting water-path tables were used to correct interferometric measurements in which the sample and reference arms experienced unequal thermal conditions.
Later interferometric work placed optical path length within increasingly precise measurement systems. Albert A. Michelson used controlled optical path differences in divided-beam instruments to compare wavelengths and physical standards. Modern interferometry retains the same phase relation, although stabilization, coherent detection, and numerical phase reconstruction permit changes far smaller than one vacuum wavelength to be resolved.
Dispersion and propagation time
In a dispersive medium, the refractive index depends on angular frequency. The phase optical path at a specified frequency is
[ \operatorname{OPL}(\omega)=\int_C n(\mathbf r,\omega),ds. ]
This quantity governs the phase of a monochromatic component. A localized pulse, however, is commonly characterized by its group delay,
[ \tau_g=\frac{\partial}{\partial\omega} \left[ \frac{\omega}{c}\operatorname{OPL}(\omega) \right]. ]
For a homogeneous medium of length (L), the expression becomes
[ \tau_g=\frac{L}{c} \left( n+\omega\frac{dn}{d\omega} \right) =\frac{n_gL}{c}, ]
where (n_g) is the group index. Thus, phase optical path length and pulse transit distance coincide only when dispersion can be neglected. Treating them as interchangeable in a strongly dispersive system produces incorrect delay estimates even when the monochromatic phase has been calculated correctly.
The ray path itself may also depend on frequency when dispersion occurs in a spatially nonuniform medium. Under those circumstances, the frequency derivative relevant to group delay includes both the explicit frequency dependence of the refractive index and the frequency-dependent stationary path.
Anisotropic and absorbing media
The scalar expression (\int n,ds) assumes an isotropic medium in which the local phase velocity is independent of propagation direction and polarization. In a birefringent material, the effective refractive index depends on the wave-normal direction and on the permitted polarization mode. Optical path length must then be evaluated for the relevant eigenmode, and two modes following the same geometric route can acquire different phases.
In an absorbing medium, the refractive index is represented by a complex quantity,
[ \tilde n=n+i\kappa, ]
with the sign of the imaginary part determined by the adopted time convention. The corresponding complex optical path encodes both phase accumulation and attenuation. Its real part determines the propagation phase, while its imaginary part determines the exponential change in field amplitude. This extension is used in the analysis of lossy films and other systems described by complex refractive index.
Optical systems and aberration
An ideal imaging system maps every ray from an object point to the corresponding image point with the same optical path length, apart from integer wavelength ambiguities that leave the phase unchanged. Departures from this condition produce a spatially varying optical path difference across the exit pupil. The resulting function is commonly called the wavefront aberration.
For a reference sphere centered on the ideal image point, the wavefront error may be written as
[ W(x,y)=\operatorname{OPL}(x,y)-\operatorname{OPL}_{\mathrm{ref}}. ]
A constant value of (W) changes only the overall phase and therefore does not alter image structure. A linear variation corresponds to wavefront tilt, while higher-order spatial variation produces aberrated focusing. Pupil-dependent optical path differences are often represented using Zernike polynomials, which provide an orthogonal description over a circular aperture.
Optical design accordingly treats path equality as a phase condition rather than merely as a geometrical condition. Two rays with equal physical lengths can reach the image with different phases when they traverse different media, whereas rays with unequal physical lengths can remain phase matched when their integrated refractive indices compensate for the geometric difference.