Length is a measure of spatial extent. It may describe the separation between two points, the extent of an object along a specified direction, or the distance traced along a curve. In mathematics, length is defined through structures such as distances, norms, and curve integrals; in physics, it is a measurable quantity whose value depends on a specified procedure and, in relativistic settings, a specified reference frame. Everyday distinctions among length, width, height, depth, and thickness identify different directions or features rather than different kinds of physical quantity.
Length is one of the seven base quantities of the International System of Units (SI). Its SI unit, the metre, is defined by fixing the numerical value of the speed of light in vacuum. This connects length measurement to time and provides a common standard for activities ranging from manufacturing and construction to atomic physics and astronomy.
The apparent simplicity of measuring an extent conceals several distinct questions: which endpoints or boundaries define it, whether the relevant path is straight or curved, how an instrument realizes a unit, and how motion, material deformation, or spatial curvature affects the result. These distinctions become especially important for irregular coastlines, flexible molecules, precision components, moving objects, and cosmological distances.
Geometric meaning and mathematical structure
In Euclidean geometry, the length of a straight segment is the distance between its endpoints. For points with Cartesian coordinates and , this distance is
The formula expresses the Pythagorean relation in three dimensions. Translations and rotations leave this distance unchanged, so an object's geometric length does not depend on where it is placed or how it is oriented. A change of scale, by contrast, multiplies every length by the same positive factor. Two geometrically similar triangles can therefore have identical angles but different side lengths.
For a vector representing a displacement, length is its magnitude, or norm. In a Euclidean vector space, the norm follows from the inner product: . Other norms are possible. The sum of the absolute coordinate differences, often called the Manhattan distance, represents travel along perpendicular street segments rather than an unrestricted straight line. Such alternatives are mathematically valid measures of distance, but they describe different structures and need not reproduce ordinary physical ruler measurements.
A metric space generalizes this idea by assigning a nonnegative distance to each pair of points. The distance is zero precisely when the points coincide, is symmetric, and satisfies the triangle inequality,
These requirements formalize basic properties of separation without requiring coordinates or a surrounding Euclidean space. Nevertheless, a metric alone does not guarantee the existence of a continuous path connecting two points, much less a shortest path. The distance between points and the length of an available route are therefore related but distinct concepts.
For physical objects, selecting a particular length usually requires a convention. The length of a rectangular beam is commonly its longest dimension, while a cylinder may be described by its axial length and diameter. The distance between a person's head and feet is termed height when measured in a standing posture, but body length in another measurement context. These labels have practical meaning without changing the underlying dimensional quantity.
Length must also be distinguished from area and volume. Although all three describe spatial extent, area scales as the square of a linear scale factor and volume as its cube. Doubling every dimension of a cube doubles its edge length, quadruples its surface area, and increases its volume eightfold. This difference underlies scaling effects in structures, organisms, and instruments.
Curves, surfaces, and irregular boundaries
The length of a curved path is not generally equal to the straight-line distance between its endpoints. A circle of radius , for example, has circumference , whereas a diameter joining opposite points has length . A road's route length can exceed both its map separation and its horizontal projection because of bends and changes in elevation. The endpoints alone are therefore insufficient to specify the length of a path.
The mathematical definition of arc length begins by approximating a curve with connected straight segments. Points are chosen in order along the curve, and the distances between consecutive points are added. The curve's length is the supremum of these polygonal sums over all finite ordered subdivisions. If this supremum is finite, the curve is called rectifiable. This definition does not assume that the curve is smooth and gives a precise meaning to the intuitive process of following a curve with increasingly short measuring segments.
For a continuously differentiable parametrized curve , with , the definition leads to the integral
For a plane graph , this becomes
The first expression also explains the connection with motion: integrating speed over time gives total distance traveled. This is not necessarily the magnitude of net displacement, since a moving object can reverse direction or return to its starting point.
On a curved surface, an intrinsic length follows the surface rather than cutting through the surrounding space. A geodesic is a locally length-minimizing path in a Riemannian geometry, although it need not be the globally shortest connection over its entire extent. Great-circle arcs provide the familiar spherical example. Between two points on a sphere, the shorter great-circle arc is generally the shortest surface route; the straight chord through the sphere is shorter still but is not confined to the surface.
Not every continuous curve has finite ordinary length. The ideal Koch curve, constructed by repeatedly replacing each segment with four segments one-third as long, has total polygonal length multiplied by at every iteration. Its limiting length is infinite despite its bounded spatial extent. Such examples are central to fractal geometry.
Physical boundaries introduce a related issue of measurement scale. A coastline measured with short steps generally appears longer than one measured with long steps, because finer measurement resolves additional inlets and projections. A real coastline is not an exact mathematical fractal at every scale, and its position also depends on tides, erosion, and the chosen shoreline definition. A reported boundary length consequently requires a scale and procedure, not merely a numerical value.
Units and the development of standards
Historical length units often originated in bodily dimensions, agricultural practice, or commercial convention. Cubits were associated with the forearm, feet with the foot, and spans with the outstretched hand. Their names did not ensure uniform size: different societies, cities, and trades maintained different realizations. Durable rods, architectural marks, and official standards helped stabilize these systems, but comparisons across jurisdictions remained difficult.
The metric system sought a more general basis. In the 1790s, revolutionary France defined the metre through an intended relationship to Earth's size: one ten-millionth of the meridian quadrant from the equator to the North Pole. Jean-Baptiste Delambre and Pierre Méchain surveyed a meridian arc between Dunkirk and Barcelona to support its determination. A platinum end standard was deposited in 1799. Later improvements in knowledge of Earth's shape did not continually alter the unit; the practical standard had become the authoritative reference.
The Metre Convention of 1875 established an international framework for maintaining measurement standards and created the International Bureau of Weights and Measures. In 1889, the first General Conference on Weights and Measures sanctioned the international prototype metre, a platinum–iridium bar whose reference length was defined by the separation between engraved lines under specified conditions. National copies allowed comparison and dissemination. The system nevertheless depended on material artifacts, their preservation, and the ability to compare them.
In 1960, the metre was redefined as wavelengths in vacuum of radiation associated with a specified transition in krypton-86. This replaced a unique artifact with a reproducible atomic phenomenon. In 1983, a further redefinition fixed the speed of light in vacuum at exactly metres per second, making the metre the distance traveled by light in vacuum during of a second. The SI formulation adopted in 2018 and effective in 2019 expresses this through the fixed numerical value of the defining constant, preserving the same unit.
SI prefixes accommodate different scales: a kilometre is metres, a millimetre metres, and a nanometre metres. The ångström, equal to metres, remains common in descriptions of atomic structures. Under the 1959 international agreement, the international yard is exactly metres and the international inch exactly metres. These exact conversion factors must be distinguished from the uncertainty of any actual ruler or measured object.
As a physical dimension, length is conventionally denoted . Dimensional analysis uses it to check relationships among quantities: speed has dimension , acceleration , and volume . Dimensional consistency is necessary for a physical equation, although it cannot by itself establish that the equation is correct.
Realizing the metre and establishing traceability
A unit definition is not itself a measuring instrument. To realize the metre, a laboratory must connect an observable process to the fixed constants of the SI. Since the speed of light is exact by definition, measuring an optical frequency determines the corresponding vacuum wavelength through
The frequency measurement is ultimately connected to the definition of the second. Stabilized lasers and optical frequency combs make this connection practical at very high precision. Exactness of the defining constant does not make a realized wavelength exact: frequency measurement, environmental conditions, and the measurement arrangement still contribute uncertainty.
Interferometry compares optical paths using the phase relationship between coherent light waves. In a simple displacement interferometer, motion of a reflecting mirror changes the outgoing and returning path. One complete fringe cycle then corresponds to a mirror displacement of approximately half a wavelength in the propagation medium. By counting fringes and measuring fractional phase, the instrument can resolve changes far smaller than the spacing of marks on a conventional scale.
The distinction between vacuum wavelength and wavelength in air is important. In a medium with refractive index ,
For ordinary air, is of the order of a few parts in at visible wavelengths. That difference is large in precision dimensional work. Air temperature, pressure, humidity, and composition affect the correction. An interferometer may therefore operate in vacuum, measure the environmental variables needed to calculate refractive index, or incorporate an independent refractive-index measurement.
Material standards remain essential even though they no longer define the unit. Gauge blocks provide accurately finished lengths that can be compared interferometrically or mechanically. Their flat faces can be wrung together to form combinations of lengths, with the details of contact and wringing included in high-accuracy work. Line scales, step gauges, calibrated spheres, and other artifacts transfer dimensional references to instruments with different measurement geometries.
Metrological traceability links a measurement result to a reference through a documented, unbroken chain of calibrations, each contributing to the uncertainty. A workshop instrument need not be compared directly with a primary laser system. It may be calibrated using an artifact that was itself calibrated against a higher-level standard. Traceability concerns the result and its calibration chain; merely possessing an instrument bearing a calibration label does not establish the reliability of every measurement made with it.
Calibration also differs from adjustment. Calibration establishes relationships between indications and reference values under stated conditions. Adjustment changes an instrument so that it behaves in a prescribed way. A calibrated instrument can have a known nonzero correction, which must be considered when interpreting its readings.
Measurement methods and uncertainty
A length measurement begins by defining the measurand: the particular quantity intended to be measured. For a metal cylinder, “length” might mean the distance between two fitted end planes, the greatest axial extent, or the separation measured at a specified location under a specified contact force. These definitions can produce different results if the ends are tilted, rough, or deformed. Instrument precision cannot resolve an ambiguity in what is being measured.
Direct-contact instruments include rulers, calipers, micrometers, and coordinate-measuring machines. A micrometer converts rotation of a precision screw into displacement; a coordinate-measuring machine samples surface points and calculates geometric features from them. Noncontact methods include optical imaging, laser triangulation, interferometry, and time-of-flight ranging. In a simple round-trip ranging arrangement,
where is the propagation speed in the relevant medium and is the corrected round-trip travel time. The factor of two accounts for travel to the target and back.
Geometry can introduce errors even when the scale is accurate. If a displacement is projected onto a measuring axis inclined by angle , the indicated component may be , producing a cosine error. Abbe error arises when the reference scale is offset from the measurement line and the mechanism undergoes small angular motion. For offset and small angle , the resulting error is approximately , with the sign determined by the geometry. Precision instruments therefore pay close attention to alignment and the placement of reference axes.
Measurement uncertainty expresses the dispersion of values reasonably attributable to the measurand. Its contributors can include scale calibration, repeatability, finite resolution, temperature, contact force, optical corrections, and imperfect knowledge of surface form. Repeated readings estimate some effects statistically but do not reveal every systematic influence. A ruler with a stable scale error may produce highly repeatable readings that remain displaced from the reference value.
For a measurement model , the combined standard uncertainty is commonly evaluated by propagating input uncertainties. If the inputs are uncorrelated, the first-order expression is
Correlated inputs require additional covariance terms. An expanded uncertainty is often reported as , together with the coverage factor and the basis for its interpretation. Reporting extra decimal places does not reduce uncertainty.
Digital processing adds further choices. An image-based diameter depends on pixel calibration, lens distortion, illumination, and the rule used to identify an edge. Filtering a profile can remove roughness or alter the apparent boundary. Consequently, two instruments can disagree because they implement different definitions of the feature, rather than because one simply has a faulty scale.
Material dimensions and engineering use
A manufactured object's length is not generally a fixed property independent of its surroundings. Changes in temperature, mechanical loading, moisture, and internal structure can change its dimensions. For sufficiently small temperature changes over a range in which the coefficient is approximately constant, linear thermal expansion is represented by
where is the linear expansion coefficient. Many steels have room-temperature values near –, although composition and temperature matter. A one-metre steel component can therefore change length by roughly millimetres for a temperature change of kelvins.
The standard reference temperature for geometrical and dimensional properties in industrial metrology is . Measurements made at other temperatures may require correction to that reference. Uniform temperature is not always enough: a gradient through a component can cause bending, while incomplete thermal equilibration can make its dimensions change during measurement. Low-expansion materials help in standards and precision structures, but no material should be treated as dimensionally invariant under all conditions.
Mechanical strain relates deformation to original length. For a uniformly extended bar, engineering strain is
Within a suitable linear elastic range, axial stress and strain are approximately related by Young's modulus, . Contact probes can therefore change the very dimension they measure, especially for soft polymers, thin walls, and biological samples. Permanent deformation, creep, and stress relaxation introduce time dependence that cannot be described by a single elastic correction.
Engineering drawings specify not only nominal dimensions but permissible variation. A shaft and hole must have compatible size limits to achieve a clearance, transition, or interference fit. Assemblies also depend on accumulated dimensional variation: several individually acceptable parts can combine into an unacceptable overall length. Geometrical specifications distinguish size from straightness, flatness, orientation, and position, since a component can have an acceptable average dimension while still failing to mate properly.
Length enters design through scaling relationships as well as direct dimensions. For geometrically similar solid objects made of the same material, mass scales as the cube of linear size, whereas the area carrying a load scales as its square. Under self-weight, stress therefore tends to increase with scale. Likewise, the bending response of a beam depends strongly on its length and cross-sectional geometry, not simply on the amount of material it contains.
Some engineering “lengths” are model parameters rather than endpoint separations. Hydraulic diameter characterizes flow passages through their area and wetted perimeter; effective buckling length represents the influence of support conditions on column stability. Such quantities have the dimension of length and can be measured or calculated consistently, but their physical interpretation depends on the model that defines them.
Atomic, molecular, and quantum length scales
At microscopic scales, length remains meaningful, but boundaries and measurement processes differ from those of rigid macroscopic objects. An atom does not have a hard outer surface. Its electrons are described by quantum states with spatial probability distributions, and different definitions—covalent, metallic, or van der Waals radius—characterize different interactions. A quoted atomic radius is therefore not simply the reading of a miniature caliper.
A useful atomic scale is the Bohr radius,
It is a characteristic length in the quantum description of hydrogen, rather than the radius of a sharp boundary. Typical chemical bond lengths are of the order of metres. Structural methods such as X-ray crystallography infer interatomic separations from diffraction data and physical models. The inferred positions and distances carry uncertainties and can reflect thermal motion, disorder, and averaging.
Flexible molecules require several distinct measures of extent. A polymer's contour length follows its backbone, whereas its end-to-end distance joins its terminal points directly. Its radius of gyration describes the distribution of material about its centre of mass. These quantities need not change together: a long polymer can form a compact coil without appreciably changing its contour length. Double-stranded DNA in the common B form has an axial rise of about nanometres per base pair, so a molecule's base-pair count supplies an approximate contour length under specified structural conditions, not its instantaneous end-to-end separation.
In quantum mechanics, spatial localization is linked to momentum uncertainty. For a position coordinate and its corresponding momentum component,
The uncertainty principle is a property of quantum states, not merely a statement that all microscopes are poorly made. Resolving small structures often involves probes with short wavelengths and correspondingly high momenta. Yet achievable resolution also depends on interactions, signal strength, aberrations, and the reconstruction procedure.
Other characteristic lengths identify the regime in which particular processes become important. The reduced Compton wavelength connects a particle's mass with a relativistic quantum scale. A mean free path describes the average travel distance between interactions in a specified medium. A correlation length describes how fluctuations or ordering remain related across space. None is necessarily the physical diameter of an object.
The Planck length,
combines constants associated with quantum theory, gravitation, and relativity. It is important in dimensional arguments about quantum gravity. It has not been experimentally established as a smallest possible distance, nor does its definition alone demonstrate that space is discrete.
Terrestrial and astronomical distances
On Earth, long-distance measurement must specify the reference surface and route. A distance measured along uneven ground differs from its horizontal projection and from a distance reduced to a reference ellipsoid. Because Earth is not a perfect sphere, high-accuracy surveying uses geometrical models together with information about the gravity field. Satellite positioning initially supplies positions within a reference frame; obtaining the desired ground distance or height requires appropriate transformations and corrections.
A geographical boundary may be defined by coordinates, a river channel, a watershed, or a physical monument. Its measured length depends on that definition. River length is especially sensitive to the selected source, the chosen channel through branching reaches, and the scale used to follow bends. Shorelines also move with tides, sediment transport, and sea level. Such quantities should not be interpreted as immutable constants merely because they are reported to several digits.
In astronomy, several units are convenient for distances much larger than terrestrial dimensions. The astronomical unit has been defined since 2012 as exactly metres. It is close to the characteristic Earth–Sun separation but is no longer defined by an evolving orbital estimate. A light-year is the distance light travels in vacuum during a Julian year of exactly days, giving exactly metres.
The parsec is tied to angular geometry. It is defined as astronomical units, approximately metres or light-years. Its name comes from parallax and arcsecond: stellar parallax relates an annual apparent angular displacement to the known scale of Earth's orbit. For a parallax expressed in arcseconds, the familiar astronomical relation is in parsecs. At small parallax, even a small angular error can produce substantial distance uncertainty.
Different distance ranges require different observations. Radar and laser ranging provide travel-time measurements within the Solar System. Stellar parallax supplies a geometric foundation farther out. Distances beyond practical parallax measurements may be inferred from objects whose luminosities can be calibrated, including certain variable stars and supernovae. These methods form interconnected calibrations rather than a single ruler extended into space.
In cosmology, several different distance definitions are necessary because the universe expands and light takes time to propagate. Luminosity distance connects an object's luminosity to its observed flux; angular-diameter distance connects transverse physical size to observed angle. Under standard assumptions about light propagation and photon conservation, they satisfy
where is redshift. Comoving distance and proper distance refer to further geometrical constructions. For a distant galaxy, a numerical “distance” is incomplete unless the definition and cosmological model are specified.
Reference frames, relativity, and curved spacetime
In classical mechanics, lengths are normally treated as independent of an observer's uniform motion. This approximation is extraordinarily effective when relative speeds are small compared with the speed of light and gravitational effects are weak. Special relativity changes the relationship between measurements made in different inertial frames, because observers moving relative to one another do not generally agree on which spatially separated events are simultaneous.
The length of a moving object must be determined from the positions of its endpoints at the same time in the measuring observer's frame. If a rod has rest length and moves parallel to its axis with speed , its measured length in that frame is
This is length contraction. Dimensions perpendicular to the relative velocity are unchanged by the corresponding Lorentz boost. The rest length is the length measured in the object's own inertial rest frame, where such a frame applies.
Length contraction is not simply the appearance of a moving object in a photograph. A camera receives light that left different parts of the object at different times; the resulting image combines propagation effects with relativistic geometry. Nor is contraction described as ordinary compression caused by mechanical stress. It arises from comparing endpoint events selected by different simultaneity conventions. Actual acceleration of an extended object presents additional questions, since signals and forces cannot propagate instantaneously through matter.
The underlying spacetime relation can be expressed, using one common sign convention, as
This interval is invariant under a Lorentz transformation, although the individual time and space separations are not. Spatial length is therefore not generally an invariant assigned to an arbitrary pair of events. Instead, it is obtained by specifying an appropriate set of simultaneous events or another explicit operational construction.
In general relativity, gravitation is described through the geometry of spacetime. On a specified spacelike slice, a spatial metric determines the length of a curve through
with repeated spatial indices summed. Coordinate differences alone need not equal physically measured distances. Moreover, different choices of spatial slice can yield different extended distance assignments, especially in dynamical spacetimes.
Local measurements still have a direct operational meaning: nearby observers can compare rods, clocks, and light signals in sufficiently small regions. Extended distances require additional structure, such as a chosen reference frame, synchronization procedure, or radar-ranging convention. In an expanding universe or near strongly gravitating bodies, these distinctions determine which length is being calculated and which observations can establish it.