Velocity is the rate at which position changes with time, measured relative to a specified reference frame. In classical mechanics, it is a vector quantity: it specifies both the rate of motion and its direction. Its magnitude is speed, a scalar quantity. An aircraft traveling north at 250 metres per second therefore has a different velocity from one traveling east at the same speed. Velocity is fundamental to physics because it connects descriptions of motion with momentum, energy, forces, and the transport of matter.
The concept applies to individual particles, extended bodies, fluids, and propagating disturbances, although these applications require important distinctions. A rotating body generally has different velocities at different points; a fluid is described by a velocity field; and the velocity of a wave pattern need not equal the velocity of the material through which it travels. Relativity changes how different observers compare velocities, while quantum mechanics limits the simultaneous precision with which position and momentum can be specified.
Definition, magnitude, and units
For an object with position vector , its average velocity over a time interval is its displacement divided by the elapsed time:
Displacement is the directed difference between the final and initial positions, not the total distance traveled. An object that completes one circuit of a track returns to its starting point and therefore has zero average velocity over the circuit, although its average speed is positive. More generally, the magnitude of average velocity cannot exceed average speed over the same interval.
Instantaneous velocity is obtained by taking the limit as the time interval approaches zero:
This definition uses the mathematical concept of a derivative and assumes that the trajectory is differentiable at the time in question. In three-dimensional Cartesian coordinates,
The components , , and may be positive, negative, or zero. Their signs indicate motion relative to the chosen coordinate axes; they do not distinguish physically positive from physically negative amounts of motion.
Speed is the magnitude of this vector:
Where the path is smooth, speed also equals , with the distance measured along the path. The direction of nonzero instantaneous velocity is tangent to the trajectory. At an instant when velocity is zero, no direction can be assigned to the velocity vector itself, even if the object subsequently begins moving along a well-defined path.
The unit of velocity in the International System of Units is the metre per second, written or . Its physical dimension is length divided by time, . Common alternatives include kilometres per hour, miles per hour, and knots. Exactly , and one knot is one nautical mile per hour, or exactly .
Descriptions such as “20 metres per second” give a speed unless a direction is stated or supplied by context. In one-dimensional calculations, a signed number can fully represent velocity once the positive direction has been defined. On a position–time graph, the slope represents velocity; on a velocity–time graph, the signed area under a component curve gives the corresponding displacement.
Trajectories and acceleration
Velocity describes the instantaneous state of motion, whereas acceleration describes how that state changes:
Because velocity includes direction as well as magnitude, acceleration can occur without any change in speed. An object moving uniformly around a circle is a central example: its velocity continuously changes direction even though its speed remains constant. Conversely, an object can momentarily have zero velocity while retaining nonzero acceleration, as at the highest point of an ideal vertically thrown projectile.
The inverse relations express velocity and position through integration:
These equations show why initial conditions matter. Knowledge of acceleration alone does not determine velocity without an initial velocity, and knowledge of velocity does not determine absolute position without an initial position. The integral also preserves directional information: opposite signed velocities can cancel in displacement even though both contribute positively to distance traveled.
For constant acceleration, the familiar equations become
Near Earth’s surface, a projectile can often be approximated as having constant downward acceleration of magnitude , provided air resistance and variation of gravity with height are negligible. Its horizontal velocity then remains constant while its vertical velocity changes linearly with time. The resulting path is parabolic. The approximation is not suitable for every projectile: atmospheric drag can be decisive for a falling leaf, a parachute, or a high-speed object.
For a smooth curved trajectory, acceleration separates into tangential and normal components:
where is the local radius of curvature, is the tangent direction, and points toward the local center of curvature. The first term changes speed; the second changes direction. In uniform circular motion, only the normal term remains, giving the centripetal acceleration .
The vector nature of these relationships prevents several common errors. “Deceleration” usually means decreasing speed, but it does not necessarily mean a negative acceleration component: an object moving in the negative coordinate direction slows down when its acceleration points positively. Similarly, a sharp bend in a plotted path does not represent a physically smooth turn at finite nonzero speed. An exact instantaneous change of direction would require an unbounded acceleration in the idealized mathematical description.
Reference frames, relative motion, and rotation
A velocity always refers to a reference frame, whether explicitly named or implicitly understood. A passenger seated in a moving train has approximately zero velocity relative to the carriage but shares the train’s velocity relative to the track. Neither description is incomplete merely because the other is possible. What matters is identifying the frame and using it consistently.
In Newtonian mechanics, two nonrotating frames whose origins move at constant relative velocity are related by the Galilean transformation. If their axes remain parallel,
Time is the same in both frames, and acceleration is unchanged. An inertial reference frame is one in which a body subject to no net force moves with constant velocity. Any frame moving uniformly relative to an inertial frame is also inertial within Newtonian theory; there is no experimentally privileged state of absolute rest in this framework.
Relative velocity is useful even when neither body is at rest in the selected frame. If two objects have positions and , then the velocity of relative to , expressed in the same nonrotating coordinate basis, is
For navigation, an aircraft’s velocity relative to the ground is the vector sum of its velocity relative to the air and the air’s velocity relative to the ground. A crosswind therefore changes both ground speed and ground-track direction. Adding only the numerical speeds gives an incorrect result unless the relevant motions are collinear.
Rotating frames require additional care because their coordinate axes change direction. If a moving frame has origin velocity and angular velocity , a point at relative position has inertial velocity
with all vectors expressed in a common basis. The cross product accounts for motion caused by rotation of the frame itself. Differentiating again introduces centrifugal, Coriolis, and, when the rotation rate changes, Euler acceleration terms. The Coriolis effect is particularly important in atmospheric and oceanic motion described relative to rotating Earth.
For a rigid body, distances between material points remain fixed. Its instantaneous motion can be represented by translation of a chosen reference point plus rotation:
A rotating wheel therefore does not possess one velocity common to all its points. For ideal rolling without slipping on stationary ground, the point of contact is instantaneously at rest relative to the ground, while the top of the wheel moves at twice the speed of its center. These velocities follow from combining translation and rotation, not from any deformation of the wheel.
Momentum, energy, and dynamical laws
Velocity becomes a dynamical quantity when connected to mass, momentum, and force. For an ordinary Newtonian particle of constant mass,
Thus a net force changes velocity rather than maintaining it. This distinction is central to Newton’s laws of motion: uniform straight-line motion requires no net force. Everyday experience can obscure this principle because friction and drag often oppose motion, so a continuing applied force is needed to balance those resistive forces.
The kinetic energy of a nonrelativistic particle is
It depends on speed, not on the direction of velocity. Momentum, by contrast, retains direction. Two identical objects moving oppositely at equal speeds have zero total momentum but positive total kinetic energy. This difference is essential in collision analysis. In an isolated system, total momentum is conserved; kinetic energy is additionally conserved only for an elastic collision. In inelastic collisions, some kinetic energy becomes internal energy, deformation, sound, or other forms.
Both kinetic energy and momentum depend on the reference frame. A stationary object has zero kinetic energy in its rest frame but positive kinetic energy in a frame in which it moves. Their frame dependence does not invalidate conservation laws: the laws apply consistently within each inertial frame. For a system of particles, total momentum can be written , where is the total mass and is the velocity of the center of mass. Internal motions can contribute kinetic energy even when the center of mass is at rest.
The instantaneous rate at which a force does work is
Only the force component along the velocity contributes to this mechanical power. A perpendicular force can bend a trajectory without changing speed. For example, the magnetic part of the Lorentz force on a charged particle is . It is perpendicular to the particle’s velocity and therefore does no work on that particle. In a uniform magnetic field, it can produce circular or helical motion while leaving the speed unchanged.
Resistance frequently depends on velocity. At sufficiently low Reynolds number, a sphere moving slowly through a Newtonian fluid experiences Stokes drag with magnitude , where is dynamic viscosity and is sphere radius. In many higher-Reynolds-number situations, drag is modeled approximately as proportional to . A falling object approaches terminal velocity when drag and buoyancy balance its weight. Terminal velocity is therefore not a universal speed of falling: it depends on the object, fluid, orientation, and flow regime.
Velocity fields and the transport of matter
An extended medium cannot generally be described by one velocity vector. Instead, fluid dynamics uses a velocity field, commonly written , that assigns a local velocity to each position and time. The field represents motion averaged over a scale large compared with molecular dimensions but small compared with the flow structures being studied. It is a continuum description rather than a record of each molecule’s instantaneous motion.
Two complementary descriptions are common. The Lagrangian description follows individual material parcels along their trajectories. The Eulerian description describes quantities at fixed spatial locations. A floating marker approximates the former viewpoint, while a current meter attached to a bridge illustrates the latter. A steady Eulerian velocity field can still produce accelerating particles: water entering a narrowing channel may move faster downstream even though the velocity at each fixed point is constant in time.
The acceleration of a material parcel is the material derivative of the velocity field:
The first term describes local change at a fixed position; the second describes change encountered as the parcel moves through a spatially varying field. This convective term is a major source of nonlinearity in fluid motion. Omitting it would incorrectly assign zero acceleration to every steady flow.
Velocity also enters the equation of mass conservation:
where is density. For a constant-density incompressible fluid, this reduces to . The volume flow rate through a surface is
where is its unit normal. If velocity is uniform and normal to the surface, . In a river, however, speeds normally vary with depth and lateral position, so river discharge requires integration or a suitable sampling approximation rather than multiplication by an arbitrary point speed.
Spatial derivatives of velocity describe deformation and local rotation. The curl , called vorticity, characterizes local rotational motion. Symmetric combinations of velocity gradients describe strain rates, which determine viscous stresses in Newtonian fluids. Under the usual no-slip approximation, fluid velocity matches the velocity of a solid boundary, producing strong gradients near walls. Viscosity transfers momentum across these gradients.
In turbulence, velocity fluctuates irregularly over many scales. A mean velocity field alone does not capture all transport: correlations among velocity fluctuations contribute additional momentum fluxes, represented by Reynolds stresses in averaged equations. Distinguishing instantaneous, time-averaged, and spatially averaged velocity is therefore essential when comparing laboratory measurements, numerical simulations, and large-scale environmental observations.
Wave velocities and signal propagation
A wave has several associated velocities, and these need not be identical. The velocity of a wave pattern is distinct from the velocity of the material carrying it. In a sound wave, air molecules oscillate around local mean positions while the disturbance propagates through the air. In an ideal traveling wave on a stretched string, string elements move mainly transversely while the wave travels along the string.
For a sinusoidal wave with phase , points of constant phase travel at the phase velocity
where is angular frequency and is wave number. This is the speed of a crest or another fixed phase point. A narrow-band wave packet, formed by combining waves with nearby frequencies, often has an envelope that travels at the group velocity
When the medium is dispersive, frequency and wave number are not simply proportional, and phase and group velocities can differ.
In a nondispersive idealization with , both velocities equal . For electromagnetic waves in vacuum, is the speed of light, exactly . In a transparent optical medium, the phase velocity is commonly expressed as , where is the refractive index. Its frequency dependence produces dispersion. Under suitable low-loss conditions, the group velocity is
This relation helps describe pulse delays in optical materials and fibers.
Group velocity must not always be identified with the speed of information or energy. In regions of strong absorption, gain, or anomalous dispersion, a pulse can change shape so substantially that its peak moves with an apparently superluminal or negative group velocity. Such behavior does not establish faster-than-light transmission of a new causal signal. The propagation of a signal front and the full response of the medium must be considered.
Water waves supply another distinction. For small-amplitude surface gravity waves on deep water, neglecting surface tension,
so the group velocity is half the phase velocity. Wave crests move through the packet rather than remaining fixed within it. The fluid particles execute approximately orbital motions, with additional effects such as Stokes drift arising beyond the simplest linear description.
Relative motion between source and observer changes the measured frequency through the Doppler effect. In acoustics, the motion of each relative to the medium matters. For electromagnetic radiation, relativistic relationships replace that medium-based description. Doppler measurements usually constrain the component of relative velocity along the line of sight; motion perpendicular to that line requires other information or, at high precision, consideration of relativistic frequency shifts.
Velocity in special and general relativity
Special relativity, introduced by Albert Einstein in 1905, replaces Galilean velocity transformations with relationships consistent with the invariance of the speed of light. Suppose one inertial frame moves at speed along the -axis of another. If an object has velocity components in the latter frame, then
where . These equations follow from the Lorentz transformation, including its transformation of time as well as position.
For collinear motions, the corresponding addition rule is
If a spacecraft moves at relative to a laboratory and launches an object forward at relative to itself, the laboratory measures approximately , not . At speeds much smaller than , the denominator differs negligibly from one, recovering the Newtonian approximation. If , the result is , expressing the invariant vacuum speed of light.
The ordinary three-velocity remains , but relativity also uses four-velocity, defined by differentiation of spacetime position with respect to proper time:
With metric signature , its invariant norm satisfies . This is a geometric statement about a timelike trajectory in spacetime, not a claim that an ordinary massive object has spatial speed . A photon has no rest frame and no proper time along its null trajectory, so this massive-particle definition of four-velocity does not apply to it.
Relativistic momentum and kinetic energy are
For an object with nonzero rest mass, reaching would require unbounded energy in this description. At low speed, expanding recovers . Relativistic corrections become indispensable in particle accelerators, where large increases in particle energy can correspond to only small increases in speed.
In general relativity, velocity comparisons between distant locations require additional specification because spacetime is curved and coordinate systems need not represent direct physical measurements. A local observer measures the velocity of a passing object using a local reference frame. Massive objects locally travel below , and light locally travels at . Coordinate velocities can take other values without violating these local statements.
Likewise, cosmological recession rates inferred from the changing proper distance between widely separated galaxies can exceed . Such rates concern an extended definition of distance on chosen cosmological time slices, rather than the local passage of one object through another’s inertial frame. Treating them as ordinary special-relativistic relative velocities would confuse distinct measurements.
Microscopic velocities and quantum mechanics
In a gas, molecules have a distribution of velocities rather than a single common velocity. A gas at rest macroscopically may have zero mean molecular velocity while its molecules move rapidly in all directions. Statistical mechanics distinguishes this random thermal motion from bulk flow. If is the local mean velocity, a molecule’s peculiar velocity is its actual velocity minus .
For a dilute classical ideal gas in thermal equilibrium, the Maxwell–Boltzmann distribution gives the probability density for molecular speed:
Here is molecular mass, is absolute temperature, and is the Boltzmann constant. The most probable speed, mean speed, and root-mean-square speed are, respectively,
They differ because the speed distribution is not symmetric.
For nitrogen molecules at approximately , the root-mean-square speed is about . This is not the speed at which a recognizable parcel of nitrogen necessarily crosses a room. Frequent molecular collisions continually change individual velocities, while large-scale mixing involves bulk flow and diffusion. Molecular impacts transfer momentum to boundaries and produce gas pressure. Temperature is related to random motion relative to the bulk flow, not simply to the gas’s speed relative to an arbitrary external observer.
In quantum mechanics, velocity is represented through the dynamics of operators and states rather than assumed to accompany a precisely defined classical trajectory. For a nonrelativistic particle with Hamiltonian
the Heisenberg equation gives
For a charged particle in an electromagnetic vector potential, mechanical momentum differs from canonical momentum; in the standard minimal-coupling description, velocity is . Confusing these momenta can lead to incorrect physical interpretations.
The uncertainty principle implies, for the simple nonrelativistic relation ,
It constrains the joint statistical spreads of position and velocity, not merely the quality of an instrument. Quantum probability flow can also be described by a probability current derived from the wave function, but that current is not automatically a unique classical particle trajectory.
In crystals, an electron wave packet associated with an energy band has semiclassical group velocity . Its response depends on the band structure rather than on the free-particle relation alone. Thermal, quantum, and macroscopic velocities thus refer to related but distinct levels of description.
Measurement and uncertainty
Velocity can be measured from changes in position, from frequency shifts, or through a calibrated physical response to motion. The simplest operational method determines displacement during a known time interval. For nearly uniform straight-line motion, is sufficient. For varying motion, the result is an average over that interval, not an exact instantaneous value.
Optical gates, high-speed cameras, and motion-tracking systems estimate velocity from successive position measurements. A central-difference estimate is
Smaller time steps can improve temporal resolution but also amplify position noise when differences are divided by a shorter interval. Filtering or fitting a trajectory can reduce noise, yet introduces assumptions about smoothness and can suppress rapid changes. Exposure time, sampling rate, lens distortion, and synchronization all affect the result.
Doppler instruments infer velocity from frequency changes. For a monostatic radar reflecting from a target moving slowly compared with , the approximate shift is
where is radial velocity and is transmitted wavelength; the sign depends on the convention for approaching or receding motion. The factor of two arises from the outward and return propagation. A single radar ordinarily measures radial velocity, not the full three-dimensional velocity vector. Multiple viewing directions or additional tracking information are needed to reconstruct more components.
Laser Doppler methods use optical frequency differences to measure motion, while acoustic Doppler instruments apply related principles to moving scatterers in water or air. Particle image velocimetry estimates displacement patterns between images of tracer-seeded flow. Its output is a velocity field averaged over finite interrogation regions and time intervals. Tracer particles must follow the fluid sufficiently closely; particles with excessive inertia can depart from the motion being measured.
Satellite navigation systems can estimate receiver velocity from changes in position and from Doppler shifts in received carrier signals. In astronomy, spectroscopy measures line-of-sight motion through spectral shifts, while angular proper motion combined with distance constrains transverse motion. Converting a measured redshift into velocity requires the appropriate model: low-speed Doppler, relativistic, gravitational, and cosmological interpretations are not interchangeable.
Every velocity measurement has measurement uncertainty. For independent uncertainties in distance and time in the simple ratio , first-order propagation gives
Correlations require additional terms. Reliable reporting also identifies the reference frame, measured component, averaging interval, and relevant calibration. A numerical speed without these qualifications may not be comparable to another instrument’s output, even when both are expressed in metres per second.
Historical development
Ancient accounts of motion did not generally use velocity in its modern vector and differential form. Aristotle distinguished natural and forced motions within a broader physical philosophy, but his treatment did not formulate the inertial relationship between force and change of velocity. Medieval scholars subsequently developed more quantitative ways of discussing rates of motion. Fourteenth-century scholars associated with the Oxford Calculators formulated the mean speed theorem: uniformly accelerated motion covers the same distance as uniform motion at the average of its initial and final speeds over the same duration.
Nicole Oresme used geometrical representations of varying qualities, including motion, to illuminate such relationships. These constructions anticipated aspects of later graphs in which an area represents an accumulated quantity. They were not identical to modern coordinate graphs or differential calculus, but they helped separate the magnitude of a changing rate from the total amount accumulated over time.
Galileo Galilei made accelerated motion a central subject of mathematical and experimental investigation. His Two New Sciences, published in 1638, presented results concerning falling bodies, inclined planes, and projectiles. For motion from rest under uniform acceleration, the distances traveled scale as the squares of elapsed times. Inclined planes allowed comparatively slow motions to be studied, making timing more practicable than in a short vertical fall. Galileo also analyzed the combination of uniform horizontal motion and accelerated vertical motion in ideal projectile trajectories.
The seventeenth-century development of calculus supplied a general language for instantaneous rates. Isaac Newton treated changing quantities through fluxions, while Gottfried Wilhelm Leibniz developed differential notation. Newton’s Philosophiæ Naturalis Principia Mathematica of 1687 connected changes in motion to impressed forces and analyzed terrestrial and celestial trajectories within one mathematical framework. Although its demonstrations were largely geometrical, the resulting mechanics could be expressed through derivatives of position and momentum.
The modern vector treatment developed gradually during the nineteenth century from work on geometry, mechanics, and algebra. It made the distinction between a directed velocity and its scalar magnitude explicit and provided systematic operations for resolving components, describing rotation, and calculating relative motion. Meanwhile, kinetic theory connected molecular velocity distributions with pressure and temperature. James Clerk Maxwell introduced a molecular velocity distribution in 1860, and Ludwig Boltzmann subsequently extended the statistical treatment of gases.
Twentieth-century physics altered the interpretation rather than eliminating the concept. Special relativity replaced universal time and simple velocity addition with spacetime transformations. Quantum mechanics replaced the general assumption of a definite microscopic trajectory with state-dependent predictions and operator relationships. Across these developments, velocity retained its role as a rate of positional change, while the meaning of position, time, observation, and physical state became more precisely specified.