Until relatively recent times, it was not known whether light travelled instantaneously or at a finite speed. The first extant recorded examination of this subject was in ancient Greece. Empedocles maintained that light was something in motion, and therefore must take some time to travel. Aristotle argued, to the contrary, that "light is due to the presence of something, but it is not a movement". Euclid and Ptolemy advanced the emission theory of vision, where light is emitted from the eye, thus enabling sight. Using that theory, Heron of Alexandria advanced the argument that the speed of light must be infinite, since distant objects such as stars appear immediately upon opening the eyes.
     Early Islamic philosophers initially agreed with the Aristotelian view that light had no speed of travel. In 1021, Islamic physicist Alhazen (Ibn al-Haytham) published the Book of Optics, in which he used experiments related to the camera obscura to support the now accepted intromission theory of vision, where light moves from an object into the eye. This led Alhazen to propose that light must therefore have a finite speed, and that the speed of light is variable, decreasing in denser bodies. He argued that light is a "substantial matter", the propagation of which requires time "even if this is hidden to our senses".
     Also in the 11th century, Abu Rayhan al-Biruni agreed that light has a finite speed, and observed that the speed of light is much faster than the speed of sound. Roger Bacon argued that the speed of light in air was not infinite, using philosophical arguments backed by the writing of Alhazen and Aristotle. In the 1270s, Witelo considered the possibility of light travelling at infinite speed in a vacuum but slowing down in denser bodies. A comment on a verse in the Rigveda by the 14th century Indian scholar Sayana mentioned a speed of light, about 186,400 miles per second, that was chosen so that light would encircle the Puranic universe in one day, making it "the most astonishing 'blind hit' in the history of science!". In 1574, the Ottoman astronomer and physicist Taqi al-Din concluded that the speed of light is constant, but variable in denser bodies, and suggested that it would take a long time for light from the stars, which are very distant, to reach the Earth.
     In the early 17th century, Johannes Kepler believed that the speed of light was infinite since empty space presents no obstacle to it. René Descartes argued that if the speed of light were finite, the Sun, Earth, and Moon would be noticeably out of alignment during a lunar eclipse. Since such misalignment had not been observed, Descartes concluded the speed of light was infinite. Descartes speculated that if the speed of light were found to be finite, his whole system of philosophy might be demolished.

     In 1629, Isaac Beeckman proposed an experiment in which a person would observe the flash of a cannon reflecting off a mirror about one mile (1.6 km) away. In 1638, Galileo Galilei proposed an experiment, with an apparent claim to having performed it some years earlier, to measure the speed of light by observing the delay between uncovering a lantern and its perception some distance away. He was unable to distinguish whether light travel was instaneous or not, but concluded that if it weren't, it must nevertheless be extraordinarily rapid. Galileo's experiment was carried out by the Accademia del Cimento of Florence in 1667, with the lanterns separated by about one mile, but no delay was observed. Based on the modern value of the speed of light, the actual delay in this experiment would be about 11 microseconds. Robert Hooke explained the negative results as Galileo had by pointing out that such observations did not establish the infinite speed of light, but only that the speed must be very great.

     The first quantitative estimate of the speed of light was made in 1676 by Ole Christensen Rømer, one of a group of astronomers of the French Royal Academy of Sciences who were studying the motion of Jupiter's moons. From the observation that the periods of Jupiter's innermost moon Io appeared to be shorter when the earth was approaching Jupiter than when receding from Jupiter he concluded that light travels at a finite speed, and was able to estimate that would take light 22 minutes to cross the diameter of Earth's orbit. Christiaan Huygens combined this estimate with an estimate for the diameter of the Earth's orbit to obtain an estimate of speed of light of 220,000 km/s, 26% lower than the actual value.
     Isaac Newton also accepted the finite speed. In his 1704 book Opticks he gives a value of "seven or eight minutes" for the time taken for light to travel from the Sun to the Earth (the modern value is 8 minutes 19 seconds). The same effect was subsequently observed by Rømer for a "spot" rotating with the surface of Jupiter. Later observations also showed the effect with the three other Galilean moons, where it was more difficult to observe, thus laying to rest some further objections that had been raised.
     Between 1725 and 1728, James Bradley, while searching for stellar parallax, observed the apparent motion of the star γ Draconis (Eltanin) depending on the season of the year. He realized that the motion (about 39 arcseconds) could not be a parallax (it was in the wrong direction at any given time) and, after ruling out several other possible causes, produced the theory of the aberration of light, a vector addition of the velocity of light arriving from the star and the velocity of the Earth around its orbit. The effect is that an observer on the Earth will see the light coming from a slightly different angle than the "true" value which, for a star in the sky, means a slightly different position. The effect is greatest near the orbital pole which, for the Earth, is close to γ Draconis. Bradley was able to predict the aberration for several other stars, and confirm his predictions by observation. His observations on γ Draconis gave a ratio of the speed of light to the mean linear speed of the Earth's orbital motion: Bradley's figure was that light travelled 10,210 times faster than the Earth in its orbit (the modern figure is 10,066 times faster) or, equivalently, that it would take light 8 minutes and 12 seconds to travel from the Sun to the Earth.

     The first successful entirely earthbound measurement of the speed of light was carried out by Hippolyte Fizeau in 1849. Fizeau's experiment was conceptually similar to those proposed by Beeckman and Galileo. A beam of light was directed at a mirror 8 km away. On the way from the source to the mirror, the beam passed through a rotating cog wheel. At a certain rate of rotation, the beam could pass through one gap on the way out and another on the way back. But at slightly higher or lower rates, the beam would strike a tooth and not pass through the wheel. Knowing the distance to the mirror, the number of teeth on the wheel, and the rate of rotation, the speed of light could be calculated. Fizeau reported the speed of light as 313,000 km/s. Léon Foucault improved on Fizeau's method by replacing the cogwheel with a rotating mirror. Foucault's estimate, published in 1862, was 298,000 km/s.
     In 1879, Albert Michelson performed a similar experiment at the U.S. Naval Academy. He measured the speed of light in air to be 299,864±51 kilometres per second, and estimated the speed of light in vacuum as 299,940 km/s, or 186,380 mps.
     In 1887, Michelson and Edward Morley performed an experiment to detect differences in the speed of light due to the Earth's motion through the luminiferous ether, at what is now Case Western Reserve University.[86] Its failure is generally considered to be the first strong evidence against the ether theory.

     During World War II, the development of the cavity resonance wavemeter for use in radar, together with precision timing methods, opened the way to laboratory-based measurements of the speed of light. In 1946, Louis Essen and A.C. Gordon-Smith used a microwave cavity of precisely known dimensions to establish the frequency for a variety of normal modes of microwaves. As the wavelength of the modes was known from the geometry of the cavity and from electromagnetic theory, knowledge of the associated frequencies enabled a calculation of the speed of light.
     The Essen–Gordon-Smith result, 299,792 ± 3 km/s, was substantially more precise than those found by optical techniques, and prompted much controversy. However, by 1950 repeated measurements by Essen established a result of 299,792.5±1.0 km/s, which became the value adopted by the 12th General Assembly of the Radio-Scientific Union in 1957.

     An alternative to the cavity resonator method to find the wavelength for determining the speed of light is to use a form of interferometer, indicated schematically in the figure. A coherent light beam with a known frequency (ν), as from a laser, is split to follow two paths and then recombined. By carefully changing the path length and observing the interference pattern, the wavelength of the light (λ) can be determined, which is related to the speed of light by the equation c = λ ν.
     The main problem with interferometry is to measure the frequency of light in or near the optical region. This was first overcome by a group at the NIST laboratories in Boulder, Colorado, in 1972. By a series of photodiodes and specially constructed metal–insulator–metal diodes, they succeeded in linking the frequency of the caesium transition used in atomic clocks to the frequency of a methane-stabilized laser (nearly 10,000 times higher). Their results were

• frequency ν = 88.376181627(50) THz
• wavelength λ = 3.392231376(12) µm
• speed of light c = 299 792 456.2(1.1) m/s

     The 1972 measurement of the speed of light, with a relative uncertainty of 4×10-9, was not only a feat of experimental precision, it also demonstrated a fundamental limit to how precisely the speed of light could be measured at that time using any technique. The remaining uncertainty in the value was almost completely attributable to uncertainty in the length of the metre.
     Since 1960, the metre had been defined as a given number of wavelengths of the light of one of the spectral lines of a krypton lamp,[Note 10] but it turned out that the chosen spectral line was not perfectly symmetrical. This gave an uncertainty in its wavelength, and hence in the length of the metre. By analogy with a metal measuring stick, it was as if the stick were slightly fuzzy at each end, although if it were a real measuring stick, the fuzziness at the ends of a one-metre stick would only be apparent at the atomic scale.
     To get round this problem, the 15th Conférence Générale des Poids et Mesures (CGPM) in 1975 recommended the use of the value 299,792,458 m/s for "the speed of propagation of electromagnetic waves in vacuum". The 17th CGPM in 1983 decided to redefine the metre to be "the length of the path travelled by light in vacuum during a time interval of 1/299,792,458 of a second".
     The effect of this definition gives the speed of light the exact value 299,792,458 m/s, which is nearly the same as the value 299,792,456.2(1.1) m/s obtained in the 1972 experiment.