The ancient Greeks were the first to propose theories of light and vision. Democritus suggested that atoms swarm from objects into the eye, the first particle theory. Other philosophers of Antiquity thought that some ray from the eye scanned the object. Neither theory explains why we don’t see in the dark. Arab writers were the first to suggest that light emanates from some objects (the Sun and stars, fire) and is reflected into the eye. But the nature of light itself was not obvious –– it could be particles such as the ones Democritus suggested or it could be some other entity.
Some 17th-century scientists promoted a wave theory of light, as had Leonardo da Vinci earlier. Robert Hooke and Robert Boyle had observed the colors appearing in oil on water. By 1665 both had explained this by suggesting interference of the rays reflected from the two surfaces of the thin film of oil. Newton rejected the wave theory of light, although his theory allowed for wavelike constructs. But Newton did not believe that waves could cast sharp shadows. In principle, he was right. In reality, light does bend around sharp edges the way that waves do, but this could not be observed in Newton’s time. The amount of bending is too small. As a result, Newton believed that light must consist of particles.
Particle theories were popular in England in the 18th century, but a series of experiments by Thomas Young, starting in 1801, demonstrated that light is a wave phenomenon. Young showed convincingly that rays of light could interfere with each other, a phenomenon only possible for waves.
But Newton was not entirely wrong. In 1905 Albert Einstein showed that the photoelectric effect could be explained by assuming that light is composed of particles. In the photoelectric effect, electrons are knocked from a substance by light. Then Louis de Broglie suggested that entities recognized as particles could also be understood as waves. Experiments later confirmed this idea. Finally, physicists agreed that whether light behaves as a wave or as a particle depends on the property that is being measured.