Science is modelling in action (recall the quote ‘science is the name, modelling is the game’).
Whenever natural scientists try to understand phenomena that they observe, they make models. For example, geometrical optics has emerged in this way.
In order to describe the course of light rays through prisms, mirrors and lenses, the assistance of geometry was called in. With simple rules about
reflection (angle of incidence equals angle of reflection), refraction (Snell's law) and lenses (parallel rays go through the focal point) a geometrical system
has been worked out with which one could understand and design telescopes and eyeglasses. Geometrical optics is a model of the operation of the ‘real’ optics
that enables us to understand that optics. But it also has limitations: one cannot explain deviations such as chromatic and spherical aberration. And the
relationship between geometrical optics and light as a wave phenomenon is certainly not trivial. Nevertheless, the model is still very useful.1
Models are also used to make new predictions and to generate ideas. In 1907 Einstein conceived a model system in which he could express the basic idea of his general theory of relativity: people in an elevator without windows. Those persons cannot distinguish between a stay on earth and a state in which the elevator is inside a rocket of which the motor generates an acceleration equal to the gravitational acceleration on earth. In both cases, the person feels a force in the direction of his feet. Also, the person cannot distinguish between an elevator in free fall and in an orbit around the earth.
Video: Einstein-elevator2
Einstein's insight via this model system was that if a person cannot see or feel any distinction between the two situations, then there does not exist any distinction from physics point of view. This insight led Einstein ultimately to his theory of general relativity in 1915.