Once the scientist has classified the phenomena of interest, his next task is to attempt to elucidate the relationships between those phenomena. He tenders these as a hypotheses. A hypothesis is a falsifiable statement. The statement of a hypothesis may be accessible using the vernacular. In most cases, however, the scientist must either refine the definition of his terms, or invent new terms, to ensure that the scope and implications of the relationship are precisely established.

For illustration, consider the proposition "the stars rise at sunset". This has two meanings in the vernacular: it may be an astronomical hypothesis (the celestial stars) or a sociological hypothesis (stars of Hollywood). We need to specify terms to ensure we are understood. Yet if vague, still the hypothesis is falsifiable. On rainy nights, the stars may not appear to rise at all. Does this mean that the stars aren't there?

Another hypothesis, with more predictive power, is that the acceleration due to gravity is constant everywhere near the surface of the earth. Given the definition of acceleration, this leads to a prediction for the time-dependence of a falling object's position.