What Science Does
There is another type of scientific statements called (perhaps improperly) "laws". Those are rules or formulas that express patterns that are observed in experiments, but not yet explained with any theory. Such tentative rules are less reliable than laws based on theories. However, often a pattern is observed before a theory is constructed, so many laws were initially of this type, and later became incorporated in a theory. Mendel's law would be an example, as Mendel was able to express the ratio of phenotypes before he came up with a theoretical explanation. Another example is Hubble's law in astronomy, which says that other galaxies are moving away from us (based on the red shift of the light we receive from them) and that the farther they are from us, the faster they are moving away. When this law was originally stated, it was just a concise expression of an observed pattern, without a theory behind it; today it is part of the theory of the expanding Universe that started from a Big Bang.
Some experts make a distinction between theories and hypotheses in terms of reliability. According to this classification, a hypothesis is an idea in an early stage of development, sufficient to produce testable predictions, but not yet thoroughly tested. A theory, on the other hand, is a more developed idea that has already withstood many experimental tests. This distinction is sometimes useful, but it does not accurately reflect the actual use of those terms by scientists. For example, string theories in physics have not yet been tested by experiment; however, they are complex sets of ideas that apply to a broad range of phenomena, so they are called theories rather than hypotheses.
Parsimony and Occam's Razor
Testability (falsifiability) is the most important criterion an idea must satisfy to be a scientific theory. But there is one other criterion that is almost as important - the criterion of parsimony, also known as Occam's razor. First formulated by English philosopher William of Occam (1285-1349), this is a requirement that entities in a logical system should not be multiplied beyond necessity. In simpler, everyday language: one should not make more assumptions than needed. If a glass I left on the table in the evening is on the floor in the morning, it may be because somebody broke into the house skillfully, causing no damage, put the glass on the floor to confuse me, and left without disturbing or stealing anything, or it may be because my cat jumped on the table and knocked the glass down. I may not find evidence that contradicts either explanation, but the second one is far simpler (more parsimonious, i.e., involves fewer assumptions), and thus preferred based on Occam's razor.
In science, there are often multiple theories trying to explain the same phenomena. If we can find observations that contradict one of them, we can reject that theory and favor the alternative. But if such observations are not available, we prefer the simpler theory. The Ptolemaic theory of geocentric Universe and the Copernican theory, in which the Earth and other planets circle around the Sun, can both produce equally correct predictions, but the former only at the cost of great complications, and the latter in a simple and concise form. By Occam's razor, the heliocentric system of Copernicus quickly prevailed despite the Church's initial resistance.
The Ptolemaic theory suffered another methodological deficiency. It had to be modified and new rules added to it many times after discoveries of new celestial objects. It lacked clear delineation of what assumptions are allowed in it, so it was possible to adapt it almost without limit and incorporate new information in it. Modern scientific theories are expected to be constructed in a more disciplined way so that they cannot be arbitrarily expanded and complicated.
