Testing a scientific hypothesis involves at least four different activities. First, the hypothesis must be examined for internal consistency. A hypothesis that is self-contradictory or not logically well-formed in some other way should be rejected. Second, the logical structure of the hypothesis must be examined to ascertain whether it has explanatory value, i.e., whether it makes the observed phenomena intelligible in some sense, whether it provides an understanding of why the phenomena do in fact occur as observed.

A hypothesis that is purely tautological should be rejected because it has no explanatory value. A scientific hypothesis identifies the conditions, processes, or mechanisms that account for the phenomena it purports to explain. Thus, hypotheses establish general relationships between certain conditions and their consequences or between certain causes and their effects. For example, the motions of the planets around the sun are explained as a consequence of gravity, and respiration as an effect of red blood cells that carry oxygen from the lungs to various parts of the body.

Third, the hypothesis must be examined for its consistency with hypotheses and theories commonly accepted in the particular field of science, or to see whether it represents any advance with respect to well-established alternative hypotheses. Lack of consistency with other theories is not always ground for rejection of a hypothesis, although it will often be. Some of the greatest scientific advances occur precisely when it is shown that a widely-held and well supported hypothesis is replaced by a new one that accounts for the same phenomena that were explained by the preexisting hypothesis, as well as other phenomena it could not account for. One example is the replacement of Newtonian mechanics by the theory of relativity, which rejects the conservation of matter and the simultaneity of events that occur at a distance—two fundamental tenets of Newton's theory.

Examples of this kind are pervasive in rapidly advancing disciplines, such as molecular biology at present. The so-called "central dogma" holds that molecular information flows only in one direction, from DNA to RNA to protein. The DNA contains the genetic information that determines what the organism is, but that information has to be expressed in enzymes (a particular class of proteins) that guide all chemical processes in cells. The information contained in the DNA molecules is conveyed to proteins by means of intermediate molecules, called messenger RNA. David Baltimore and Howard Temin were awarded the Nobel Prize for discovering that information could flow in the opposite direction, from RNA to DNA, by means of the enzyme reverse transcriptase. They showed that some viruses, as they infect cells, are able to copy their RNA into DNA, which then becomes integrated into the DNA of the infected cell, where it is used as if it were the cell's own DNA.

Other examples are the following. Until very recently, it was universally thought that only the proteins known as enzymes could mediate (technically "catalyze") the chemical reactions in cells. However, Thomas Cech and Sidney Altman received in 1989 the Nobel Prize for showing that certain RNA molecules act as enzymes and catalyze their own reactions. One more example concerns the so-called "co-linearity" between DNA and protein. It was generally thought that the sequence of nucleotides in the DNA of a gene is expressed consecutively in the sequence of aminoacids in the protein. This conception was shaken by the discovery that genes come in pieces, separated by intervening DNA segments that do not carry genetic information; Richard Roberts and Philip Sharp received the 1993 Nobel Prize for this discovery.

The fourth and most distinctive test is the one I have identified, which consists of putting on trial an empirically scientific hypothesis by ascertaining whether or not predictions about the world of experience derived as logical consequences from the hypothesis agree with what is actually observed. This is the critical element that distinguishes the empirical sciences from other forms of knowledge: the requirement that scientific hypotheses be empirically falsifiable. Scientific hypotheses cannot be consistent with all possible states of affairs in the empirical world. A hypothesis is scientific only if it is consistent with some but not with other possible states of affairs not yet observed in the world, so that it may be subject to the possibility of falsification by observation. The predictions derived from a scientific hypothesis must be sufficiently precise that they limit the range of possible observations with which they are compatible. If the results of an empirical test agree with the predictions derived from a hypothesis, the hypothesis is said to be provisionally corroborated; otherwise it is falsified.

The requirement that a scientific hypothesis be falsifiable has been called by Karl Popper the criterion of demarcation of the empirical sciences because it sets apart the empirical sciences from other forms of knowledge. A hypothesis that is not subject to the possibility of empirical falsification does not belong in the realm of science.

See my "On the Scientific Method, Its Practice and Pitfalls," Hist. Phil. Life Sci. 16 (1994), pp. 205-240.