The first step towards the investigation of the extent to which science has in fact taken and may continue to take the place of religion is to decide the sense in which we are to understand the word science. First, we are to discuss not scientia but ‘natural science’, and this is not a sharply definable conception. Natural science is the knowledge and theories that are obtained by the scientific method. The scientific method starts from observations, verifiable by repetition or, where repetition is not possible, by the making of similar observations. These observations lead by the method of induction to statements about classes of observations, which statements we dignify with the name of laws. These laws may be used as material for the deduction of other laws which are only accepted if verified by their prediction of what experiment confirms.

The observations of science are never wholly accurate (except when they are enumerations), but their degree of error is usually known and so we can arrive at comparatively reliable statements about what has been observed. The inductions made from these observations are of a degree of reliability dependent on the number of observations used and the accuracy with which these are expressed by the generalisation. Some scientific laws, then, make statements which are very unlikely to be appreciably modified, while others are considerably less certain.

The scientific laws are not a disconnected set of rules; they are unified by the theories or suppositions concerning the nature of what science studies. Thus the atomic theory links together hundreds or thousands of laws in physics and chemistry, explains them, and suggests new lines of investigation. These theories are the more likely in proportion as they link together a wider range of laws (and so of observations) and as they are fruitful in leading to new knowledge. Some theories, such as the atomic theory, are practically accepted as fact; others, such as Wegener’s theory of floating continents, as fruitful conjectures.