HYPOTHESES AND THEORIES 39 



good health and takes plenty of exercise and nourishing food, 

 has a son resembling himself, then what is the relation between 

 the facts? To what shall we attribute the likeness between 

 father and son ? Is the son strong and healthy merely because he 

 was 'born' with the characteristics with which the father was 

 ' born,' and because he has lived like him ? Or has the life led by 

 the father altered the germ-plasm and improved the child's heritage ? 

 Or are both factors to be taken into account ? Obviously, our task 

 in ascertaining the relations between the facts is harder here than 

 in anatomy. We are obliged to draw a ' mediate ' inference. The 

 immense number of erroneous and discarded hypotheses with which 

 the past of heredity is littered is evidence of the difficulty found by 

 its students in drawing correct inferences. 1 



70. An inference is a hypothesis. But at first it is only a 

 working hypothesis, a mere guess. Before we can be sure it is true, 

 before we can convert it into an established theory r , we must test the 

 thinking by which it was reached, and so ascertain whether it is 

 correct. Now there is one way, and one way only, by which the truth 

 of hypotheses can be tested. As we test facts by appealing to facts, 

 so we must test thoughts by appealing to thoughts to inferences. 

 We must proceed by two steps. First, we must make a rigorous 

 deductive inference of consequences. Next, we must ascertain by 

 an appeal to reality in the world around us whether these predicted 

 consequences actually do occur. 2 For example, when families 

 dwell under unhealthy conditions, as in the slums of great cities, 

 the parents tend to be sickly and the children puny. In this case 



1 The necessity of adopting diverse methods when observing and classifying 

 the facts of the various sciences is hardly realized by some students. Occasionally 

 one hears an anatomist, zoologist, or botanist, a ' systematist,' lament the rage for 

 speculation and declare that the function of science is merely to observe and 

 classify facts by which is meant of course the kind of observing and classifying 

 that is usually adequate for the purposes of the systematic sciences. But, 

 obviously, such observing would be of little use to the chemist, and such classifying, 

 when applied to a subject so abstruse as heredity, would totally fail to establish 

 the relations between the facts. The difficulty of simply observing compels the 

 chemist to experiment ; the difficulty of classifying compels the student of 

 heredity to speculation, which, however, when founded on strictly ascertained 

 facts and rigorously tested by a deductive inference of consequences, is not un- 

 scientific, but merely an especially strenuous effort to think accurately to classify 

 correctly. See 819 et seq. 



2 " Every hypothesis is an attempt to find meaning in observed phenomena, 

 to constitute reality in a rational way. It follows that the fundamental condition 

 of a valid hypothesis is that it should explain and give meaning to the facts of 

 observation. And it can only do this if it embraces those facts in that systematic 

 whole which is the one form under which it is possible to think the universe. This 



