_Comparative Method._—Claude Bernard completed this critical work by laying down the laws of experiment on living beings. He commended as the rational method of research the _comparative method_. This should be, and is in fact, the daily instrument of all those who work in physiology. It compels the investigator in every research bearing on organized beings to institute a series of tests, such that the conditions which are unknown and impossible to know may be regarded as identical from one test to another; and when we are certain that a single condition is variable, it compels him to discover the character of the condition we are dealing with, and to learn to appreciate, and to measure its influence. It is safe to say that the errors which are daily committed in biological work have their cause in some infraction of this golden rule. In physical science the obligation to follow the comparative method is much less felt. In most cases the _witness test_[3] is useless. In physiology the witness test is indispensable. [3] In an article on the experimental method recently published in the _Dictionnaire de Physiologie_, M. Ch. Richet writes as follows:—“We must therefore never cease to carry out comparative experiments. I do not hesitate to say that this comparison is the basis of the experimental method.” It is in fact what was taught by Claude Bernard in maxim and by example. It is no exaggeration to assert that nine-tenths of the errors which take place in research work are imputable to some breach of this method. When an investigator makes a mistake, save in the case of material error, it is almost certainly due to the fact that he has neglected to carry out one of the comparative tests required in the problem before him. The following is an instance which happened since the above pages were written:—Several years ago a chemist announced the existence in the blood serum of a ferment, lipase, capable of saponifying fats—that is to say, of extracting from them the fatty acid. From this he deduced many consequences relative to the mechanism of fermentations. But on the other hand, it has been since shown (April 1902) that this lipase of the serum does not exist. How did the error arise? The author in question had mixed normally obtained serum with oil, and he had noted the acidification of the mixture; he assured himself of the fact by adding carbonate of soda. He saw the alkalinity of the mixture, serum + oil + carbonate of soda, diminish, and he drew the conclusion that the acid came from the saponified oil. He did not make the comparative test, serum + carbonate of soda. If he had done so, he would have ascertained that it also succeeded, and that therefore as the acid did not come from the saponification of the oil, since there was none, its production could not prove the existence of a lipase.