INDUCTION. 594 obsen-er may simply look at things as they offer themsflves to his notice, or ho may set about to interfere with the spontaneous course of events in order that he may the l>etter observe the connection. The latter sort of observation is called nn txperiment. When experiment is posisiblc, it is usually of fjreat assistance in col- lecting data for jjeneralization. Indeed, almost nil the great advances in modern sc-ienec have been due to experimentation rather than to mere observation. In intentional observation, whether it be simple or ex|ieriMiental, it is extremely important for the observer to know what he is to look for. Not every detjiil in the complex of details under observation is pertinent to the matter in hand ; and yet no a priori rules can be given to direct an investigator. Much depends on experience and skill not subject to rules. But by analyzing successfully conducted inductions, logicians are able to formulate several important niles of procedure, which thus are a posteriori. Vox our present purposes we can do no better than quote .John Stuart Mill's canons of induc- tion. These state the conditions under which generalization may be made. Hence the observer should be on the watch for the occurrence of these conditions. The canons are as follows: First Canon, for the Method of .Agreement: If tiro or more itifitancrs of the phrnomrnnn under invctti/jation hare onli/ one circunifitanre in common, the cir- cumstance in irhieh alone all the innlaneea atirre, is the cause (or effect) of the tfiren phenomenon. Second Canon, for the Method of Difference: // an instance in which the phenomenon under inveslifjdiion occurs, and an instance in which it does not occur, hare ever;/ circumstance in common .9(ir<' 07ie. that one occurring only in the former, the circumstance in which alone the two in.stances differ, is the effect, or the cause, or an indispensalile part of the cause, of the phe- vomcnon. Third C:inon, for the .Joint Method of Agreement and Difference: If two or more in- stances in which the jihenomenon occurs have only one circumstance in common, while two or more instances in which it does not occur have nothing in common save the absence of thnt cir- cumstance; the circumstance in which alone the two sets of in-ttaners differ, is the effect, or the cause, or an indisprnsahle part of the cause, of the phenomenon. Kourth C;inon, for the Mi'thod of Residues: Suhduet from ani/ phenomenon such part as is knoii-n b;/ prei^ious inductions to be the effect of certain antecedents, and the resi- due of the phenomenon is the effect of the re- maining antecedents. Fifth Canon, for the Method of Concomitant Variations: Whatever phenomenon varies in anil manner wherever an- other phenomenon varies in some pnrliiular man- ner, is either a cause or an effect of that phe- nomenon, or is connected tc^th it throutjh some fact of cau-iation. These are statements of the conditions mder which scientific men make in- ductive inferences. Hence in our observations for the purposes of induction we should keep the following questions in mind : Is it the ease that any instance in which the phenomenon under investigation occurs, and any instance in which it does not occur, have every circumstance in common save one? Is it the case that two or more instances of the phenomenon under inves- tigation have only one circumstance in common? Is it true that two or more instances in which INDUCTION. the phenomenon under investigation occurs have only one circumstance in conunon, while two or more instances in which it does not occur havit nothing in common save the absence of that circumstance? Is a part of this phenomenon under investigation known to be the ellecl of certain causes f Does the phenomenon under in- vestigation vary in any manner whatever when- ever some particular circumstance varies in some particular manner? In case any of tlii-se ques- tions can l>e answered in the allirmativr. an induction may be made. It must not, however, be supposed that an iiivcstigalor sets about his work with only such general clews as are fur- nished by the.se questions. In any particular investigation, a knowledge of exactly what has already been done by previous workers is indis- pensable. If he has this knowledge, the invi's- tigator is in a position to ask much nuue delinite concrete questions than the five we have just mentioned. But, however much more delinite his problem may be, it may always be reduced to general expression, and when so reduced it will assume the form of one of the live questions stated above, which give the headings for the treatment of the methodology of observation. .Any more specific methodology can l)e worked out only on the basis of the practiced experience of the exi>ert in any particular line of research. - 11. It has already l)een said that when any one of the live questions given in the last para- graph can be answered in the aflirmativc, wo universalize the actually observed relation be- tween phenomenon and circumstance. But now it will probably 1k> asked what warrant there is for this uiiiversalizalion. The answer is some- what as follows: To universalize is an instinctive tendency of thought. The child 'jumps at con- clusions' in verj' much the same way as the kitten jumps at a moving string. What the child does, the ordinary adult also does, and unless corrected by exnorience the probability is that every one would genenilize every observed relation. But experience does chwk the tendency in a measure. The child who generalizes the ob- served connection between shape and rebound- ingness, or color and rel)Oundingness. comes by later experience to find that his generalization is invalid. The truth of the universal judg- ment implied in his expectation is not borne out by later developments. But he is nothing daunted. He proceeds to generalize some other observed connection until at last he gets a i/en' eral law which is uncontradicted b;/ experience, and which he therefore accepts as true. The first generalizations are naive, instinctive" at- tempts to find uniformity in the world of ex- perience. The attempt in any particular direc- tion may be baffled, but only to be followed by an attempt in another direction. In other words, some attempts fail and others succeed. To succeed is to find that subsequent experience conforms and continues to conform to the ex- pectation arotised by previous experiences. To fail is to find that subsequent experience disap- points this expectation, while in either case the expectation is. from the point of view of lojric, an unrefle<tive f^>neralization. In later life consciousness becomes reflective; it looks back upon its own processes and finds that in cases of sticcessful peneralizjition certain conditions are present, while in eases of failure these conditions are absent. The same instinctive generalizing
