422 ACTUAL WORKING IN ORIGINAL RESEARCH. 



circumstance is the cause of that phenomenon. The 

 instances must not diger in more than one particular, for 

 if they vary in several respects we cannot tell to which 

 the effect is due. The great rule for discovering causes 

 by means of experiment is to vary only one circumstance 

 at a time, keeping all the others just as they were. If by 

 adding or removing a single circumstance a change occurs, 

 that circumstance is either a cause or condition of the 

 event or of a part of it ; but if we add or remove two 

 circumstances at once, we cannot be sure that the two 

 changes had not neutralised each other's effect. Usually 

 we find that there are several circumstances, on excluding 

 each of which the phenomenon is affected ; and in that 

 case one or more must be the cause and the remainder are 

 conditions of the event. We employ this method largely 

 when making experiments. 



We often cannot vary only one circumstance at a time, 

 and therefore cannot employ the Method of Difference ; in 

 such cases we use the Combined Method of Agreement 

 and Difference, which is as follows : If one set of instances 

 has only one circumstance in common, and an effect 

 occurs ; and if another set has nothing in common except 

 the absence of that circumstance, and the effect does not 

 occur, that circumstance is the cause of that effect. 

 Usually, however, there are several inseparable circum- 

 stances which are the cause and conditions of an effect. 



The Method of Coincident Variations is If one phe- 

 nomenon varies in a particular manner, whenever another 

 varies in a particular manner the latter is probably the 

 cause of the former. The variation may be either direct, 

 inverse, or even at a different rate, provided it is invariably 

 coincident or successive. This is a powerful method of 

 discovering inseparable conditions, and we employ it 

 when we compare series of constants in order to detect 



