DETERMINATION OF CAUSES. 419 



Dractice, when there are exceptional instances which ordi- 

 nary causes will not explain, to assume the existence and 

 operation of wider and deeper-seated causes. 



We cannot perceive forces themselves, but we can per- 

 ceive their effects, and the order of them, and are thus 

 nabled to infer the actions of those forces, and the order 

 of the actions. Our power of perceiving acts of causation 

 depends upon our consciousness of mental change ; and we 

 >erceive effects of causes, and the order of their succession, 

 Because they are acts of change and produce changes in us. 

 To discover the cause or causes of a new phenomenon 

 with certainty, the whole of its conditions and circum- 

 tances must be individually investigated. The discovery 

 >f the true cause or causes is often only arrived at at an 

 idvanced stage of a research; to discover them we have 

 usually to invent the correct hypothesis, and draw the 

 right conclusion ; sometimes this is done even before a 

 research is commenced ; at other times not until it is 

 ar advanced or nearly completed ; and usually we have 

 X) invent many hypotheses before we obtain the right one. 

 All the conditions of the phenomenon must, as completely 

 as possible, be isolated one by one, and the effect of ex- 

 luding each condition and circumstance must be carefully 

 bserved and noted. The cause or causes must exist among 

 the invariable antecedents or concomitants of the effect, 

 whether we can perceive them or not. Whatever can be 

 omitted is no part of the cause ; the cause cannot be any- 

 thing which is present in cases where the effect is not 

 produced, unless a preventive cause or condition is also 

 present ; it is often suggested by analogy, as well as by 

 observing that something varies as the effect varies. 



At the outset we should choose for investigation the 

 simplest form of the phenomenon, and we must then first 

 separate as completely as possible all the unessential and 



B B 2 



