420 ACTUAL WORKING IN ORIGINAL RESEARCH. 



uninfluencing circumstances, and we can judge to some 

 extent what those are likely to be, by means of our previous 

 knowledge of science. To know the cause of failure or of 

 interference is often the greatest step towards ensuring 

 success. When we possess no knowledge of the circum- 

 stances attending an event, we can form no idea respecting 

 its cause. We must not assume without sufficient evi- 

 dence either dependence or independence of the pheno- 

 menon upon any one of the conditions or circumstances 

 present, nor the absence of all interference ; because many 

 of the most obvious circumstances attending a phenomenon 

 have often no real connection with it. Even two variations 

 may follow the same law, and yet be the result of different 

 causes, and not be actually connected ; and, on the other 

 hand, the most obvious circumstances may be the real 

 cause. The only invariable antecedent or antecedents of 

 a phenomenon constitute the probable cause of it ; there 

 may, however, be a hidden or latent cause ; and also a 

 plurality of causes. Every cause which follows a different 

 law of increase or decrease to that which produces the 

 true effect has to be excluded, because it disguises the 

 one we are examining ; and, if desirable, this interfering 

 cause may be separately investigated. Of the insepa- 

 rable conditions and circumstances which now remain, 

 we must ascertain which are united by the bonds of causa- 

 tion and necessary condition, and which by coincidence only. 

 If, after separating all the causes which appear to interfere 

 with the effect, the same conditions do not uniformly pro- 

 duce the same result, there must remain some unobserved 

 difference ; therefore, we should never omit to describe every 

 circumstance which accompanies the effect; and if we cannot 

 readily discover the cause, we should make radical changes 

 in the experiment. If also, after having in any case 

 excluded all the results of known causes, an effect still 



