CONCLUDING REMARKS. 197 



natural phenomena until those generalizations were sure and 

 unobjectionable in other words, were laws science would 

 be lost in a complex mass of unconnected observations, 

 which would probably never disentangle themselves. Excess 

 on either side is to be avoided ; although we may often err on 

 the side of hasty generalisation, we may equally err on the 

 side of mere elaborate collection of observations, which, 

 though sometimes leading to a valuable result, yet, when cu- 

 mulated without a connecting link, frequently occasion a cost- 

 ly waste of time, and leave the subject to which they refer in 

 greater obscurity than that in which it was involved at their 

 commencement. 



Collections of facts differ in importance, as do theories : 

 the former, in many instances, derive their value from their 

 capability of generalisation ; while, conversely, theories are 

 valuable as methods of co-ordinating given series of facts, 

 and more valuable in proportion as they require fewer excep- 

 tions and fewer postulates. Facts may sometimes be as 

 well explained by one view as by another, but without a 

 theory they are unintelligible and incommunicable. Let us 

 use our utmost effort to communicate a fact without using the 

 language of theory, and we fail ; theory is involved in all 

 our expressions ; the knowledge of bygone times is imported 

 into succeeding times by terms involving theoretic conceptions. 

 As the knowledge of any particular science developes itself, 

 our views of it become more simple ; hypotheses, or the in- 

 troduction of supposititious views, are more and more dis- 

 pensed with ; words become applicable more directly to the 

 phenomena, and, losing the hypothetic meaning which they 

 necessarily possessed at their reception, acquire a secondary 

 sense, which brings more immediately to our minds the facts 

 of which they are indices. The scaffolding has served its 

 purpose. The hypothesis fades away, and a theory, or gen- 

 eralised view of phenomena, more independent of supposition, 

 but still full of gaps and difficulties, takes its place. This in 



