430 THE PRINCIPLES OF SCIENCE. 



man and his Creator. The course of nature is regarded 

 as being determined by invariable principles of mechanics 

 which have acted since the world began, and will act for 

 infinite ages to come. Even if the origin of all things be 

 attributed to an inteUigent creative mind, that Being is 

 regarded as having yielded up arbitrary power, and as 

 being subject like a human legislator to the laws which 

 he has himself enacted. Such notions I should describe 

 as superficial and erroneous, being derived, as I think, 

 from false views of the nature of scientific inference, and 

 the degree of certainty of the knowledge which we acquire 

 by inductive investigation. 



A law of nature, as I regard the meaning of the 

 expression, is not a uniformity which must be obeyed by 

 all objects, but merely a uniformity which is as a matter of 

 fact obeyed by those objects which have come beneath 

 our observation. There is nothing whatever incompa 

 tible with logic in the discovery of objects which should 

 prove exceptions to any law of nature. Perhaps the best 

 established law is that which asserts an invariable cor 

 relation to exist between gravity and inertia, so that all 

 gravitating bodies are found to possess inertia, and all 

 bodies possessing inertia are found to gravitate. But it 

 would be no reproach to our scientific method, if something 

 were ultimately discovered to possess gravity without in 

 ertia. Strictly defined and correctly interpreted, the law 

 itself would acknowledge the possibility ; for with the 

 statement of every law we ought properly to join an esti 

 mate of the number of instances in which it has been 

 observed to hold true, and the probability thence calcu 

 lated, that it will hold true in the next case. Now as we 

 before found (vol. i. p. 299) no finite number of instances 

 can warrant us in expecting with certainty that the next 

 instance will be of like nature ; in the formulas yielded 

 by the inverse method of probabilities a unit always 



