TIIK LAWS OF SCIENCE 41 



" steel " implies that these properties are invariably 

 associated, and it is just these invariable associations, 

 whether of " properties " or of " events," that are 

 expressed by laws. In the same way " rust " implies 

 another set of associated properties and another law ; 

 rust would not be rust, unless a certain colour were 

 associated with a powdery form and insolubility in water. 



And we can proceed further and apply the same analysis 

 to the ideas that were employed in stating that the 

 properties of steel are invariably associated, or, in other 

 words, that there is such a thing as steel. For instance, 

 we spoke of a magnet. When we say that a body is a 

 magnet, we are again asserting an invariable association 

 of properties ; the body will deflect a compass needle 

 and it will generate an electric current in a coil of wire 

 rotated rapidly in its neighbourhood. The statement 

 that there are magnets is a law asserting that these 

 properties are invariably associated. And so we could 

 go on finding that the things between which laws assert 

 liable relations are themselves characterized by 

 other invariably associated properties. 



This, then, is one of the ways in which laws may be 

 both the original subject-matter and the final result of 

 ice. We find that certain events or certain propert ies, 

 A and B, are invariably associated ; the fact that they 

 are so associated enables us to define a kind of object, 

 or a kind of event, which may be the proper sub] 

 matter of science. If the object or the event ( 

 of A and B without any invariable association bet 



might be a particular object or a particular 

 md might form an important part of the 

 popular conception of the external world, but it 

 be proper subject II UN 



D Napoleon and the battle of \\ 

 bjcct and an event consisting of various propci 

 and events ; but these properties and events are not 



