30 



cover those vastly divided particles of the water, which are not 

 influenced by the still more minutely divided portions of the acid. 



59. But a larger is sometimes totally changed, to all appear- 

 ances, by a lesser quantity : here the proportions of the real 

 agents are in fact the same. Thus a lesser agent may be diffused 

 in the way just described among the particles of a larger, ap- 

 parently changing the qualities of the whole; and another agent 

 being combined equal to the lesser already combined, may have a 

 relation of influence only with the lesser, and by destroying its 

 influence in the general mass, may appear to aftect the whole 

 quantity, so much greater than itself. 



60. There is yet another, and perhaps a more striking class 

 of examples of an influence communicated by a minute, apparently 

 to the whole of a considerable, quantity: I allude to those of con- 

 version. Thus a single spark of fire will ignite a whole barrel of 

 gunpowder. This cannot be, according to our law of proportions, 

 by a communication of the divided portions of the spark to all 

 the divisible particles of the gunpowder: what then is the mode 

 of this instance of causation? 



61. It is simple increase by affinity. To describe the process 

 more minutely : gunpowder contains latent fire (made latent by its 

 combination in this form); eliminated fire (as the spark) is related 

 with its own quantum of the latent fire, and unites with it by the 

 force of superior affinity. The single spark is thus increased by it* 

 union; these conjoined quantities have a similar relation with the 

 latent fire within their sphere, and the ignition of the whole is 

 produced by a quick repetition of the same process. These points 

 will be still further scrutinized when we come to speak of the sub- 

 jects to which they apply. To proceed at present with the other 

 topics of causation: 



62. An effect is identified by its causes; but it is so closely 

 connected with other things, that, from meeting with them con- 

 tinually in this close connection with the effect, we are apt to 

 regard them as causes. Every individual cause of an effect, 

 unfortunately, is liable to have these adjuncts and associations, 

 which, from the regularity and frequency of their alliance, are apt 

 to be mistaken for the causes which are necessary to the existence 

 of the effect. There is no way of evading this obstacle to a just 

 discrimination in our pursuit of science, but by defining exclusively 

 the particular effect we mean to consider, and then examining its 

 real dependences. Thus, for example, we may say, atmospherical 

 air maintains life, or is a cause of life; if atmospherical air should 

 be found to consist of three constituents, the real relation with, or 

 real cause of, life may be only one of these; while, previous to an 

 analysis of air, the whole properties, or their combined result, may 

 be assigned as the cause of life; or, after such an analysis has been 

 made, the concurrence of these three properties may be assigned 

 as necessary to life, and they may, from the effect which they pro- 

 duce in their union, be equally considered as causes, until the 



