DARWIN, AND THE ORIGIN OF SPECIES 1 85 



is that which takes its departure from that conception of specific 

 adaptive types, which leads us to seek the origin of the " events " 

 that exhibit the type in the physical properties of living matter, 

 and to seek the origin of the statistical " law of error " which forms 

 an adaptive type out of these events in the order of external nature. 

 I shall now try to illustrate the way in which the order of nature 

 forms adaptive types, or zoological species, out of the events which 

 are afforded by the properties of living beings. 



• While no one doubts that the paths of all the drops in a shower 

 of rain might have been predicted from mechanical data, one who 

 knows none of these data may, by an umbrella, make determinate 

 the paths of all the drops which immediately concern him. In statis- 

 tical language we may say that, even if we know nothing of the 

 causes of the events, we may make an adaptive type out of them by 

 means of an umbrella ; and if we move the umbrella to another 

 place, we may make a new type, identical in adaptive value, out 

 of a different set of events ; for the causes of the events have 

 nothing to do with the use to which we put them, except in this, 

 that we could not use them unless they occurred. 



If before a long line of machine guns, scattering bullets to all 

 quarters of the field, we set up a target, exhaustive knowledge of 

 machine guns might enable one to say how many balls will strike 

 it in a given time, and how they will be distributed, but, as we use 

 words, we say certain balls chance to hit, for the target does not 

 affect in any way the course by which a ball reaches it. If we 

 now put before the target a screen with a hole in it, no one safely 

 behind the screen would wish to show his face at the hole incon- 

 siderately, since, so far as it affects him, the course of the balls which 

 concern him has been made determinate. 



Now imagine an unlimited series of similar screens set in line, 

 each within range of the next, and suppose, furthermore, that while 

 each ball that hits a screen drops and is lost ; each one that goes 

 through a hole grows into a new machine gun. No two objects, 

 natural or artificial, are exactly alike, and among the original 

 machine guns some would put more balls through the first 

 screen, and have more descendants than others, even if they 

 had been set, one after another, in the same place before the 

 target. 



