12 Ellenj W. Davis 



Heads 10 9876543210 



Tails 0123456789 10 



Coins 1 10 45 120 210 252 210 120 45 10 1 



Suppose, now, that any coin is ever so little more likely to 

 slip down upon a side on which it had previously slipped most 

 often. Give ten more spinnings. 



Then either 252 would get smaller, some other numbers in- 

 creasing, or, at least, it would be acquiring a tendency to do so 

 which, on constant repetition of the ten spinnings, would at 

 last show itself. 



Mark now that the tendency is constantly accelerated, grow- 

 ing with what it feeds upon. 



At last we can imagine half-heads-half -tails coming to have 

 as little likelihood and all-heads or all-tails as great likelihood 

 as you please. 



A species whose type was half-heads-half-tails has been 

 differentiated into two, of types all-heads and all-tails, formed 

 at the expense of the parent type. Chance has worked against 

 itself to produce a law. We could not at the start say whether 

 a coin would turn up heads or tails; we can now say that 

 probably any coin of the one set would show heads; any of 

 the other, tails. 



I have calculated the results of an extreme supposition as 

 to these coins. This, namely: that as they have fallen in any 

 one set of ten spinnings, so will they probably fall in the next. 



The upper line of the subjoined table gives the set number; 

 the left hand column, the number of times a coin shows head 

 in ten spinnings; the body of the table, the number of coins 

 in sets indicated at top showing heads as indicated to the left. 



142 



