310 SEE— THE CAUSE OF EARTHQUAKES. [Octobe. ,9, 



probability that the direction of a piece of the chain will fall in any 

 one of these divisions of the right angle is one-tenth ; and the prob- 

 ability that all the n pieces throughout the whole chain will fall in 

 the same angular division is (i/io)". When n is a large number, or 

 the chain is long and close to the sea, this probability becomes prac- 

 tically zero. In such a range as the Andes it is observed that all the 

 pieces of the chain do fall in the first division of the angle, between 

 0° and 9°; and, hence, in general, the probability is (10)": i that 

 the observed lay of the chain so exactly parallel to the shore of the 

 sea is not the result of mere chance, but depends directly on some 

 physical cause which has made the chain essentially straight as well 

 as laid out the general course parallel to the sea coast. 



If the chain had the short bends in it here assumed to be pos- 

 sible, the total length would be greater than that of the existing 

 chain, and n would be correspondingly increased. The data used, 

 therefore, make n a minimum, and P= 1/(10)" a maximum for the 

 given chain everywhere so closely following the seashore. Thus 

 the calculated value of P is too large rather than too small, with the 

 existing lay of the chain. 



Another way of reaching analogous results is to consider what 

 the deviation from strict parallelism is in so long a chain, and the 

 probability that the coincidence would be so exact throughout, when 

 all angles between i" and 324,000" (the equivalent of 90°) are 

 equally probable. If no physical cause is involved depending on 

 the sea, there is no reason why the chain should not run at any angle 

 across the shore line. Now, the Andes are made up on the average 

 of at least two parallel chains, and, the probability of this double 

 parallel trend throughout would be only 



II I 



324,000 324,000 114,976,000,000 



In any case, we see that for the double chain, the chances are hun- 

 dreds of billions to one against strict parallelism to the seashore. 

 In the first method of treatment each part of the chain is con- 

 sidered, without regard to the rest ; in the second method, the chain 

 is viewed more as a Avhole, and the individual parts neglected. In 

 respect to the first method, it might be claimed that as mountains 



