XCii PROCEEDINGS OE THE GEOLOGICAL SOCIETY. [May I9O9, 



The chemical composition was obtained by taking the average of 

 such analyses as were available at the time, but no information is 

 given as to the details of the calculation : we do not know how the 

 analyses were weighted, we do not even know the names of the 

 rivers which furnished them. On the whole it would seem that 

 of all the data on which we have to depend this is the least 

 trustworthy. It gives 24,106 tons of sodium per cubic mile 

 of river-water, and we thus have a total annual delivery of 

 24,106 x 6524 == 15,727 X 10 4 tons nearly, and consequently 



14,000 xlO 12 __ ina . .. 



15,727X104 = 90X1 ° a PP roximatel y> 



or the time which has elapsed since the first formation of the Ocean 

 amounts to about ninety millions of years. 



After a very full discussion of the various factors which might 

 modify this result, Joly was led to assume ninety-six millions of 

 years as the most probable estimate. 



It may be worth while to probe this method x a little deeper, 

 especially as regards the quantity of sodium brought down by 

 rivers. 



To obtain a completely satisfactory result, we require to know 

 the average volume cud composition of all the important rivers of 

 the world ; failing this, we might endeavour to obtain an average 

 by a thorough study of the rivers of a single continent. Let us 

 make the attempt, selecting North America, as likely to afford us 

 the largest amount of information. 



The data we require are (1) the average annual discharge of 

 each river, as determined by measurements made near its mouth 

 over a period of about forty years ; and (2) the average annual 

 composition, as determined by analyses made at the same stations 

 and at the same times as the measurements of discharge. 



As regards (1) we possess, except in the case of Canada, a 

 number of sufficiently trustworthy averages; but, unfortunately, 

 it is not always possible to make use of them in connexion 

 with (2). For, though there are many rivers of which the average 



from the calculations. When these are taken into account the total land-area 

 found by my method is 56,046,730 square miles, an excess of about 1*2 per 

 cent, over the true area. This gives a check on the correctness of the results. 

 It will be seen that if we take 025 instead of 0'222 as the fraction representing 

 the run-off, the final result is not very different from Sir John Murray's. 



1 For an interesting criticism, see Rev. O. Fisher, Geol. Mag. dec. 4, vol. vii 

 (1900) pp. 124 & 132. 



