CHEMICAL CHANGES DURING DECOMPOSITION. 



515 



It is therefore clear that even the large losses of the elements repre- 

 sented by the above tables are much too small. The amount of the error 

 in a given case for a certain element is large in proportion as the element 

 is relatively insoluble. For instance, in the phonolite, where the potassium 

 is considered fixed, the loss in alumina is given as 1.60 per cent; but if the 

 loss of the potassium is, for instance, 30 per cent — and it is probably more 

 than this — the loss of the alumina was more than 30 per cent. But in the 

 case of the sodium, the loss of which is calculated as 66 per cent, this 

 amount may be considerably too small, but it can not be more than 100 

 percent; hence the relative, error is much less in this case. For instance, 

 in the diabase (Table V) the loss of sodium, 98.57 per cent, is almost, and 

 of potassium, 100 per cent, is quite, correct. Where the losses are given 

 as small in the foregoing tables, therefore, it is probable that the calculated 

 amounts are but a fraction of the real losses ; but where the losses are given 

 as large, the errors, while very considerable, are probably not so great 

 proportionally, and the calculated amounts give a rough approximation to 

 the minimum percentage of depletion of the rocks in these elements. 



In order to show the calculated comparative losses of the six important 

 elements, aside from aluminum, the following table is compiled for those 

 rocks in which the calculation has been made upon the supposition that the 

 aluminum is fixed. 



Table showing calculated jiercentage losses of important elements on the supposition 

 that the aluminum is constant. 



(i Averages of potash and soda made from eight tables, asi 

 and Na. 



the fifth it is impossible to state relative amounts of K 



