490 Subsurface Geologic Methods 



tion is sampled. For heterogeneous rocks, the entire sample must be re- 

 duced to a reasonably fine size by careful crushing; mixed, sampled, and 

 this sample reduced further in particle size; then remixed and resampled, 

 the process being repeated until the operator is satisfied that the few milli- 

 grams taken for analysis are representative of the entire sample. For 

 relatively homogeneous rocks such as compact limestones the sampling 

 requirements are perhaps less rigid than those for rocks containing a var- 

 iety of minerals of different crushing hardnesses and different sizes of liber- 

 ation, but the operation is nevertheless tedious. 



The only satisfactory method for use in analyzing nonconducting 

 powders consists in placing the powder within a cup bored in an elec- 

 trode made of some conducting material. In this case the arc (the high- 

 tension spark cannot be used) is struck between the electrode holding the 

 powder and another electrode of the same material. The high temperature 

 reached by the cupped electrode (the anode or positive electrode) gradu- 

 ally volatilizes the constituents of the powder, and the spectrum lines of 

 the constituents of the powder are excited in the arc column. High-purity 

 silver and copper have been used as electrode materials, but the most 

 convenient and easily obtained substance is graphite. This is available 

 in regular shapes, usually as rods of various dimensions, at reasonable 

 cost, and of very low impurity content. It has the further advantage of 

 easy machinability. 



For quantitative analysis all powder samples must be weighed. This 

 is tedious work, and any method that reduces time spent in weighing with- 

 out materially sacrificing accuracy is to be commended. The Roller-Smith 

 spring balance has been found most useful in this respect, for it is ex- 

 tremely rapid in operation and is quite consistent in behavior. Samples 

 weighing five milligrams can be weighed with an accuracy of 0.2 percent. 

 Specimens can be loaded into the cupped electrodes directly from the 

 balance pan through a small glass funnel or by means of a shaped assay- 

 silver or platinum launder. Even with these aids, however, the time con- 

 sumed in weighing is unduly great. It should be possible to expedite weigh- 

 ing and loading by using a machine that produces pellets of the rock 

 powder. As the accuracy of spectrochemical analysis is not of a very 

 high order, limited accuracy in reproducibility of pellet weight would be 

 acceptable. Care would be necessary that pellets were not contaminated 

 by dust from preceding specimens. 



Arc analysis of limestone powders presents an unusual problem. The 

 rapid heating in the arc gives rise to copious evolution of carbon dioxide, 

 which, if the lower electrode is poorly designed, tends to blow part or 

 all of the sample from the cup. This difficulty has been overcome by 

 using a relatively deep cup -^ so that the carbonates are calcined at low 

 temperature before volatilization of the constituent elements commences. 



The increased exposure time thus enforced adds considerably to gen- 



^' Slo89, L. L., and Cooke, S. R. B., op. cit. 



