H. H. Robinson — Chemical Analyses of Igneous Rocks. 265 



The general distribution curve (fig. 1) is based, of course, on 

 the work of many analysts. Some of these, however, have 

 contributed much more greatly than others. This is partic- 

 ularly true of the chemists of the United States Geological 

 Survey ; fully one-third of all the analyses here used have been 

 made. by them. Thus a distribution curve for the summations 

 of the analyses made by the Survey chemists may be drawn. 

 Such a curve for the work of five Survey chemists is shown 

 by I, fig. 2. For comparison the distribution curve (II) for 

 the work of all chemists, exclusive of those five analysts, is 



Fig. 2. 



Z2. 1J37 ft? 62. <jf87 /Q0./2. /Q0.37 /00.6Z J00.Z7 /O/./Z 



added. The number of analyses used for I is 1,012 ; for II it 

 is 2,497. For convenience the analyses are grouped by differ- 

 ences of 0*25 per cent, the significance of which unit has been 

 noted in an earlier paragraph. 



Clearly the Survey chemists have a much larger proportion 

 of summations at or near the maximum than have other 

 chemists. The relative percentages of analyses falling in the 

 several summation groups are as given on the following page. 



The superior results as to summation shown by the chemists 

 of the U. S. Geological Survey are due to several reasons 

 among which not the least is the experience gained from con- 

 tinuous work. Also, poor analyses (supposing such may 

 occasionally be made) are evidently not allowed to leave the 

 laboratory until their errors have been found and corrected. 

 And it may be said that if all chemists followed this same 



