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



Table I. 



The number of analyses from 97"00^ to lOS'SO^ for groups of summations 

 differing by 0*05 per cent. 



Summations. 



97 



98 



99 



100 



101 



102 



103 





00- 



04 



1 



2 



19 



139 



26 



3 







05- 



09 







15 



143 



29 



4 



__ 





10- 



14 



1 



1 



29 



144 



10 











15- 



19 





4 



22 



151 



20 









20- 



24 





1 



30 



128 



12 



3 



._ 





25- 



29 



1 



1 



29 



131 



18 







_. 





30- 



34 







4 



34 



133 



12 



2 



1 





35- 



39 







1 



36 



114 



17 



2 



__ 





40- 



44 





2 



49 



83 



8 



1 



1 





45- 



49 



1 



1 



42 



88 



9 



2 



1 





50- 



54 







4 



49 



90 



2 



1 







55- 



59 





3 



47 



77 



7 



3 







60- 



64 



1 



4 



71 



69 



5 



1 



.. 





64- 



69 



1 



3 



82 



69 



3 



__ 







70- 



74 



1 



4 



95 



67 



3 



1 



.. 





75- 



79 





6 



96 



59 



6 







.. 





80- 



84 







5 



116 



48 



3 



1 







85- 



89 



1 



2 



119 



43 



2 







.. 





90- 



94 



2 



4 



142 



36 



1 









•9S--99 



-- 



8 



131 



34 



2 



-- 



-- 



Totals 



10 



60 



1253 



1846 



195 



24 



3 



Gn 



md 



total 



3391 















sufficient analyses, their effect on the general average is equal 

 to zero. Washington* gives figures for the allowable varia- 

 tion in duplicate determinations of the various oxides, all of 

 which may be either plus or minus, in which case they would 

 tend, of course, to balance one another purely as a matter of 

 chance. Dittrich, f on the basis of special determinations, 

 decided that errors of a certain size and direction were to be 

 expected for six of the common rock elements expressed as 

 oxides. However, part of the analytic work was not done by 

 Dittrich and was poor, and as a whole it hardly warranted 

 quantitative conclusions. The qualitative statement of the 

 results is that a minus error is to be expected for alumina, a 

 plus error for ferric iron, and a plus or minus error for lime, 



* Washington, H. S., Chemical Analysis of Kocks, p. 24, 1904. 

 f Dittrich, M., Ueber Genauigkeit von Gesteinanalysen, Neues Jahrb. f. 

 Min. u. Geol., vol. ii, p. 69, 1903. 



