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



in those following they are percentages of the total number of 

 anal} r ses ; in all the figures the summation groups are plotted 

 by their mean value. 



In general the departures from the fitted curve would appear 

 small in view of the fact that the analyses (1) are the work of 

 some hundreds of different chemists ; (2) that they cover a 

 period of 32 years, although the great majority fall in the 

 last 6 years of that period, during which time many changes in 

 methods of analysis were made ; (3) that there is the certainty 

 of some constant errors, and a possibility of others being 

 present ; and (4) that all the summations are not a matter of 

 pure chance. Such departures as are shown are to some 

 extent, though not wholly, due to the unit of measure being 

 below significant size. This is evident when the data are 

 plotted by the O20 per cent unit ; the irregularities are 

 much smoothed out, although the curve as a whole becomes 

 slightly asymmetric. 



The different proportions of the entire number of analyses 

 which may be expected to fall into the several groups of sum- 

 mations on the basis of the 0*15 per cent unit are given below. 

 Corresponding figures are also given for groups by differences 

 of - 25 per cent on the assumption of an entirely symmetrical 

 distribution. 



Table II. 



0'15 per cent unit 



Summation groups 



100 

 100 



99 

 100 



99 

 100 



99 

 100 



99 

 100 



99 

 100 



99 

 101 



99 

 101 



98 

 101 



98 

 101 



98 

 101 



98 



•05-100-19 

 •20-100-34 

 •90-100-04 

 •35-100-49 \ 

 •75-99-89 \ 

 ■50-100-64 ( 

 •60-99-74 \ 

 ■65-100-79 I 

 ■45-99-59 \ 

 •80-100-94 ( 

 •30-99-44 \ 

 •95-101-09 j 

 •15-99-29 J 

 •10-101-24 

 •00-99-14 

 •25-101-39 

 •85-98-99 

 •40-101-54 ) 

 •70-98-84 \ 

 •55-101-69 j 

 •55-98-69 \ 

 •70-101-84 I 

 •40-98-54 f 



Proportion 

 of analyses 



12-8$ 

 11 



25 per cent unit 



Beyond above limits 1 



Summation groups 

 100-00-100-24 

 100-25-100-49 



99-75-99-99 

 100-50-100-74 



99-50-99-74 

 100-75-100-99 



99-25-99-49 

 101-00-101-24 



99-00-99-24 

 101-25-101-49 



98-75-98-99 

 101-50-101-74 } 



98-50-98-74 [ 

 Beyond above limits 



Proportion 

 of analyses 



20-8$ 



17-0 



10-5 



6-0 



3-1 



1-3 



•6 



2-2 



