16 BULLETIN 551. U. S. DEPARTMENT OF AGRICULTURE. 
SUMMARY. 
In this paper there are presented the complete analyses of 45 
samples of soil, representing four soil provinces. These, together 
with the analyses of 2-1 samples previously published representing 
also four provinces, are discussed from the points of view of extreme 
variation of all the samples, variation in composition within a soil 
province, variation of the same type, and the bearing of the limit 
of error hi analysis on the interpretation of analytical data. 
It is thought that the analyses discussed represent nearly the 
extremes in composition of soils in the regions in which the samples 
were taken. Marked resemblances in composition of soils from the 
same province are pointed out. It is shown that some samples of 
the same type differ considerably in chemical composition. It is also 
shown that some soils of different types may resemble each other 
in chemical composition as closely as different samples of the same 
type. 
It is pointed out that the unavoidable error in analytical operations 
is in many cases of such magnitude that when analyses are stated in 
pounds per acre differences of several hundred pounds of some con- 
stituents are not significant. 
ADDITIONAL COPIES 
OF THIS PUBLICATION MAT BE PROCURED FROM 
THE SUPERINTENDENT OF DOCUMENTS 
GOVERNMENT PRINTING OFFICE 
"WASHINGTON, D. C. 
AT 
5 CENTS PER COPY 
