REPORT OP THE STATISTICIAN. 363 



period. In wlieat it varies little from 12 bushels per acre, usually a 

 fraction above in any period taken. Corn varies little from 20 

 busliols, Yot there is a confusion of ideas of this term ''average." 

 Some, instead of reporting the average of all crops, good and bad, 

 refer to an ideal standard of production; a full crop, represented by 

 100, as in our basis of condition. In the course of time averages may 

 change with improved cultivation, but the constant enlargement of 

 area, by extension of settlement, prevents much change in the United 

 States. 



Having decided on what is an average, the old-style reporter makes 

 his returns. Instead of precision of expression, capable of exact 

 mathematical rendering, he rexDorts in language like this : Barely 

 average ; fairly good ; moderate ; middling ; light ; variable ; very 



Eoor ; indifferent ; promising ; much above average ; very good ; 

 eavy crop ; and every imaginable form of expression, utterly im- 

 possible of formulation in figures, because nine men out of ten might 

 give a different jiercentage value to nearly all of them. If every re- 

 porter were infallible in judgment, his form of expression can never 

 he interpreted by others, and therefore what is true in his judgment 

 is false as rendered. 



This is so notably the case that, unless the compiler makes a false 

 pretense of an ability to interpret language that is inexpressible in 

 figures, he contents himself with arrangement of returns in three 

 classes: one an "average," one above it, and one below. All that such 

 returns can show, therefore, is whether a larger number of persons 

 report "above" than the number reporting below. Should 80 per- 

 sons, for instance, report a prospect above average and 70 a lower 

 condition, it would popularly be assumed that the crop was slightly 

 above average. It might be, unless the 70 below average representee! 

 a larger area or rate of yield than the 80, in which case the returns 

 would really indicate less than an average crop. As there is nothing 

 to indicate how much above or below an average a return may be, it 

 is impossible to know whether the 80 returns mean as much as the 70. 

 So tliis form of return is nearly meaningless. It is not sufficiently 

 defiiiite; it does not express such discriminative judgment as is quite 

 possible to render. Some reporters, aware of this absurdity, esti- 

 mate in definite proportions: half, two-thirds, or three-fourths of an 

 average crop. This is correct in principle, but not sufficiently close; 

 it is only a little less absurd; for any man of good judgment ought to 

 be able to estimate within 5 per cent, of the trutJb., and to discrimi- 

 nate more accurately than by jumps of 15 to 25 per cent, in an esti- 

 mate. 



Such indefiniteness of expression results from these clumsy methods, 

 which foster carelessness and inaccuracy in crop estimates. This 

 loose and worthless method of reporting has had its day, and should 

 be discarded. 



It can be demonstrated, from experience with statistical returns, 

 even when they are mathematically and accurately rendered and con- 

 solidated, that there is either tendency to make a full crop the "aver- 

 age" crop, or else to make the comparison with an "average" stand- 

 ard somewhat too low. There is a practice out West of comparing 

 with a five years' average, i. e. , the real average of the good and bad 

 crops of the last five years; which would be an abnormally low aver- 

 age in the case of corn and wheat from 1880 to 1885. As a matter of 

 fact, half of such returns appear to be made on the basis of a full 

 crop instead of eon. average of several more or less crippled harvests, 



