2IO • Bulletin 307 



which shows either that the individuals of the sample chosen are not 

 fairly representative of the population, or are so variable that the mean 

 has little importance. 



Coefficient oj variation. — The mean and standard deviation describe the 

 absolute values of individuals of one group. But in order to compare the 

 values of one group with another, it is necessary to have a measure of 

 relative variability. Such a measure is found by dividing the standard 

 deviation by the mean; the result is called the coefficient of variation. 



Comparison by percentages 



Percentage frequency. — A simple frequency distribution, or frequency 

 curve, describes the absolute values of the individuals of the group for 

 which it is made. But if the frequency of the observed cases is reduced 

 to percentages, a means is given at once for comparing the values of one 

 group with another. 



Percentage distribution. — A scheme of distribution enables one to see 

 the proportion of individuals of the entire group that is measured by 

 any value or lies between two values. Such a scheme may be represented 

 graphically as in Plate II, or by a single line or column of figures, as in 

 Table 2. When such a scheme of distribution is made for a sufficiently 

 large number of cases, it is useful for determining the rank of any single 

 individual of the population. By looking for the grade of the individual 

 under consideration, we may see what percentage of the population lies 

 above it and what percentage lies below it. If comparison is made from 

 year to year, the individual gain or loss may be determined. Schemes of 

 distribution for different groups offer a good basis for comparison of dif- 

 ferent populations. 



THE ORCHARD SURVEY FROM THE STATISTICAL STANDPOINT 



The value of an orchard survey. — In making an orchard survey we are 

 trying to determine the value of various types of orchard practice. Each 

 orchard, therefore, might be considered an experiment plot in which the 

 different methods of orchard management are the experiments. To be 

 sure, the work may not be conducted so scientifically as at an experiment 

 station; but the fact that there are many experiments along the same 

 lines and extending over a great territory counterbalances the scientific 

 precision of a few more or less artificial experiments. When large numbers 

 of orchards are considered, the law of chance operates to overcome the 

 local variations of individual orchards. 



Value of results. — The value of results in survey statistics depends upon : 

 ist, the accuracy of the field data; 

 2<i, the number of observations; 



