258 



Journal of Agricultural Research 



Vol. XII, No. 5 



The biometrical constants for the several plantations are given in 

 Table II. The oranges, lemons, and apples, as might be expected, show 

 less variability than the seedling walnuts. The coefficient of variability 

 of the clonal varieties ranges from 29.72 to 41.23 per cent. This total 

 range of only 11. 51 per cent shows a marked similarity of the extent of 

 variation. 



Table II. — Variability in yield of the different individual fruit trees 



Kind of fruit. 



Naval orange (Arling- 

 ton 



Naval orange (Ante- 

 lope Heights) 



Valencia orange 



Eureka lemon 



Seedling walnut 



Seedling walnut 



Seedling walnut, aver- 

 age 



Jonathan apple 



Standard 

 deviation. 



Pounds. 

 54- 42 ± 0.8a 



55- 33 ± I- 19 

 97- 84±3- 01 

 81.38i2.03 

 47-77±l-36 

 41-94±I-I9 



40. io± 1. 14 



125. 30±4. CO 



Coefficient of 

 variability. 



39- 55 ±o- 68 



29. 72±o. 69 

 39- 72 ± I- 40 



30. o6±o. 8i 

 47. 86± I. 64 

 53- 91 ±1-92 



46. 41 ±1-58 



4i-23±l. S2 



Probable error. 



Pounds 

 per tree. 



Percent- 

 age of 

 mean. 



26.67 



20.05 

 26.79 

 20. 28 

 32-28 



31-30 

 27-81 



The probable error, expressed in pounds of fruit per tree, is the greatest 



in case of the Jonathan apple, amounting to 85 pounds, while the Valencia 



orange and the Eureka lemon fall to 66 and 55 pounds, respectively. 



Such probable errors, expressed in pounds per tree, are not comparable, 



however, unless the mean yields are approximately the same. The 



probable error expressed as a percentage of the mean is therefore added 



to Table II to make it more easily compared with tables of other writers 



who have seen fit to use this constant rather than the coefficient of 



variability. 



METHODS OF CALCULATING VARIABILITY 



The yields of the various fruit plantations have been studied, with 

 trees singly and combined into plots of various sizes. The coefficient of 

 variability and probable error have been used as the basis of comparison 

 in most cases. 



Plots of different sizes necessarily have varying mean yields per plot; 

 therefore the coefficient of variability is more readily interpreted than 

 the standard deviation. The probable error may only be used with 

 accuracy in cases where the number of variants is relatively large and 

 their distributions normal. 



VARIABILITY OF ORCHARD PLOT YIELDS 



EFFECT OF INCREASING THE NUMBER OF ADJACENT TREES PER PLOT 



The first point studied was the effect of increasing the number of 

 adjacent trees per plot, measured by the coefficient of variabihty. Based 

 on the theory of random samphng of variables, the average production of 10, 

 or even 5, trees should be a more typical sample of the orchard than that 



