Feb.4,i9i8 Variability of Yields of Fruit Trees and Field Trials 263 



Table IV. — Effect of increasing number of trees per plot in scattered units 

 FOUR TREES IN A UNIT 



EIGHT TREES IN A UNIT 



28. 98ii.30 

 17. 49±I09 

 II. 78io. 89 

 14- 85 i I. 30 

 10. O9io. 96 



25. 27i2. 16 



10. 94il. 27 



11. 3S±I. 63 

 6.68ii. 13 



II. oii2. 01 



24. 99i2. 39 

 13. 20il. 69 

 10. 99il. 7S 

 10. 54il. 90 

 4-94±I-0S 



24. 27io. 77 

 12. 84io. s6 

 II. 27io. 63 

 9. S4±o. 57 

 7-9S±o.S4 



The 4-tree unit apparently gives a greater degree of accuracy than the 

 8-tree unit with the same total number of trees. This point is clearly 

 shown by the curves. In considering a total number of either 8, 16, 

 or 24 trees the curve for the 4-tree unit approaches more nearly the 

 theoretical curve than the curve of the 8-tree unit. With a total num- 

 ber of 24 trees, for example, the 4-tree units would be scattered about 

 regularly in six different places, while the 8-tree units would be located 

 in three different places. The larger number of unit plots thus gives a 

 more typical sample of the area than half as many units with twice the 

 number of trees in a unit. 



In combining both the 4- and the 8-tree units in the regular scattering 

 of ultimate plots, an attempt was made to throw both high- and low- 

 yielding small plots into a combination, although a systematic distri- 

 bution was maintained. The fact that the curve of the 4-tree units 

 drops below the theoretical in one place indicates that this attempt 

 was successful. An ordinary regular scattering of the ultimate plots 

 might not approach the theoretical curve of random sampling so closely 

 if a knowledge of the relative productivity of the soil was not available 

 before arranging the plots. A 4-tree unit might not be practical in 

 tests of cultural methods, fertilizer, or irrigation trials. In such trials 

 an 8-tree plot is usually the smallest practical unit. In the case of 

 walnuts, however, which should be planted at least 50 by 50 feet, a 

 4-tree unit in a single row would allow for a space 50 feet wide by 200 

 feet long, and, if guard rows were planted between the experimental 

 trees, the plots would be 100 by 250 feet, a very practical size upon 

 which to handle orchard machinery. For the trial of rootstocks, prun- 

 ing experiments, variety tests, etc., the 4-tree plot is a practical-sized 

 27808°--18 3 



