Feb.4,i9i8 Variability of Yields of Fruit Trees and Field Trials 247 



Recent investigations, which are reviewed in succeeding pages, have 

 thrown light upon the nature of the variability inherent in experiments 

 conducted with groups of plants or of animals. So far as the writers 

 know, however, few studies have yet been published upon the uncer- 

 tainty with which one deals in attempting to carry on plot trials with 

 orchard trees. In connection with plans for extending the trial plots of 

 this Experiment Station, the writers have attempted to study the ques- 

 tion of variability of tree yields and to formulate some plan to determine 

 the probable reliability of the results. It is highly desirable to make 

 the tree plots as small as possible without sacrificing too much accuracy 

 on account of the large amount of land required for each individual. 



A fruit tree is possibly more affected by environmental conditions than 

 an annual crop growing from seed to maturity in one season. The tree 

 roots penetrate more deeply into the soil and may be affected by soil 

 differences to a considerable depth. Climatic conditions during the rest- 

 ing period may have marked influence on the crop production of the trees. 



Pruning also introduces variation. Uniform pruning is desirable; 

 yet different trees need different types of pruning, and more uniform 

 results will be obtained if this is recognized. 



Individual trees have apparent idiosyncrasies in fruiting habits. A 

 tree may yield large crops of fruit in alternate years, with very light 

 crops in intervening years. The character of the stock upon which the 

 particular variety was budded to form the tree may profoundly influence 

 the type, habit, and productivity of the adult tree. Furthermore, we 

 can not overlook the possibility of errors in yields due to predacious 

 animals which devour or otherwise destroy more or less fruit. 

 Fruit trees therefore present opportunity for more variability than 

 would be expected in the growth of annual plants. 



PREVIOUS STUDIES 



The varying productivity of fruit trees has been called to our attention 

 by the writings of Macoun (1904),^ Munson (ipoy), Shamel (1912), 

 Hedrick (1912), Stewart (191J, p. 552-554), Fletcher {1913), Coit (Jpjo), 

 Whitten (19 15), Lewis and Vickers (19 15, p. 30-31), Barre {1915), 

 Gourley {19 15, p- 72-73), Kraus (19 16), and others. The above observa- 

 tions have been largely made in connection with plant breeding and 

 orchard management problems, and were not made to bear necessarily 

 on the accuracy of plot trials. 



Holtsmark and Larsen {1906) were among the first to call attention to 

 the errors of field trials. They recognized the inevitable variation of 

 field results, and showed how it may be estimated by the use of the 

 standard deviation and the coeflficient of variability. They also showed 



1 Bibliographic citations in parentheses refer to "Literature cited", pp. 382-283. 

 27808°— 18 2 



