CROPPINC. SYSTEMS FOR TOBACCO 21 



Appendix 



Note on Statistical Interpretation of Data 



By taking sufficiently wide class intervals it was found that the yield data gave approximately 

 normal distribution curves, and were thus capable of statistical treatment. The mean yield of each 

 plot for the period 1926-1931 was considered as a unit in the calculation of statistical constants. 

 Thus the effect of seasonal variation was eliminated or reduced in a manner suggested by Engledow 

 and Yule.* The following formulae were then used in calculating constants: 



Standard deviation 



_ / Sd' 



Probable error of a single plot = i_ .67-15 standard deviation 

 Probable error of the mean (p. e.) = +_ .6745 standard deviation 



v 



The following constants were obtained : 



Mean Standard Probable error of Probable error 



Cropping system yield deviation single treatment of the mean 



lb. lb. lb. % lb. % 



No cover (check) 1603 76.0 51.2 i.l 16.2 1.0 



Manure 1693 106.6 71.8 4.2 25.4 1.6 



Timothy cover 1620 108.5 73.0 4.7 23.1 1.4 



Red top cover 1697 65.2 43.8 2.6 19.5 1.1 



Rye cover 1624 38.5 25.9 1.6 11.7 .7 



In order to see if the difference between any two means was significant, the probable error of the 

 difference was obtained by using the formula : 



Probable error of the difference = 



y'^(p.e.j)^+ (p.e.,)' 



To be considered significant, that is due to the treatment and not to chance, any difference 

 must be at least 3.8 times its probable error. 



*Engledow, F. L., and Yule, G. U. The principles and practices of yield trials. Empire Cotton 

 Growing Corporation, London (1930.) 



Explanation of the Grade and Crop Index 



Grade Imlcx: 



In comparing the quality of tobacco grown under different treatments, it is very difficult to keep 

 in mind the percentage of six or more commercial grades of tobacco from one treatment , and compare 

 them with a like number from another. To simplify the comparison use has been made of a grade 

 index* similar to that developed and used at the Connecticut Experiment Station. The grade index 

 is a single numerical expression for the grading quality of the tobacco grown under a given treatment, 

 and is comparable, but not equivalent to the average selling price per pound based on percentage of 

 grades. It is based on the percentage of the carefully assorted commercial grades and the relative 

 price value of the different grades. Although market prices vary from year to year, the ratio of 

 prices of the different grades remains fairly constant. The following table gives the commercial 

 grades and numbers denoting relative price values : 



Comparative 

 Commercial Grade Symbol value 



Light wrappers (19" up) L 1 .00 



Medium wrappers (21' up) M .60 



Long Seconds (19" up) LS .60 



Short Seconds (15" - 17") SS .30 



Long Darks (19" up) LD .30 



Short Darks (U" - 17") SD .20 



Fillers F .10 



Brokes _ B 10 



The grade index is obtained by multiplying the percentage of each grade by the corresponding com- 

 parative value and adding the products. 



