8 MASS. KXPERIMENT STATION l5ni.LK IIN 2«;7 



Discussion of the Results of tlie Experiments 



The correct interpretation of the results of field experiments such as are reported 

 in this bulletin is extremely difificult. Unavoidable variations in soil and climate 

 are the most stubborn factors the agronomist must deal with in the interpretation 

 of his results. Every feasible device must be used to obtain a reasonable degree of 

 assurance that the differences in results are due to the treatments considered 

 instead of to chance variation in soil, climate, or some other variable. The effects 

 of soil variation can be reduced but not eliminated by the replication of plots 

 given a certain treatment; likewise, chance variation due to climate can be reduced 

 by repeating the experiment a sufficient number of years to give a fair sample of 

 the climate. This experiment is concerned with the eftect of different cropping 

 systems on yield and quality of tobacco. 



1 . Effect of Cropping Systems on Yield 



In Tables 8-18 are given the yields and grades of tobacco grown under the differ- 

 ent cropping systems. For convenient comparison the results are summarized 

 in Table 3 and averaged for the different periods. I^^air comparisons can be made 

 only among results of treatments which coveretl the same period. Thus, the 

 treatments "continuous tobacco: no cover (check)," "manure," and "timothy 

 cover" ran the entire course of the experiment. 



It will be seen from the table that for each of the pcrio Is the manure treatment 

 yielded more tobacco than either of the others. Further, timothy cover yielded 

 slightly less than no cover or check for the periods 1924-1927 and 1924-1931, but 

 slightly more for 1926-1931, and considerably more for 1928-1931. 'i'his indicates 

 a recovery in the latter part of the experimental period from the inhibiting effect 

 of timothy shown in the first part of the period and reported by Jones (13). The 

 figures for all treatments show that red top cover produced the most pounds of 

 tobacco for the period 1926-1931 and likewise for the period 1928-1931. Rye was 

 slightly superior to timothy as a cover for 1926-1931, but the relation was reversed 

 for 1928-1931. 



It will be observed that there is little ilift'erence between some of the average 

 yields cited. The question naturally arises as to what significance attaches to the 

 differences. How much higher nmst one average yield be than another to be 

 significant? A statistical study4 has been made of the results, which throws some 

 light on the question. By replicating plots of the same treatment several times, 

 and by conducting the experiment for several years, the probable error of a single 

 plot has been reduced to less than 5 per cent, and the probable error of the mean 

 (average) to less than 1.5 per cent, both expressed as percentages of the mean. 

 This means that there must be a difference of about 6 per cent or about 95 pounds 

 before one average can be considered as significantly greater or less than the other. 

 In other words, differences in averages much less than 95 pounds, for the period 

 1926-1931, may be due to chance variation instead of to the treatment in question. 

 By this criterion the average yields from red top cover and manure are the only 

 ones sufficiently near the requirement to make it reasonably certain that the 

 difference is due to the treatment instead of to chance. For the same period, 

 yields from no cover, timothy cover, and rye cover are so close that no particular 

 significance can be attached to the differences. 



The yields of tobacco from the two rotations are so much below those of the 

 (heck that there is lit lie question about the significance of the difTcrences, although 



*See appendix for note on method of statistical interpretation. 



