CLIMATE AND PLANT GROWTH. 



15 



vicinal specimens or in plants grown in a single battery. These vari- 

 ations may be due to such features as slight differences in the fertility 

 of the soil, but mainly they are accounted for by the natural tendency 

 of individual plants to vary. In order to eliminate such iildividual 

 variations, it is necessary to average the results secured from a large 

 number of plants. In the present experiment the number of plants 

 was sufficiently large to render the probable error of the average 

 measurements of any battery at a given station much less than the 

 difference in measurements between the plants of this battery and 

 those of the corresponding batteries of the other stations. This fact 

 is brought out in Table 4. 



Table 4. — Conrpurison ieticeen average error of the plant measurements at a 

 given type station and the difference in experimental results of the respective 

 type stations} • 



Plant. 



Type station. 



riant 



Error 



Per cent 



measure- 



in 



average 



ments. 



average. 



error. 



Mm.. 



Mm. 





12, 569 



1,358 



10.80 



19, 103 



2,708 



14.12 



5,064 



818 



16.13 



2,420 



141 



5.82 



4,296 



290 



6.75 



3,359 



136 



4.04 



155 



6 



3.85 



127 



4 



3.10 



57 



2 



3.50 



Differ- 

 ence 

 between 

 measure- 

 ments 

 at type 

 stations. 



Brome grass 2. 



Wheat 2. 



Peas 3. 



Oak-brush 

 Aspen-fir. . 

 Spruce-fir. 



Oak-brush 

 Aspen-fir. . 

 Spruce-fir . 



Oak-brush 

 Aspen-fir- . 

 Spruce-fir . 



Mm. 



6,534 

 14, 039 



1,876 

 937 



1 The formula used in Tables 4 and 5 in deriving the average error of the mean is: the summation of all the 

 variations from the mean, regardless of sign, divided by the number of cases times the square root of the 

 number of cases. 



2 Average leaf length per plant. 



3 Average stem length per plant. 



Table 4 shows the average leaf length (which is taken to repre- 

 sent leaf area) of typical specimens of mountain brome grass and 

 wheat, and the a^verage stem length of specimens of field peas grown 

 in the respective type stations. From these data are computed the 

 variations froni the mfian, the percentage of average error, and the 

 difference between the measurements obtained at the type stations. 



The greatest experimental error occurs in the case of mountain 

 brome grass. This is accounted for by the fact that only 5 specimens 

 of brome grass were grown in each type station, while in the case of 

 the peas and wheat 10 and 20 specimens, respectively, were grown at 

 each station. In each instance, however, the experimental error due 

 to individual variation within a type is much less than the difference 

 between two groups of different types. 



