CROPPING PATTERN AND WATER RELATIONS 



361 



multiple correlation functions in fact does not make up the full 100% for 

 every value of the production parameters. The appHcation of these formulae 

 must be considered as an approximation. 



ThePjeliability of the Results 



If the reasoning about the fundamental properties of the cropping pattern 

 is not convincing, then still a fitting of curves through the scatter diagrams 

 remains, in v^hich Xq and A give some information with respect to the 

 position and slope of the curves, irrespective of any parabohc relation. 



If the result of different areas is compared, thi? proves something about 

 the reahty of these parameters of the curves. In Fig. 3 for two areas the 



CANISVUETPOLDEB. ZEEUWS- VLAANDEREN 

 Frequency '/• 

 30 1- 



26 

 22 



IB 

 14 



I III 



Frequency •/• 



I I I I I 



4 6 8 12 16 20 26 30 



LAND VAN HEUSDEN EN ALTENA 



Fig. 3. The vertical component of sliift log A, expressed in a frequency scale, plotted 

 against the same constant for another area, can apparently be determined rather 

 accurately. 



values for x^ and log A are plotted against each other. The graph shows 

 that within the margins of accuracy that may be expected in these investiga- 

 tions, the sliift parameters A and x^ show in each area the same mutual 

 relation. 



This result shows that the parabola, as a principle to bring order in the 

 independent choice of the crop to be grown by individual farmers, works 

 out well, even if there exists a distinct difference in the type of farming in 

 the two areas. 



