We have applied the principal component analysis (P. C. A.) to the data of the 66 sampling points of Cryptomeria forest (C. japonica D. Don), for the purpose of investigating whether the forest site unit classified by means of the homogeneity in environmental factors such as climate, geological formation and topography well shows also the local characteristics of these soil properties or not.
P. C. A. summarizes a number of the variables into a fewer principal components by means of exchanging the axis of co-ordinates and further classifies an individual sample into several types according to the the value of the principal component which is calculated from the next equation:
Z_K=lk₁x₁+lk₂x₂+……+lk_nx_n+……+lk_px_p
where, Z_K is the principal component of No. K
lk_n is the Eigen vector of No. n
x_n is the variable of No. n (Normalized value)
The results of P. C. A. were as follows:
(1) On the physical properties, we could summarize the variables of 16 items into the three principal components which account for 77.9 percent of the total variance, and on the chemical properties, we could summarize the variables of 10 items into the three principal components which account for 77.5 percent of the total variance by P. C. A.
(2) The pattern of the chemical and physical properties of the soils investigated by these summarized principal components was found to be closely related to the forest site unit. Accordingly, it may be considered that the classification of the forest site unit we had done was also a well-grounded method from the viewpoint of soil properties.