As carrots prefer a cool season, they are usually grown during the autumn and the early spring on Japanese farms. Hokkaido farmers, however, cultivate carrots during the cool summer and harvest them in the autumn.
After being harvested, large quantities of carrots are sent from Hokkaido to markets all over the country, standing unrivaled among the carrot producing regions at the same season. The author has been studying the geography of truck farming, above all from a locational approach, so he investigated the carrot producing regions in Hokkaido, especially the Furano district. The carrot production in the Furano district increased rapidly in the 1970's, and this district has become the representative carrot-producing region in Hokkaido.
As the farms of Furano are comparatively larger than the farms of Honshu, they raise carrots with a higher productivity of labor, but with low productivity of the land. It was a stimulus for carrot growing that rice production has been restricted by the government since 1970. The Furano farmers chose to grow carrots and onions instead of rice.
Farmers who raise carrots and onions employ many women. Labor intensive farming in Hokkaido including carrot growing, usually is located near the cities, which can supply the many women employees. The main supply region of the agricultural employees in the Furano district is the Furano urban area itself and the coal mine cities which are situated to the northwest, 30∼40km away from Furano city.
Farmers who cultivate carrots in their fields plow the land with a powered cultivator, sow with a drill seedes, and weed with powered spray. The employees thin out the carrot plants and pull up the carrot roots, both with their hands. The machinery operation which needs skilled labor gets a higher reward, while the wages of the manual laborers stay in the lower levels. Thus, the farmer manages carrot production by employing relatively cheaper labor.
The size of the acreage under cultivation generally restricts the farming type. The average farm acreages in Hokkaido are arrranged in the order of: dairy (33.3ha), upland field cropping (16.5ha), paddy field cropping (5.9ha) and vegetable growing (4.2ha). As the farmers in the Furano district have field acreages in the 5 to 11ha, range, they choose paddy field cropping or vegetable production.

View full abstract

The findings of factorial ecology confirm, for western cities, the invariability of the socio-economic status and family status constructs. Also cities with a substantial ethnic minority can usually be differentiated by the ethnic status factor. In a spatial context, factorial ecologies have demonstrated how the spatial patterns of the three invariant constructs were characterized: socio-economic status is characterized by sectorial zonation, family status by concentric zonation, and ethnic status by cluster distribution.
Factorial ecology pays particular attention to the pattern of factor loadings and the distribution of factor scores through the application of factor analysis. However, spatial patterns of original variables contributing to explain those factors have been rarely examined. It is important to analyze the similarities between spatial patterns of input variables as well as the factor constructs derived from the correlations between them.
This paper attempts to investigate the residential structure for the Tokyo City Area by performing canonical trend surface analysis which is able to synthesize social structure (factor constructs) and spatial structure (spatial patterns). All the data collected from the 1980 Population Census of Japan were compiled into 562 cells of one square kilometer, covering the 23 wards of Tokyo (Fig. 1). According to previous studies (Takano, 1979; Ueno, 1982), it is assumed that the ecological structure of the Tokyo City Area could be well represented by the three underlying factors of socio-economic status, family status and residential zone. Then eight input variables related to these three factors were selected (Table 1). The paper is divided into three parts: 1) comparison of spatial patterns of input variables, 2) factorial ecology approach to risidential structure, and 3) investigation of residential structure by means of canonical trend surface analysis.
1) Comparison of spatial patterns of input variables
Ordinal trend surface analyses were applied in order to compare spatial patterns of eight input variables. Fig. 2 shows their spatial patterns, and their trend surfaces. Two types of classification of the spatial patterns were then performed by means of cluster analysis (the Ward method) of Euclidean distances measured on the two indices: i) nine standard coefficients of cubic trend surface, and ii) the increase in the cumulative percentage of variance explained by adding linear, quadratic and cubic trend surface terms.
Classification on the basis of the cubic thend surface coefficients yields two groups of spatial pattern, concave and convex patterns (Fig. 4). Also, classification on the basis of the percentage of explained variance produces three groups corresponding to the three hypothesized factors (Fig. 5). Taking into account variances explained by linear through cubic trend surfaces (Fig. 3), the following were observed: the spatial pattern of variables related to the socio-economic status is a sectorial one such that a simple linear trend of southwest to northeast is dominant; that related to the family status shows a zonal pattern characteristic of a quadratic; and that of the residential zone is a complicated pattern with two poles, but is less explained by a cubic trend surface.
2) Factorial ecology approach to residential structure
The data matrix pertaining to the eight variables was factor-analyzed, and the initial factor matrix indicated three orthogonal dimensions (components) with eigenvalues greater than 1.0, which accounted for 86.8% of the total variance (Table 2). Table 3 shows the normal varimax-rotated loadings together with the final communalities. They indicate that the ecological structure of the Tokyo City Area could be well represented by the three factors hypothesized previously. That is, “socio-economic status”, “family status”, and “residential zone” appear in the first

View full abstract