The purpose of the present paper is to come to a more general understanding of the urban structure of Hiroshima City by investigating urban factorial ecology from the viewpoints of both residential area and economic space. The urban area of Hiroshima City, with a population of 852, 607 in 1975, is divided into 39 areal units by using the aggregated census tracts and the former municipalities which were recently amalgamated. As shown in Tabs. 1 and 2 the writers carefully selected about 30 input variables including the changing ratio data in both analyses. The results of factor and cluster analyses using these input variables are summarized as follows: 1 ) Within the five factors which were extracted by factor anaysis (principal axis method and Varimax rotation) of residential area characteristics and explained 83.6 per cent of total variance, only three factors were clearly interpreted. These were urbanisation, urbanity and economic-occupational factors The third one, named economic-occupational factor, would correspond to the dimension of socio-economic status which are generally labelled by many studies. The factor structure analysed Is considerably different from the result of the previous investigation in which 55 smaller areal units (census tracts) within the former city boundary excluding suburban areas were used. It may be based on the difference of areal boundary and areal units, though the input variables also were not the same. 2 ) The standardized factor scores of each of these three factors are distributed as shown in Figs. 2, 3 and 4. The urbanisation factor has high values in the suburban area with strongly advancing urbanisation, whereas it has low values in the central area with obvious decrease of inhabitants and in the rural area where the waves of urbanisation seem not to have arrived. Both the distributions of urbanity and economic-occupational factors are close to a concentric pattern. But the latter factor has the two areas of highest score values in the built-up area. 3 ) Though seven factors extracted from the factor analysis of economic-space characteristics explain 82.3 per cent of total variance, only four main factors are clearly discerned. These are retail and eating-drinking, rural (rurality), industrial activity and manufacturing dominant factors. The score distribution of the retail and eating-drinking factor shows a concentric pattern with remarkably high values in the CBD and may be similar to the distribution of land values. It is noticeable, however, that commercial areas in the periphery of the central area which were formerly developed as sub-centres, tend to be stagnant or declining in the recent retail turnovers as shown in Fig. 10. Although the rural factor is also distributed in the form of a concentric pattern, its pattern is quite contrary to that of the first factor. On the other hand, industrial activity and manufacturing dominant factors show irregular patterns. While the values of the former are high in the areas with many establishments com-pared to inhabitants, those of the latter show the highest values in the coastal industrial zone and relatively high values in the areas along Routes 2 and 54. 4 ) As anticipated from the distribution pattern of factor scores for each factor mentioned above, the urbanisation factor in residential areal characteristics and the manufacturing dominant factor in economic space characteristics are scarcely correlated with other factors, so that each factor appears to make an independent dimension. Moreover, new groups can be classified from the results of Tab. 6 and Fig. 12, which were depicted by combining all five groups classified by cluster analysis of Ward's method; CBD (A), marginal area of CBD (B), the area specializing in administrative functions and with a complex of large apartment houses (B'), manufacturing area within the built-up area (C), residential area in the marginal

(View PDF for the rest of the abstract.)