Abstract
The central place system within an urban area seems to be different from that of the ordinary regional level due to hierarchical and spatial structure and changing process, though we often have delt with both systems as the same one. Moreover, the intra-urban central place system itself is changing now obviously in correspondence with structural changes in retail trade brought about by the entrance of supermarkets. Accordingly the writer intends to determine the present-day structure of the intra-urban central place system in Hiroshima city based on the analysis of trade areas and then to clarify an important problem of central place studies.
The results analysed are summarized as follows:
1) Within the old municipal area of Hiroshima city there are more than 40 centres of convenience goods (vegetables and fruits) as shown in Fig. 2. In new residential areas of the urban periphery there are large trade areas with over 1km-radius, where consumers usually use both the nearest neighbour store and the supermarket in the centre. In trade areas of middle goods (underwears), however, the city centre expands its trade area remarkably at the cost of the most of the other centres so that only a few other centres have their small trade areas of middle goods (Fig. 3). Moreover, the trade area of the city centre includes not only almost all of the urban area but also the western half of Hiroshima prefecture in level of specialized goods (suits). Therefore, the typical hierarchical nesting structure of the central place system as seen in Christaller's theory cannot be seen within the urban area of Hiroshima (Fig. 8); there exist even centres with smaller areas of middle goods than of convenience goods. This fact means that the concept of threshold population, though it is basic for explaining the central place theory, does not actually play an important role in the location of central functions.
2) Although the writer did not analyse the central functions of each centre with the direct count method, the results of classification of central places obtained by this method seem to be obviously different from the results of this analysis; centres with central functions of higher order do not always have a large trade area of higher order goods. By this analysis a system of three levels in the order of central places is recognized.
3) The trade area of middle goods for the city centre is not delimited in the Densely Inhabited District of Hiroshima but expands mainly toward suburban area with poor central functions. The fact that the boundary of the urban area is ignored by the trade areas of intra-urban central places reveals clearly the unsuitableness of Beavon's model and his concept of subnormal profit.
4) The increasing ratio of retail sales from 1970 to 1976 is higher in the city centre and suburban area than in the outlying centres of higher order, as shown in Fig. 10 and Tab. 2. In addition, the decreasing density of population in the surrounding area of the city centre affects the growth of these centres negatively, which have acted for the central functions of the city centre destroyed perfectly by atomic bomb. Nowadays, suburban centres such as Gion, Kabe and Itsukaichi are growing so that it seems that the location of the outlying centres of higher order are moving from the old traffic nodes of the inner city to suburban centres rejuvenated by population increase of the surrounding areas. If so, the normal spatial pattern of Christaller's type should be recognized more actively for the intra-urban system of central places than the dendritic pattern which is expected to be developed in an area with inclining distribution of land value and regional demand.
