Introduction - Methods for Multilevel Analysis and Visualisation of Geographical Networks - page 2

Fig. 1.1 The city as the connector of multiple networks

Thus, cities can be considered to be “connectors” that favour the emergence
of new links and maximise interaction, and a city's wealth depends on the
heterogeneity that this valuable combination creates ( Allen , 1999 ; Castells , 1996 ;
Jacobs , 1969 ). Each type of network develops its own particular rules, regulations,
conventions and values ( Castells , 2009 ; Storper & Salais , 1997 ; Uzzi , 1997 ).
From a geographical perspective, the assumption that transformations of places
strongly depend on their components' internal and external relations with other
places makes the positions of places in networks at different geographical scales
crucial. Geographical interactions can be studied through networks that are repre-
sented as graphs. To date, most geographical network studies have been based either
on “gravitation models” ( Wilson , 1967 ) or general measures of graph connectivity
or accessibility ( Kansky , 1963 ). Graph theory - in particular, Watts' small-world
networks model ( Watts , 1999 ) - can contribute to this research area and enrich
spatial analysis by complementing systems theory.
Through graph theory, network analysis has also been associated for a long time
with sociology ( Freeman , 2004 ). Graph visualization was employed much earlier
than is typically reported. Freeman identified kinship trees ( Morgan , 1851 )and
simple graphs of two-step marriage prohibitions ( Macfarlane , 1883 ) as seminal
graphical representations. Hobson's ( 1909 ) study of the interlocking directorates
of the De Beers and Rand Mines industry in South Africa was the first example of
two-mode data collection ( Hobson , 1909 ).
Freeman ( 2004 ) noted that these pioneers were followed by a succession
of twentieth-century schools that incorporated new issues and methods into the
study of social networks - among them, Lund University's leading geographer,
Hägerstrand.His approach to spatial diffusion ( Hägerstrand , 1952 ), which assumed
that the social network acted through space and time, highly influenced structural
approaches to geography in the 1960s( Berry , 1964 ; Dacey , 1964 ; Garrison , 1960 ;
Garrison, Berry, Marble, Nystuen, & Morrill , 1959 ; P. Gould & White , 1974 ;
Tobler , 1965 ). However, Freeman criticised the lack of communication between
social geographers and other social scientists that prevented the exchange and