Click to increase image sizeClick to decrease image size Acknowledgements Acknowledgement is due to Peter Wilkinson and anonymous referees for providing comment on earlier versions of this article. Responsibility for any errors or omissions remains that of the author. Notes 1. The population categories referred to in this paper were established under the apartheid regime but have been maintained by the present democratic government in official documentation, including the national census, to provide a means of monitoring social change, particularly efforts to transform the legacy of the apartheid era. The term ‘Coloured’ generally refers to persons of mixed ethnic origins drawn mainly from first-nation Khoi and San peoples, slaves from present-day Indonesia and European colonialists; ‘Black African’ to persons descended from one or more of the Bantu-speaking peoples; and ‘White’ to persons descended from European settlers. In the interests of consistency these terms are used throughout the paper, even in instances where the nomenclature used in source material is different. 2. Due to the author's involvement in the drafting the chapter dealing with the design of movement networks a different approach, similar to that discussed later in this paper, is adopted in this version. It is argued here, however, that this approach has yet to become widely implemented or at least recognised by practitioners as best practice. 3. An ‘open’ road network is defined here as a network of roads that intersect freely with one another, and thus offers a choice of numerous alternative equidistant routes between any two points. ‘Closed’ road networks, conversely, adhere to principles of hierarchical access management, and establish clearly defined routes between any two points within the network offering few or no equidistant alternatives. 4. Indeed an innovative form of neighbourhood unit, the Radburn ‘superblock’, went one step further and assumed that pedestrian and vehicle conflict could be removed altogether by the design of grade-separated roadway and footway networks within local neighbourhoods. 5. In the case of multi-mode trips, the ‘main’ mode is typically defined as that which covers the greatest travelled distance. The relative importance of walking as a travel mode is underestimated in main mode use analysis as walking trip segments are invariably attached to both ends of public transport trips and sometimes to an end of motor-car trips. 6. ‘Inner’ locations refer to residential neighbourhoods with a mixed pattern of land use, and an open, pedestrian-oriented street pattern providing local access (meaning within a 2.5 – 3-km walking distance) to commerce and employment opportunities. ‘Outer’ locations refer to residential neighbourhoods with little or no local access to (formal) commerce and employment opportunities, and a closed, car-oriented street pattern. 7. This is not surprising given the findings of research on the relationship between vehicle speed and pedestrian fatality. Studies have found that in collisions between vehicles and pedestrians, the pedestrian's chances of being killed rise dramatically with an increase of vehicle speed. The probability of a pedestrian fatality is 5% at 32 km/h, rising to 37% at 48 km/h and to 83% at 70 km/h (O'Flaherty, 1997 O'Flaherty, C. 1997. “Physical methods of traffic control”. In Transport Planning and Traffic Engineering, Edited by: O'Flaherty, C. London: Arnold. [Crossref] , [Google Scholar]). 8. It is interesting to note that in Tanzanian cities, where walking shares of modal split are very similar to those of low-income households in South African cities (46% in Dar es Salaam and 67% in Morogoro versus 61% in Cape Town), it is recommended that the parallel frequency of ‘collector/distributor’ routes is up to 5000 m (Rwebangira & de Langen, 1999 Rwebangira, T. and de Langen, M. 1999. “Urban road design standards: recommended non-motorized transport elements in Sub Saharan Africa”. Georgetown East Paper presented at the Transportation Research Board 78th Annual Meeting, Session: Global Transport Forum 1: Urban Mobility in Sub Saharan Africa [Google Scholar], p. 5). This reflects much greater sympathy to the severance impact of arterials on pedestrian movement in lower-income environments than the South African convention of 1500 – 2000 m.