yo,

I suppose you are all familiar with the **“image of the city”** by Kevin the Lynch (+). If not better check it out before you get to 2nd year of uni.

Last night I came across a paper by

**Ruth Conroy Dalton and Sonit Bafna**

from:

Georgia Institute of Technology, USA

which is based on the book & some new injections and potential ideas

**Abstract**

*This paper presents a study of the relationship between city elements, as defined by Lynch, and the spatial descriptors commonly used in space syntax research, leading to a proposed relationship between the hitherto unrelated concepts of intelligibility and imageability. The paper starts by demonstrating how each of Lynch’s five city elements (the node, path, district, edge and landmark) may be redefined using a selection of spatial notations, primarily the axial line and the isovist. Furthermore, by precisely defining the relationship between the axial line and the isovist, it can be shown that all of Lynch’s elements may be redefined using a single, coherent family of tightly-related spatial entities. A case study of Boston, circa 1950, is used to test an application of these redefinitions and the relationships between the various spatial descriptors and Lynch’s elements. In turn, this leads to a hypothesis concerning the relationship between the concepts of intelligibility and imageabilty, concepts that were previously considered to be independent. Finally, the paper concludes by building upon the relationship between intelligibility and imageabilty to conclude that this relationship provides strong evidence for an underlying cognitive basis to space syntax.*

**1. Introduction**

Kevin Lynch, in The Image of the City (Lynch, 1960) argued for *legibility *being a significant quality of the city. According to him, the legibility of the city, or “the ease with which [a city’s] parts can be recognised and can be organised into a coherent pattern”, is significant not only for aiding practical tasks such as way-finding, but also that it is central to the emotional and physical well-being of the inhabitant population, personally as well as socially. He continues by equating the legible environment with an “imageable” one. Imageability, according to him, is “that quality in a physical object which gives it a high probability of evoking a strong image in any given observer. … It might also be called *legibility.*”

**2. First and second order elements**

An obvious starting point is to attempt to re-interpret Lynch’s five environmental features in terms of the basic spatial descriptors commonly used in space syntax research. The motivation behind reinterpreting Lynchian elements in space syntax terms is not reductive—we are not claiming that all of the Lynchian elements can be completely characterised in terms of syntactical variables—but rather the effort is to find out how much of Lynch’s theoretical approach and his specific findings can be accounted for by a syntactic approach. Our study, therefore, has a two-pronged approach. On the one hand, we will try to determine whether there is a syntactic logic underlying the basic descriptive elements through which Lynch describes the inhabitants’ cognitive maps of cities, and on the other, we will try to offer a syntactical argument for why particular environment-features of the selected study area are picked up by Lynch’s subjects and not others.

Our argument is that Lynch treats his five elements in a resolutely empirical sense; he seems to have identified them as being the best elements available for characterising how inhabitants map their cities, without making a case for how they come together to give a systematic mapping of cities. Despite his reticence on the subject, however, it is easy to see that the elements are anything but random. We can sort these elements into two groups. At the first level are the *spatial *descriptors— nodes, paths, and districts. They are elements of zero, one, and two dimensions that observers acquire and utilise as anchors for location, and the relationship of these elements to the observer is topological. Not only can the observer position himself in space in terms of basic topological relationships (“to the front of”, or “to the right of”) but also ‘at’, ‘on’, or ‘inside’ them. At the next level are primarily *visual *descriptors—edges and landmarks. The observer’s relationship to them is of a higher geometrical order (at least projective), in that he/she locates her position vis-à-vis these using a rough polar or vector orientation (a sense of both distance and direction), 59.3 but does not actually occupy them.

**3. Axial lines and isovists**

The question that naturally arises, at this point, is *which *syntactical descriptions will best help our case. The axial line representation seems the ideal descriptor, partly because it is the commonest representation in the space syntax repertoire and empirically the most successful for urban contexts, and partly because it is the closest to computational studies of wayfinding (Kuipers, 1996; Kuipers, Tecuci and Stankiewicz, 2003; Penn, 2003). However, Lynch’s stress on the visual characteristics of elements does raise the question of whether we should be including other spatial representations in this paper, and the isovist appears a potentially useful candidate.

What we would like to clarify, in this context, is recent work, which can be held to bridge the conceptual gap between the concept of the axial line and the isovist**2**. The traditional definition of the axial line is that it represents the fewest and longest lines of sight that pass through every space comprising any system. The definition of an isovist is that it is the field of view, available from a specific vantage point; a horizontal slice through this field of view is then calculated, usually taken at eye height and parallel to the ground plane. It is the resulting polygonal representation of this two-dimensional, visual ‘slice’ that is referred to as an isovist. It is worth noting that axial lines and isovists are one and two-dimensional representations respectively. (Viewsheds, used predominantly by geographers, are on the whole, three-dimensional and represent the entire field of view from a single location).

* * *

If we consider *briefly *isovist integration or VGA analysis, there is a relationship between isovists and convex shapes. If an isovist is generated from an array of vantage points forming a regular grid filling all possible navigable space, then the relationship of isovist connectivity (or mutual visibility between points) can be established. In work by Turner et al they established that each clique in the graph is analogous to a convex space. That is to say, a space inside of which any pair of points are mutually visible.

A formalisation that used the single concept of the line of sight to create both a consistent convex partitioning and a definite axial line structure for any spatial setting has been suggested by Peponis and his colleagues (Peponis, Wineman et al., 1997; Peponis, Wineman et al., 1998). Using only the *extension *part of the lines of the all-line map as boundaries, Peponis and all show how a given space can be partitioned into informationally stable shapes called E-spaces (or using a more limited set of all line extensions, into S-spaces), and these in turn used to automatically

generate axial lines.

The main body of the work is very long for a “blog post”

I will try to split it up in 3 or 4 parts. Maybe we will get you a link to download it from here soon

ciao minions