SASI Home| Publications | Data | Maps | SASI Staff
Dorling, D. (1991) The Visualization of Spatial Structure, PhD Thesis, Department of Geography, University of Newcastle upon Tyne

Chapter 2: People, Spaces and Places

The practices through which social structure is both expressed and reproduced cannot be divorced from the structuring of space and the use of spatial structures. Previously structured space both constrains and enables the reproduction of social practices and social structure.
The social becomes the spatial.
The spatial becomes the social.
[Pred A. 1986 p.198]

2.1 Which People

I am interested in the lives of the people of Britain over the last two decades. This is because I am one of those people. I have been counted as a birth in 1968, a child in two censuses, as a migrant by the Health Service, as a claimant of unemployment benefit, as a voter in a general election, and so on. As I have been counted, so have been millions of others. Surely all these numbers can be turned into a picture of the people in the country in which I grew up?

What is it about these people that we wish to understand? Trivially, it is who these people were, how old they were , what they did. Fundamentally we ask what was the relationship between these people, what was the structure of the society in which they lived? You cannot have grown up in Britain in these decades without having felt the weight of this structure, how it affected your life, defined your opportunities, and altered your destiny. Whether you were male or female, where you went to school... you knew it made a difference, but you did not know quite what difference it made.

More importantly you could not have known the effect of the social structure on everybody else. It is the relationship between the gains of some people, and the losses of others, which would escape any single individual's perception. They may be able to imagine some of it, but not to picture all of it. Here I attempt to show the social structure; not to explain or to understand it to any great extent, but merely to see what there is to be understood, and how its most obvious workings are organised. To illuminate the situation.

It is only by first seeing what you wish to comprehend, that you can begin to understand why and how it exists. Just to create an image of the most simple manifestations of the structure of the society that these millions of people make up, is a difficult undertaking (Figure 4). The story that is told should treat everybody's part in it as equally important, as all their lives should, at least in the way that their history is told, have equal value. This is a picture book which begins that story.

Why was British society chosen as the subject of this work? That we should study all the people in this place, and only the people of this place, was a choice made from both the practicalities of the exercise and the experience of the writer. That the line chosen divides land from water is convenient, but not the reason for its imposition (Print XXX). The line around this country divides the experiences of most of its people from those outside. The sea may present no great economic or political barrier any longer but it is still a very strong social one.24

It is argued the the boundary of Britain is a much stronger social divide than those lines drawn within the state, whether between Scotland and England, or inner and outer London. The internal differences are more often seen and measured as just that — differences. The divisions over the water separate political systems which are hard to compare, standards and styles of living which are outside the experience of most of Britain's inhabitants. Some of the sharpest divisions are the closest; separating Northern Ireland from Britain, the United Kingdom from Europe, although these are much less in magnitude than inequalities further afield. Eventually it may be possible to undertake this kind of study across those lines. But it must be done at the local level, not used as another excuse for amalgamation — creating the average English, Scottish or Welsh man and woman (Print XXXI). Before we can begin to understand world society, it is prudent to delve into our own, to see just what we are comparing others with.

2.2 Why Study Places?

If we are interested in the relationships between people, why should we wish to consider places? It is not due to the ease with which we can collect information about people in a myriad of places that we should do this (Print XXXII). It is because, just as your place in time so obviously constrains and determines your life, so your place in space limits and creates the possibilities in your world. It is not the actual position in space, as it is not the actual position in time which does this, but who else shares that place, who else shares your time (Howe G.M. 1986d, Johnston R.J. 1986a, Agnew J.A. 1987, 1989, Agnew J.A. & Duncan J.S. 1989).

However, times and places are fundamentally different things. As we live we must all share the same moment in time, but, in existing, must be spread over space into different places. This diffusion of settlement in space, juxtaposed with the concentration in time, defines the dimensions over which experiences can differ. The organisation of people over space, and through time, is what makes place important.

Constrained by the limits of time, people are forced to live close to where they work. In cities they work together, but live apart. We used to live together in villages and work apart in fields; now we live apart in suburbs and work together in offices. This spatial organisation reflects the need for people to live together and the wishes of some to be apart. As they are rewarded unequally at work (if they work), this inequality is reflected in where, and how they live.

Thus neighbourhoods are created, areas where most of the people have a comparable income, live in houses which are alike, have similar backgrounds and, to some extent, a common future. When a firm closes down only those neighbourhoods from which its employees came are directly affected. If the supply of labour is spatially compact, so too will be the impact. Only pupils in the neighbourhood of a certain state school will go to that school. People downwind of a particular source of pollution will be most affected by it.

Not everything is so spatially determined. If people in one place suffer, so eventually will all others in some way. It is the spatial reinforcement of these trends which makes them so obvious. People in poor neighbourhoods are less likely to have their houses improved or their schools maintained. Then, more subtly, as the relative power of people in different places affects those places, so does the movement between them.

Just as the pattern of commuting allows the neighbourhoods to exist, the pattern of migration serves to exacerbate their differences (Print XXXIII). As a few people in poor neighbourhoods do well, they move to richer ones. More importantly, the vast majority of migration is between similar places in the spatial social hierarchy, reinforcing and perpetuating the existing differences.

The terms neighbourhood, community and locality define, here, a group of people who live in close proximity. They do not necessarily have to know or even recognize one another. What the study of human geography has shown is that they will tend to have more in common with each other than with outsiders25. This is because of what put them there, keeps them there, and moves them away — the forces that sort people in space, the institutions of social structure.

Place is important to the understanding of the social structure of society because it is through places that that structure is most directly visible. Not only is it visible in our everyday lives, but some of its many facets can be made visible on paper (Figure 5). So also must the complex picture of flows be portrayed; the flows of people which together allow, maintain and reinforce the spatial picture.

2.3 What Are Spaces?

Spaces are constructed from the relationships between places. Just as the individual attributes of people are not the main interest, so the collective attributes of single places do not hold the key to our understanding of society. Places are not things which can be rigidly defined and have a meaning of their own. They are abstract collections of people whose depiction can shed some light on the spatial social structure of our lives. Thus we cannot talk about the situation of a place in isolation, but we can talk about the context created by the collection of all such places, making up a space. It is the divergence and convergence, clustering and scattering of people between places in spaces on which we should concentrate (Muehrcke P.C. 1972, Warntz W.W. 1976, Petchenik B.B. 1979, Borchert J.R. 1987, Beruchashvili N.L. 1987).

Aspatial views of society can capture many things. Gender relationships within households may not express a national geography. Those whose schooling was in the private institutions will never have experienced the spatial inequality of the state education sector, in which where you live determines how well you learn. Aspatial views which, say, equate how people vote with their class or job may find many single strong relationships, but will fail to put them all together as a single image. Those whose schooling was privileged will have grown up in privileged places and so on (Print XXXIV). The near certainty with which we can so easily make these statements illustrates, in itself, how strong the spatial structure is.

In two dimensions we can see the entire environment in which people live as a whole, not only how they vote and work, but how everybody around them votes and works, and how other things about them are distributed — where they and their neighbours came from, how they live, who they are. Perhaps it is because this spatial structure is so strong, so well known, that we so often seek to find more ethereal aspatial relationships. We should first take a look at the wood before trying to classify the trees as if they were not part of it.

There are, however, some fundamental problems to be overcome in trying to see the structure of society through its spatial apparitions26. To begin with, there is the problem of drawing a line around a group of people to be called a society, for which Britain was chosen for this dissertation. Then, there is the question of how to cut up that space, and what effects such divisions can have. Numerous lines have been drawn across Britain defining communities and cities, regions and villages (Prints XXXV, XXXVI, XXXVII). How the spaces of interest can be rebuilt from the dissection of the nation is the central problem in relating places to people.

2.4 Drawing Lines

Just as the line which surrounds Britain is dictated to us, so too are most of those which divide its people27. Most seriously, for many purposes these lines divide the country into areas which contain far too many people to come anywhere near to the idea of the neighbourhood, locality or community outlined above. At a ridiculously aggregated scale Britain is often divided into ten or eleven regions and numerous statistics periodically published for average people in these areas — for example, the average Yorkshire and Humberside person.

Other large areas often used include the seventeen planning regions, sixty four counties, four hundred and fifty nine districts (Print XXXVIII), six hundred and thirty three mainland parliamentary constituencies (Print XXXIX), and so on (Figure 6). These areas define regions the size of small towns, often of a hundred thousand people or more. The spatial social structure hardly manifests itself here; for in these areas, to sustain such areas, are people from all parts of the social structure. Towns and cities are made up of the rich and poor. There are not rich and poor towns, not in the sense that there are rich and poor people.

What is more, the boundaries of these many areas rarely coincide, neither with each other in space nor themselves in time. So we are left with most of our geographical knowledge being mixed into a plethora of ephemeral places. The shapes and sizes of which do as much to alter the appearances of what is happening, as do the numbers which are gathered about the people within them; even though they may be very accurate measures. The average for a large area destroys knowledge of the variations and defines people who do not exist; a bit of everyone and all of no one (Nystuen J. 1965, Clarke J.I. 1975, Rhind D. 1975a, Coombes M.G., Dixon J.S., Goddard J.B., Openshaw S. & Taylor P.J. 1978, Fotheringham A.S. 1989, Tobler W.R. 1989a).

One way to try to overcome this confusion is to draw another set of lines, but this time defined not for the convenience of collection, to administer government or distribute services, but on some functional grounds28. Zones have been created which try to encompass the areas within which people live and work, or between which they tend to migrate. Although such exercises might usefully tell us something about the patterns of commuting and migration, they do not serve well for seeing social structure any better than the existing divisions — for several reasons.

Primarily, they have tended to create even fewer places than have commonly been used and so the gross aggregation of communities is sustained, again creating the illusion that there is little spatial division, but worse when it is claimed that there is some rationality behind these lines. Whereas administratively defined areas tend at least to collect the same numbers of people within their boundaries, these functionally defined entities exhibit some of the greatest variations in population of any set of boundaries in use. Finally, the very nature of these areas, created from flows of people, dispenses with the spatial aspect of society by creating a set of (semi-autonomous) places which can be studied on their own, ranked and (supposedly) profitably listed in tables. Functional regions are an aspatial concept, an attempt to take out the effect of movement and the relationship between places, where the real differences are to be found29.

There has been too much drawing of lines. Now is the time to begin rubbing them out, to see the real divisions, not the ones imposed by the cartographer's pen, administrator's necessity or geographer's algorithm.

2.5 Picturing Points

To show the real spatial distribution of people we need information about places which are at least as small as the communities we envisage these spaces to contain. If you think of the number of neighbourhoods in a town of two hundred thousand people, you should be able to count at least ten, if not twenty, distinct estates. Places ranging in size from as few as one hundred to as many as twenty thousand people are what we know as neighbourhoods. It is with places such as these that we can build comprehensive pictures of the national spatial structure. Even the smallest, conventionally used places are usually too large for our purposes (Print XL).

The boundaries of such small places may well be arbitrarily defined. That is not a great problem. It is the relationship between the places in which we are interested. We should look at what they show collectively, not their individual characteristics — there are far too many of these to examine each one, anyway. With enough small places we can create methods which are robust to the effects of the arbitrary lines drawn on the map, devising techniques by which we can paint realistic pictures of social spaces (Petermann A. 1852, Wright J.K. (ed.) 1938, Clarke J.I. & Rhind D.W. 1975, Rhind D. 1975b, 1983b, Census Research Unit 1980).

The most important factor about the localities that we use is that they vary little in the number of people which they contain, for again we require equal representation. Here we find that the second traditional response of geography in its rejection of arbitrary administrative divisions fails. At the same time as "functional" regions were being created, other researchers were turning their hand to placing lattices over the land area of the British Isles and counting attributes of the number of people in each small square. This method of division was called grid mapping, and one kilometre squares were most often used.

Grid mapping is no less arbitrary a practice than any other technique, although the practicalities of its execution are the simplest. That it creates a stable set of units over time is a trivial defence of the method; any set of lines you draw and leave on a map is "stable". What makes this type of spatial division stand out as more flawed than any other is the huge inequalities in populations it creates between areas, at the added expense of creating so many areas30. Unlike any other geographical division, the more places that are added in this one, the greater the inequalities become, with most spatial units containing no people at all.

What we require to cover Britain then, are roughly ten thousand to one hundred thousand localities. Fortunately, the lowest tier of administrative geography gives us the first set — wards31 (see Appendix F), and that of census geography the second — enumeration districts32. There are other practical alternatives, for example postcode sectors, on which much market research is based, number over eight thousand (Prints XLI, XLII & XLIII). But in general the absolute wealth of information that is available at the time of the census for enumeration districts (and for wards during the decade), makes these places the natural choices from which to paint spaces33.

2.6 Population Space

How should these pictures of spaces of people in places show us social structure? What should they look like, how should they reflect society? To answer these questions we must have some idea of what it is about the space we want to see — the nature of society.

People in space create a near continuum, especially if we view the distances between them in a relative sense. The spatial nature of our society is such that nearby places usually exhibit very similar characteristics in their populations, but occasionally they diverge widely. To see these patterns we have to stretch and squeeze the space of physical geography into becoming the landscape of human experience, opening up the cities, exorcising the empty space from the image (Prints XLIV to XLVII, Bachi R. 1975, Tukey J.W. 1976, Tikunov V.S. 1987).

The spatial nature of the society we live in holds more than divisions and continuity, trends and correlations. It is patterned. Social patterns of power, control, deprivation and monotony are all reflected in spatial mosaics. Rings of the wealthy, holes of the poor, lines of accessibility, enclaves of distinction. However, none of this would be seen if we did not seek to see it. We must know something of what we are looking for before we can know how to look.

Pictures of spatial social structure should have the power to reflect the complex tapestry and delicate lace-work of the relations between people through the places in which they live and the spaces they create. All social organization must take place somewhere, and aspects of that somewhere strongly shape what structures are formed.34

2.7 Adding Time

Finally, space and place is not enough. As this chapter was introduced by arguing that the constraints of space are as important as those of time, so it should end, re-introducing time into the picture. This detailed spatial structure is something which takes time to form, and is deformed in time (Pred A. 1984, Taylor P.J. 1991).

Neighbourhoods are temporal phenomena, the cohort in time is like the community in space. People who live at the same time will live comparable lives of comparable quality. Far more so for people who live at the same time and in the same places. These three dimensional spacetime pockets of existence are the ultimate level of containment in this work. They can be delimited by neighbourhoods and generations, measured in a few hundred people and a few dozen years.

The history and geography of modern Britain, which tells the story of all its people, can only do so successfully, through all the sources we have, based on the concept of ever changing communities35. Definition in time is, however, much less clear cut than definition in space. Communities merge into one another, evolve and disappear. The idea of rigid boundaries is even more ridiculous when applied to time than when applied to space. Here an infinity of units must be used to create a spacetime continuum, to reflect the temporal spatial structure of society.

A fixed three-dimensional image of an evolving, finely organised society is the final suggestion of this work, relating the times through which people live their lives to the places in which they live them. For now, creating the two-dimensional spaces to reflect simple two-dimensional instances of a complex structure is enough of a challenge, and the focus of the next chapter.



XXX Land use in Britain by 1km square grid (Colour).
XXXI Level II European Regions — annotated base map shaded by unemployment rate.
XXXII Counties and Scottish Regions — annotated base map shaded by unemployment rate.
XXXIII Family Practitioner Committee Areas — annotated base map shaded by unemployment rate.
XXXIV Local Education Authorities — annotated base map shaded by unemployment rate.
XXXV Functional Cities — annotated base map shaded by unemployment rate.
XXXVI Local Labour Market Areas — annotated base map shaded by unemployment rate.
XXXVII Travel-to-work Areas — annotated base map shaded by unemployment rate.
XXXVIII Local government districts — annotated base map shaded by unemployment rate.
XXXIX Parliamentary Constituencies — annotated base map shaded by unemployment rate.
XL Amalgamated Office Areas — annotated base map shaded by unemployment rate.
XLI Postcode Areas — coloured at random (Colour).
XLII Postcode Districts — coloured at random (Colour).
XLIII Postcode Sectors — coloured at random (Colour).
XLIV The British mainland rail network on an equal land area projection.
XLV The British mainland rail network on an equal population projection.
XLVI The British primary road network on an equal land area projection.
XLVII The British primary road network on an equal population projection.


A program was written to perform conventional automated cartography to modern computer standards. The program combined information from geometry and lookup data files with CSV text-files describing the topology and other attributes of places, to produce the desired map as a drawfile, which could then be manipulated further. The shading and names of places was given in the textfile and a set of simple rules applied to annotate the areas. Names were split at spaces, hyphens, commas, before the syllable "shire", and wherever an underscore had been inserted. The text was then scaled to fit within a rectangle enclosing the place, and centred.

Places could be represented by complex "paths" rather than simple polygons, so all the islands and lochs of Western Scotland were easily combined as a single list of the boundaries of Strathclyde. Paths could also overlap, and the space covered did not have to be completely allocated to areal units, as it often does in traditional systems. The places did not have to be given in any particular order as each was tagged in the drawfile with a unique identification code, permitting many other graphical possibilities.

Creating the maps as drawfiles also meant that they could be produced on almost any printer available, or "dragged into" documents such as this.

Figure 4: Drawing the Maps

The entire ward geometry of Britain was generalized using an algorithm which recorded a pair of vertices every five kilometers along the length of each (original) boundary. The list of boundaries was then stored with the following coded information:
left ward, right ward, original length, number of
vertices,and (X,Y) vertices pairs...
In spite of programming at the binary level the file was 1,099,648 bytes in length.

A lookup table was also constructed as a binary file giving ten different area codes for each ward. These were its district, county, amalgamated office area, parliamentary constituency, travel to work area, local education authority, European level II region and Family Practitioner Committee area.

A drawfile for a particular geography was produced using the draw library and a linked-list of relevant boundaries constructed using the lookup table. The list for each area type was transformed into a series of polygons and the resulting path tagged by its area identification.

The software can handle over 10,000 units, and complex structures such as lakes within islands within lakes. The drawfile can be shaded or edited on the screen later, or further manipulated by other applications.

Figure 5: Storing the Geometry

The hierarchy of areal units in Britain used in this thesis is shown here. Arrows indicate which units contain which. Dashed connections imply the use of a very similar geography. All of these divisions cover the whole of Britain:

Figure 6: The Areal Hierarchy


24 [a] Great Britain or the United Kingdom?:

It should be noted that Great Britain (England, Wales, and Scotland) is not the United Kingdom (which includes Northern Ireland as well). But, given the fact that Ireland as a whole constitutes a separate land mass, that it was historically governed as a colony of Great Britain, that the division of Northern and Southern Ireland occurred only in 1922, and that Northern Ireland itself contains but six counties — for these reasons, we restricted the study to the single land mass of Great Britain for which the requisite data were readily available. [Massey D.S. & Stephan G.E. 1977 p.352]

[b] Divisions, boundaries and borders are a help and a hindrance:
Few demographers define areal populations by statistical analysis before proceeding to analyse them demographically, but many of the more local boundaries for enumeration purposes (e.g. enumeration districts, census tracts) are delineated with demographic considerations in mind, and areal populations are aggregates of these small units. Moreover, among larger populations the use of administrative or geographic boundaries for population analysis is not always demographically undesirable, as many such boundaries have real demographic significance, and some, as for example between U.S.S.R. and China, between East and West Germany and between Bangladesh and Burma, are real demographic divides. [Clarke J.I. 1975 pp.2-3]

25 [a] Place is not everything, but:
To insist on the continuing importance of place, therefore, is not to deny that processes beyond the locality have become important determinants of what happens in places. But it is still in places that lives are lived, economic and symbolic interests are defined, information from local and extra-local sources is interpreted and takes on meaning, and political discussions are carried on. [Agnew J.A. 1987 p.2]

[b] Places can be as much social fabrication, as physical reality:
Paradoxically, the early British attitudes that viewed all West Indians as Jamaicans may be becoming true. The Jamaicanisation stems more at this moment from an identity forged by white racism than from a Marxist class consciousness. [Peach C. 1984 p.228]

[c] When, for instance, looking at an area the size of a London borough:
The point should be clear enough. People live in different worlds even though they share the same locality: there is no single community or quarter. What is pleasantly 'old' for one person is decayed and broken for another. [Wright P. 1989 p.290]

26 [a] A well known source of error is the ecological fallacy:
Many of these inferences occur in descriptive studies, in which it is very easy to confuse the characteristics of areas with the characteristics of people who live there. [Openshaw S. 1984 p.18]

[b] We cannot study individuals when dealing with places:
At least as long ago as 1950, it was demonstrated that not only the strengths of relationships between different variables change at different levels of aggregation but that, in some cases, the sign of the relationship also changes! Even worse, the effects of country-wide variations in size of areal units within the same level of the administrative hierarchy certainly provides variable levels of resolution and may also induce severe biases. [Rhind D. 1975 p.9]

[c] Analysis of individuals requires unaggregated information:
The only safeguard against false interpretation of statistical maps is to be able to get back to individual unit data, located at point accuracy in order to check on one's opinion. It is especially unsound to base policy decisions on statistical maps unless these maps work to the lowest practicable level of spatial, numerical and temporal generalization.
It must be constantly remembered that the spatial, numerical and temporal filters are varying together in virtually all statistical maps of change, and within the spatial filter itself there is a further source of variation within each map especially if one is using irregularly shaped and sized areal units.
Statistical maps at high level of aggregation are at best distorting mirrors held up to reality. The "simple" portrayal of a distribution is a difficult enough problem, as has been shown. A second layer of problems rests upon the first. ... [Forbes J. 1984 p.100]

[d] We have to learn to work round these problems:
The dependence of the correlation coefficient upon the territorial base is well known to statisticians and it is just one example of the general yoking of statistical quantities to the size of the collecting areas. Many attempts have been made, without success, to solve the problem. But it is inherently unsolvable because collecting areas or territories are man made and therefore arbitrary rather than 'natural' units. The same comment applies to time-series. Days, weeks, months and quarters are equally arbitrary subdivisions of a year and the smaller units go to make up the larger. [Cliff A.D. & Haggett P. 1988 p.163]

[e] An alternative view is that:
It is not the areal units which are to blame. The difficulty is that the method of analysis used was inappropriate. This tautology is immediate. If the procedure used gives results which depend on the areal units used then, ipso facto, the procedure must be incorrect, and should be rejected a priori. [Tobler W.R. 1989 p.115]

[f] The use of visualization may well help overcome the modifiable areal unit problem:
A fourth solution to the MAUP is much more speculative in that it involves the abandonment of traditional statistics in favour of other ways of representing the inherent information in the data. What exactly these other ways are is still uncertain but increasingly emphasis is being placed on the visualization of data rather than on statistical analysis (Tukey, 1977; Sibley, 1988) and it may be that these techniques are less susceptible to the MAUP. [Fotheringham A.S. 1989 pp.223-224]

[g] Use of cartograms might also be beneficial:
There are several cartographic solutions to the small number problem: map classification, data suppression, proportional symbol maps, and cartograms. [Kennedy S. 1989 p.191]

[h] No amount of technical wizardry can substitute for the careful collection of good quality information:
Despite the current focus on error induced during the analysis of spatial data, it is far too easy to search for the classic technological fix to problems in datasets. One thing the use of GIS systems should provide us is the realisation that increased attention to the source of data themselves will alleviate more problems than all the combined technological toolkits and error correction algorithms built into commercial GIS packages. [Brusegard D. and Menger G. 1989 p.185]

27 Which collections of people should we study?:

In terms of size, the basic demographic unit (Cox, P.R. p.4) is the individual person, while the largest is the whole human species — a species, incidentally, which is clearly distinguished from all other species — but where in the continuum between these two extremes do we draw boundaries or thresholds? A demographer would see no real difficulty here as he does not demand that the population should be distinctive areally, socially or in any other way. Many of his populations are hypothetical. [Clarke J.I. 1975 p.1]

28 [a] City sized units are sometimes, unfortunately, referred to as being local. Defence of their creation is interesting:
It is a basic tenet of this work that the appropriate scale to examine spatial variations in economic development in Britain is not the major regions but the local labour market area — that is the area where the vast majority of people live and work, where they can change jobs without changing house. While location with respect to the overall national territory will obviously be important (in the sense that the development potential in otherwise similar areas in core and peripheral regions may be different) it can be argued that insufficient attention has been given in the past to more localized characteristics of places which in certain instances may enable an area to either overcome or fail to capitalize upon its broader regional situation. [Goddard J.B. 1983 p.21]

[b] Numerous arguments have been presented claiming that functional regions are natural, for instance:
The concept of functionally defined areas is not an easy one to grasp, because most people have strong attachments only to a very localized area such as a particular neighbourhood or even a single street. Possibly the best way of appreciating these ideas on the meaning of "places" is in terms of being asked where one lives by people who live a long way away. If someone from the North East goes on holiday or a business trip to the South East and is asked where he comes from, he is not likely to venture "Osborne Avenue" or even "Jesmond" but is much more likely to reply "Newcastle upon Tyne", of which that locality is a part. Similarly, a sixth-form student being interviewed for a place in the geography department at Newcastle University is much more likely to say he comes from "London" rather than from "Islington" or even "Bexley". The same applies to a person who comes from a dormitory settlement on the outskirts of a city, in that such an individual will probably volunteer the name of that city. This is the scale of the "places" with which this book largely deals. [Champion A.G., Green A.E., Owen D.W., Ellin D.J. & Coombes M.G. 1987 p.6]

[c] A problem of scales haunts the use of functional regions:
Many commentators have been pointed to the existence in the North of prosperous places and areas of deprivation within the South. However, there has been much confusion on this issue, largely because of inconsistencies in the scale of analysis adopted. The most common fallacy has been to compare inner city problems — which are characteristically localised problems of residential deprivation, some in generally prosperous labour market areas — with the problems of job shortfalls that characterise the labour market as a whole. [Goddard J.B. & Coombes M.G. 1987 p.1]

[d] The great variations in functional region size make some patterns more likely to appear than others:

Hence, an examination of the evidence on economic prosperity and power reveals a clear North-South divide: the bulk of the most prosperous LLMAs are located south of a line from the Severn to Lincolnshire, with particular concentration of the best performing LLMAs in the South East outside London and on the neighbouring fringes of East Anglia and the South West. [Green A. 1988 p.184]

[e] Other researchers have noticed some of these problems:
The CURDS use of relative indices biases their account by emphasizing the growth of rural areas. [Savage M. 1989 p.255]

[f] And other organisations have become exasperated over the images created and the claims that some comparisons cannot be made:
Alongside the image of a booming high-tech economy and stockbroker belt commuter village, relatively little attention is paid to the other face of the South East — the unemployed, low pay and poverty — on the opposite side of the 'South-South divide' (Seeds, 1987). Conversely, one commentator has claimed that if a North-South divide exists... 'Harrogate should be on the South coast, not three hours north of Watford' (Rice, 1987).
Despite the existence of some pronounced inter-LLMA variations at the intra-regional scale, the majority of those challenging the existence of a North-South divide focus their attention on micro-scale variations at the intra-urban scale, or on inequalities between deprived inner city areas and prosperous market towns. As noted in an earlier subsection, scale is of crucial importance in examining the evidence for a North-South divide, and it is significant that contrasts between conditions prevailing in selected inner city wards and prosperous market towns involve non-comparable areal units. [Green A. 1988 pp.191-193]

[g] Advocates of a functional regions perspective have made great claims:
The geographical units adopted for this study are the 280 Local Labour Market Areas (LLMAs) of the Functional Regions framework (Coombes et al. 1982). This urban-centred functional approach divides the country up into a set of real places that are relatively independent and on which the quality of life of the local inhabitants largely depends. [Champion T. & Green A. 1989 p.63]

[h] We must always remember that it is people who function — not the space which contains them:
Enumeration-district and grid-square data do provide a useful basis for elementary descriptions of areas, either cartographically (Dewdney and Rhind, 1975) or statistically (Webber, 1977). Nevertheless they must remain descriptions of what Chapman (1977, page 55) terms "areal aggregates" and not of objects or true entities per se. Essentially Chapman argues that objects of interest should have the properties of systems. In primitive terms this means that the 'whole' should be greater than the sum of the parts, or else we are dealing merely with an aggregate. This property of 'wholeness' of an object is reflected in its relative self-containment of activity. Boundaries around the object are characterised as zones of greater impermeability between the object and the outside world. Finally the key characteristic of a true object is the existence of some control mechanism within the object. Hence objects 'respond' to stimuli and therefore can be said to 'behave'. Mere aggregates do not behave. [Coombes M.G., Dixon J.S., Goddard J.B., Openshaw S. and Taylor P.J. 1978 p.1181]

29 [a] New calls for a place perspective are not for more functional regions:
I want to be clear at the outset that most recent arguments for a place-based geography are not calls for the revival of regional geography based on either the immutable physical landscapes of fixed regions at a scale intermediate between locality and national unit or descriptive inventories of regional characteristics regarded as if they are independent of social order. The conception of place involved in these approaches could not be expected to excite much interest in social science. [Agnew J.A. 1989 p.9]

[b] Other researchers also disagree with the use of such areas in analysing social structure:
However, within such areas there can be significant changes in population and, for the purposes of this study, FRs do not produce enough local detail nor are 1971 Census statistics readily available for these areas: as a result, they are not appropriate for use as an areal base in local study. [McKee C. 1989 p.6]

[c] An area slightly larger than Greater London is often classed as a single labour market:
The multivariate classification confirms that Greater London is best not taken as a single unit because it divides sharply between Inner and Outer London. [Craig J. 1980 p.23]

[d] Although other classifications split the capital in two:
The fact that Greater London has been assessed as falling in two separate travel-to-work-areas but one LLMA starkly illustrates this point. Not since the Great Plague of 1665 has a cordon sanitaire been imposed at London's periphery, blocking entry and exit! (Defoe, 1727). [Merrett S. & Sharp C. 1991 p.290]

30 [a] Advocates of grid square mapping have occasionally confused orders of magnitude:
Enormous variations in base population are typical of grid squares but also occur in EDs ... [Rhind D. 1983 p.159]

[b] Small scale density mapping is not a particularly new idea:
A time series of such proximity measures (for 1570, 1700, 1801 and 1851) were shown diagrammatically in the 1851 Census Report. A large circle represented the area of England and Wales and inscribed within each were sets of six-sided polygons whose size was proportional to the density of population over the country as a whole, and to the average distance apart and 'average amount of ground to each person'. Variations in density were shown in a detailed dot-type map — the areal unit used is not explained but it is very small and the overall effect is not dissimilar from the maps produced 120 years later when computerised cartography led to the production of enumeration district and 1 km grid square maps. The measures of proximity mentioned in the preceding quotation were also given for each of the nine divisions (analogous to regions) of England and Wales. [Craig J. 1987 p.34]

[c] It is telling that grid maps show more empty land than populated kilometre squares:
Hence, the maps in People in Britain (CRU/OPCS/GRO(S) 1980) are the first reliable maps of unpopulated areas in Britain! The second consequence is that related to data suppression: because of their fixed size, many more grid squares than EDs are suppressed (section 2.9.6) in census data and as a consequence mapping of the grid-square data is not possible for a large percentage of the rural areas of the country. [Rhind D. 1983 p.181]

[d] But the protagonists of the technique see other reasons why it can be so unpopular:
But much the most devastating disadvantage that grid squares have for these purposes is that most people do not yet like them. They are certainly unfamiliar as a means of data referencing and presentation. At the time of writing (November 1974), for instance, only one group had any 1971 Census grid square data for England or Wales. More than this unfamiliarity though, is the deep-seated, if irrational, feeling that they are distinctly 'unnatural', while 'anything which wiggles' is 'natural'. [Rhind D. 1975 p.3]

31 [a] The approach taken here has been tentatively suggested in the past:
There is however no theoretical reason why the ward or parish should not be the basic unit in a study of the country as a whole. There are some practical problems of separating the wood from the trees in that the amount of detailed information that is available is large. However it is still possible to summarise it even though some of the techniques which are helpful with a smaller amount of data (mapping, for example,) may be impractical. [Craig J. 1980 p.14]

[b] Many may consider my approach impractical:
Anyone at this stage writing their own software of this type is therefore foolhardy. All of these maps, however (with the exception of those by OPCS) deal with only up to about 1,000 or so zones on any one sheet of paper. [Rhind D., Mounsey H. & Shepherd J. 1984 p.65]

[c] It is the availability of a high resolution printer which has made such fine scale mapping possible:
All the information presented on a map is to be read by eye from paper. The resolving power of the eye enables it to differentiate to 1/10mm where provoked to do so. Clearly, therefore, conciseness is of the essence and high resolution graphics are a common denominator of cartography. There are tricks, too, of colors, lines and point symbols that suggest that there is some subtlety within the cartographic language. Here the capabilities of the high resolution plotter in a computer system offer interesting potential to extend the language as opposed only to mimicking it. [Bickmore D.P. 1975 p.331]

[d] The realization that this technical equipment could be used for more than just mimicking manual methods was also crucial:
Nevertheless, by the early 1980s it had become possible, given adequate investment in software and hardware, to replace traditional mapmaking with digital technology. Although the proportion of maps created using digital methods is still relatively small, the technology has advanced to the point where it is impossible to determine by cursory examination which approach has been used to generate the final product. The critical point was probably passed in about 1980. The hardware developments and multi-year investments in software have advanced digital technology to the point where its constraints are no more restrictive than those of pen and paper.
One of the surprising aspects of these developments, and of applications of digital technology in many other areas as well, is the degree to which perfect emulation of the conventional product appears to be regarded as a legitimate objective. Word processing technology has been developed to the point where it is a more cost-effective way to generate the conventional typed text. Surprisingly, it has not been used to explore alternative methods of presentation to the same extent. Yet neither word processing nor digital cartography are constrained by the same limitations that produced the printed page or analogue map. [Goodchild M.F. 1988 p.316]

[e] It is the commercial software, which is of poor quality — that is now coming into direct conflict with visualization:
A resulting phenomenon is the "application expert", the man or woman who knows all the ins, outs and quirks of ARC/INFO, and who is crucial within the organisation for getting questions — more or less — answered, even to the extent of writing "user-friendly" macro systems. Although this will not change much over a short period, in the long term there is not much future to it. Especially in the field of visualization, developments have already set in to a much more modular approach, in which techniques like filtering or enhancing can be applied independently to GIS applications. [Hartmann J.L. 1991 p.411]

32 [a] There is good reason for using such a fine spatial scale:
Moreover, analyses should be carried out at the scale of the operation of the processes whose manifestations are being observed (Rhind et al 1984); without this, results are very likely to be biased. If the scale at which a process is occurring is unknown, it is best to at least examine the data in the most disaggregated form available. There is, in addition, good reason for believing much of the development — through pervasion — is locally varied. It is localised conditions which dictate whether demographic and socio-economic change exist, and the exact form it takes; therefore an emphasis on local-scale study is necessary to avoid making over-generalisations, particularly in areas of high population density where large-area study will not reveal local scale change. [McKee C.H. 1989 p.49]

[b] Other advantages also accrue when working with enumeration districts:
More generally, these problems of heterogeneous geographical units and the prospect of massive ecological fallacies are common to much census analysis. It is obvious that they may be reduced — but not eliminated — by concentrating on the analysis of EDs. [Openshaw S. 1983 p.247]

[c] But some problems remain:
The problem of object location will be less troublesome if the subareas are small. If it is possible to use small subareas, the number of possible locations for an object is small and the generated patterns, though still indeterminate representations, will be more concise than otherwise. In any case, analysis must proceed using grouped data and it may be wasteful to attempt to use sample subareas of very small size. Aesthetically, there may be justification for proceeding with small areas, but the effect of size limitations on probabilities (attributes) may become more difficult to handle. [Getis A. 1974 p.85]

33 [a] It would be interesting to discover how much detail we can profitably appreciate in an image:
Of all the research reported in this review, the one factor that constantly changes and has the potential to be a deciding factor and influence is the number of statistical units contained in the test maps. For example, the number ranges from approximately 1000, (maps produced by the U.S. Bureau of the Census), to only nine fairly large areas (Cartstensen 1982:66). Perhaps a limit exists pertaining to the number of statistical units on a two variable choropleth map that the map reader can process easily. [Halliday S.M. 1987 p.37]

[b] Visualization raises the limits:
Significantly more complexity can be comprehended through Visualization in Scientific Computing techniques than through classical ones. [McCormick B.H. et al 1987 p.vii]

[c] Again, generalization is often not necessary:
We believe that it is possible to push the resolving power of the eye to the point where this detail can be understood without complicating the issue by statistical generalization. [Bickmore D.P. 1975 p.344]

[d] But problems can arise from using many small areas:
Amalgamations may also be necessary when mapping at the enumeration-district level, depending upon the size of the study area. With very large areas, for example in excess of five hundred enumeration districts, the amount of printed shading that can be applied to each unit is limited if the size of the computer output is to remain within reasonable bounds. Some smaller enumeration districts may receive only one or two printed symbols, which renders detailed inspection of individual units virtually impossible. [Kirby A.M. and Tarn D. 1976 p.510]

[e] In particular, cartograms greatly reduce the misleading impressions created by small populations:
It is also obvious that mapping at high geographical resolution maximizes the chance of encountering small groups of people in each area: indeed, this is the rationale for high resolution, since in dealing with census data we are interested in people rather than areas. With small groups of people, however, percentages which are not based on land area ('percentage of the population which is male' c.f. 'density' of the population per km2) are frequently very unstable: the highest and lowest values, and consequently the most visually striking areas, on a map may therefore commonly be based upon very few people. Most mapping of detailed census data has ignored this effect which is rampant in use of ED and 1 km2 grid-square data. Indeed, many authors have been even more misleading: where the population is small and also clustered in one part or another of the area being mapped, they have none the less shaded the whole area and greatly extended the misleading effect of small populations. [Rhind D. 1983 p.183]

34 [a] Making the point more strongly:
Therefore, and this point is played down by Giddens, place is not just locale, as setting for activity and social interaction, but also location. The reproduction and transformation of social relations must take place somewhere. [Agnew J.A. 1987 p.27]

[b] Place and space are evolving concepts:
New Zealand is not on the other side of the world from England. Mentally shrinking the oceans to one-tenth their size places England closer to New Zealand than to Hungary. The return to the seas "shrank the earth" and turned the land inside out. [Bunge W.W. 1973 p.280]

[c] It is as much the movement between places, as the people within them, that interests us:
Place, therefore, refers to discrete if 'elastic' areas in which people can identify. The "paths" and "projects" of everyday life, to use the language of time-geography, provide the practical "glue" for place in these three senses (Pred 1984). [Agnew J.A. 1987 p.28]

35 [a] Remember, it is people, not land, we are looking at:
Cities do not only exist in the physical property of the cities, they also exist where the citizens of the city move. We are so property bound that we think a man lives at the address of his bed, even if he ever visits it. The cities could be defined as their people as apposed to the property of the city. Using such a definition, the city has moved if the people evacuate it, say on a weekend holiday. [Bunge W.W. 1973 p.293]

[b] But many aspects of the places, in which people live, have an influence over their behaviour:
... the "factors" causing political behaviour cannot just be added up in linear fashion (census class, census age, census ethnicity, etc.) to constitute an adequate explanation. To the contrary, it is how these factors "come together", take on meaning for people, and determine political outcomes that constitutes a satisfactory political analysis. In other words it is in places that causes produce the reasons that produce political behaviour. Many of these causes, as I have argued emanate from other places beyond the confines of the locality. But it is in the locality and through the choices of local populations that the various causes of political behaviour, from the shifting economic position of the US automobile industry, in the case of Detroit, to Castro's Revolution in Cuba, in the case of Miami, work to structure political expression. [Agnew J.A. 1987 p.213]

[c] And we have still to consider time properly:
A geography of regions within the modern world-system must locate its regions in both space and time. [Taylor P.J. 1991 p.28]