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 1: Envisioning Information

We must create a new language, consider a transitory state of new illusions and layers of validity and accept the possibility that there may be no language to describe ultimate reality, beyond the language of visions.
[Denes A. 1979 p.3]

1.1 Visual Thinking


Envisioning means bringing into the condition of vision, making visible, to enable visualization. It is what this thesis practises. Here the theory behind it is presented. Envisioning is about giving information to people who can see10. I argue that there are dramatic potential advantages in using visual images to allow people to unravel the spatial patterns in complex social structures (Muehrcke P.C. 1969, Arnheim R. 1970, Bertin J. 1981, 1983a, Marr D. 1982, Tufte E.R. 1990).

Although I have used computers a great deal in this work, I am not going to concentrate on the mechanics of getting information into the machine, but how you get it out to people (Prints XI & XII). To communicate with people you must involve their senses of sight or hearing, the former transmitting far more information than the latter. Language, along with music, the most sophisticated use of hearing, is an excellent means of conveying ideas and thoughts, but cannot present a large amount of information in a structured form at speed11.

When you look out of the window you can see a great deal in an instant. The mind has an extremely powerful system for processing imagery which can instantly analyse a pattern of colours, of light and shade, and know that these are trees, houses or people out there. How long would it take to describe all that you can see in words? Yet we still have to argue, that in the study of societies, there are many things which cannot be eloquently described in words or succinctly captured by equations.

This very thesis is only held together by its text. We have come a long way with our little symbols, which, after all, exist only because they were easy to scratch with a stick or form quickly with lips and tongue. Did our ancestors develop the most efficient means of communication, or did they make do with what was possible? Communication, which holds a society together, is still developing. We are only beginning to realize what there is to see.

The spatial structure of British society, which is envisaged in these pages, is made up of far more than a few large regions which can be named, and divisions which can be measured. Social structure has a texture to it, a fine pattern, an elaborate organization, not unlike the pattern of chaos (Print XIII). Such intricate structures cannot be captured by writings which say which towns are supposedly faring worst, or coefficients that tell of a simple widening of the divisions. If we want to know the how and why of things, the best we can do, before letting our imaginations take over, is to take a look at what we are talking about.

We depend on vision, we think visually, we talk in visual idioms and we dream in pictures, but we cannot easily turn a picture in our mind into something other people can see. An artist will take days to paint a single portrait. Suddenly, just as the last generation was given the camera, we have received the computer, which can turn a huge amount of data into pictures — snapshots of our society. In the future we will be able to speak visually. For now we still have to learn how12.

1.2 Pictures Over Time


Visual communication was possible in the past, but enormously time consuming and often limited by poor materials and little information. These limitations led to restricted experimentation and strongly established conventions as to the right way to paint. Our first permanent communications were cave paintings and our first textual scripts made of pictures. Today the computer window system which abounds with icons is the modern cave wall (Print XIV); we have rushed forward to the beginnings of visual communication13 (Peddle J.B. 1910, Riggleman J.R. 1936, Royston E. 1956, 1970, Lockwood A. 1969, Herdeg W. 1974, Feinberg B.M. & Franklin C.A. 1975, Beniger J.R. 1976, Beniger J.R. & Robyn D.L. 1978).

The first maps were drawn on clay. They were invaluable objects for the control of territory or the projection of religious truth about the world. Maps were accumulations of innumerable stories, reams of parchment and hordes of figures. Spatial information about the world and its people has always been at the forefront of visualization. As map-making developed into the art of cartography, rules were formalized and conventions defined (Peuker T.K. 1972, Friis H.R. 1974, Bertin J. 1978, Howe G.M. 1986c). Cartography is no longer a major discipline or even an important aspect of geography. Its modern tools can be used by children (Print XV) and its conventions are being challenged as stale.

The nineteenth century saw the strongest moves, in science, against pictures. The graphs, which instruments traced onto paper, were immediately turned into supposedly more accurate and readable tables. Diagrams were for people without mathematical imagination. Nevertheless statistical graphics did germinate in these surroundings. The graph, bar chart and scatter diagram were invented. These, too, were formalized, rules for their construction produced, while their supposed subservience to more advanced methods was made clear. Now the cycle has come round again, and there is a new breed of statisticians who see visualization as paramount (Fienberg S.E. 1979, Young F.W., Kent D.P. & Kuhfeld W.F. 1988, Buja A., Asimov D., Hurley C. & McDonald J.A. 1988, Crawford S.L. & Fall T.C. 1990, Hirsh N. & Brown B.L. 1990).

Computer graphics in the 1960s changed the picture14. Swirling images were produced from the most simple formulae (Davis P.J. 1974, Mandelbrot B.B. 1983, Andrews D.F., Fowlkes E.B. & Tukey P.A. 1988). It was immediately obvious that reading an equation told you little about what secrets it held. Before computer graphics, people were blind to the behaviour of relationships they thought they could easily understand (Print XVI). The programmers then went on to render reality — creating photographs from numerical descriptions of what we can already see around us. They now turn their efforts to the possibilities of rendering abstract worlds.

Visualization grew out of all of this, but a similar philosophy underlied much of it. Graphics have come in and out of favour in cycles through time (Pickett R.M. & White B.W. 1966, Baecker R.M. 1973, Neal M. 1988, Nielson G.M. 1989, Anderson G.C. 1989, Voegele K. 1990a). Their resurgences usually have more to do with taking advantage of new printing technologies (Figure 1) and the availability of more abundant information, than a basic understanding of their value. What is required now is to harness the potential of the computer, that both provides and renders new information, for a deeper knowledge.

1.3 Beyond Illustration


Visualization is a way of working, a methodology. Not only does it differ from the use of script and figures — reading and calculating to understand — but also from conventional graphics which aim to illustrate. Illustration is used to convey a discovery from one person to another which was found by other means. Visualization is the transformation of numbers into pictures in order to see what a mass of figures could not tell us, let alone inform others. Visualization is how discovery is made. The method is the message15 (McCormick B.H. et al. 1987, Prueitt M.L. 1987, Winkler K.H.A., Chalmers J.W., Hodson S.W., Woodward P.R. & Zabusky N.J. 1987, Wolff R.S. 1988, Forer P., Poiker T., Penny J. & Deeker G. 1990, Robertson P.K. 1990, Nielson G.M., Shriver B. & Rosenblum L.J. (eds) 1990, Foley J.D., Dam A. van, Feiner S.K. & Hughes J.F. 1990).

Most visualization research today relies on huge quantities of numerical information. Before you have such information, you can only write about what you think is happening. Now you have counted what is happening, who does what, who has what — how do you understand it? How should we analyse the information? Statistical analysis gives you single figures, averages, correlations, parameters of assumed relationships, probabilities, and so on. They are only of use if you know exactly what you want, but knowing what questions to ask is much harder than finding the answers. Social science is not about defining and testing simple hypotheses; it is about understanding complex societies.

There are many ways to begin studying society. All involve some form of ordering, of which the spatial is the most common. Having projected our figures onto the plane in some way, we can paint pictures of this ordering and see what patterns emerge, what structure there is (Print XVII). These patterns usually turn out to show complex and subtle relationships that tax our mental capabilities to comprehend and explain. This is not a bad thing — stretching the mind forces the imagination. The visual methods I am discussing take hundreds of pages full of tables of thousands of digits, and turn them into a single picture with little loss of detail, in order to see what there is to see. In terms of storage, most of the pictures in this dissertation required more disc space individually, than the entire (typeset) text.

Illustration is to clarify — to make clear, pure or transparent. Visualization does not aim to see through our information, it aims to see into it. Methodology is about transforming reality to fit inside particular conceptions. The more we simplify, the more reality is blurred. Turning people and the events of their lives into numbers is bad enough. Throwing away almost all of those numbers is worse, and yet this is what we must do, in one elaborate form or another, if we are to try and understand without images.

1.4 Texture and Colour


If we are to envisage information we must first know what can be seen as well as what there is to see. To decide how to turn numbers into pictures we must know what pictures can contain and what is seen in them. The most simple pictures are constructed of pure black and white from basic geometrical shapes (Bachi R. 1968, Hunt A.J. 1968, Tobler W.R. 1973b). What they contain, what the eye searches for, is pattern — from order, repetition, grouping and texture16. What the eye then does, is to find breaks in that order, discover inconsistencies while ignoring irrelevancy. The eye does this because that is what it evolved to do, and to do so extremely quickly.

The eyes are constantly engaged in focusing, panning and zooming. They compare different sections of the image and home in on interesting detail (the eyes are designed to scan continuously — they cannot focus for long on a fixed point). The resolution of the eyes is enormous, but far finer at the point on which they are centred. This action can be mimicked and aided when pictures are electronically produced, and can be instantaneously enlarged or reduced. Focusing is one of the simplest attributes of vision, yet we know very little about how even it operates17.

Colour is an invaluable embellishment to basic vision (Figure 2). It is wrong to think of it either as adding another dimension or merely supplying some further minor tagging of data to existing features of the graphic. It alters the character of the image (Staudhammer J. 1975, Mersey J.E. 1984, Lindenberg R.E. 1986, Atkinson D.S. 1988, Kumler M.P. 1988, Levkowitz H. 1988, Pham B. 1990, Hopgood F.R.A. 1991). Different colours are perceived variably and convey loaded meanings on their own, even more so in combination. The human eye is poor at focusing on blue. Red and green do not combine to form reddish-green, and so on. Colour adds another level, but not dimension, of complexity. The careful use of colour can convey more of the depth of organization we wish to comprehend. In particular, when used in bivariate and trivariate mapping18 (Print XVIII).

When we only wish to show a single ordering, grey scale shading is most effective and appropriate. To depict the bivariate relationships between two variables in the same place and between places, colour has been shown to produce effective keys when carefully employed. This thesis makes great use of trivariate colour schemes to show the combination of individual levels of up to three independent characteristics (Print XIX). This is both a contentious and potentially highly effective technique (Sen P.K. 1960, Dannatt L.K. 1981, Cowen D.J. 1984, Olson J.M. 1987a, Halliday S.M. 1987, Dawsey C.B. 1989, Fels J.E. 1990). It has been suggested that the printer's primary triplet of cyan, yellow and magenta be employed (or the computer's red, blue and green)19. The most intuitively appealing combination was found to be the painter's red, blue and yellow — which fortunately also coincided with Britain's major political parties' symbols.

1.5 Perspective and Detail


The most powerful ability of the eye-mind combination which is employed by visualization is generalization (Tobler W.R. 1968, 1969a, 1989b, Rhind D. 1975c, Card S.K., Pavel M. & Farrell J.E. 1985, Lavin S. 1986, Bracken I. & Martin D. 1989, Gilmartin P. & Shelton E. 1989, Herzog A. 1989). The brain only ever sees and understands through constant physical generalization of the light intensities which are measured by the retina. These are smoothed by the mind to allow instant assumptions to be made, before more careful inspection is undertaken20 (Prints XX, XXI & XXII). Such ability is essential to our survival in everyday life; it was even more so in the past. Through visualization we are utilizing one of the most finely tuned pieces of evolutionary good fortune.

We live in a three-dimensional world, despite having essentially two-dimensional vision. Perspective is the name given to the effect of projecting a three-dimensional scene onto our two dimensional retinas; we use it to try to reconstruct three-dimensional form. Although we do have binocular vision, if you close one eye you lose little feel for the three-dimensional reality. We generally only move about in two-dimensions and, in fact, have a far weaker grasp of the real three-dimensional world than we may imagine.

It is often claimed that expensive equipment which allows volumes to be created and seen is at the forefront of visualization. Stereoscopic vision, though, might not be as great an asset to visualization as it is often thought to be in seeing the real world. Stereo vision works well at gauging position when nothing is moving behind or in front of anything else. Once things begin to move though, it becomes an irrelevancy. In visualization, if we want things to move, then, through animation, they move.

Animation can be used for much more than understanding three-dimensional form. As the creation of a changing or moving image it can add another level of sophistication to two-dimensional visualization. However, like colour, it is not the same type of dimension as the spatial. In animation things must change smoothly and relatively slowly. If objects change their colour it can confuse; if too much is happening we will not have enough time to comprehend it. Surprisingly, animation takes us back towards illustration. It requires simplicity to work. Far more useful is interactive graphics — moving pictures which the viewer controls. This is not only control over how fast or slow or where the pictures move, but simultaneously over what they contain and how it is presented. This is the next step in visualization.

1.6 Pattern and Illusion


We do not think in a three dimensional geometry — many tests have shown this (Parslow R., 1987). The geometry of visual thinking is essentially two-dimensional. We also have a poor visual memory; we remember what we extract from images rather than the images themselves. Furthermore, the emotional overtones of colour are perceived differently by different people. The colour blind cannot see the full trivariate range.

Visualization can achieve a great deal, even when limited to static two-dimensional images (Wong W. 1972, White R.D. 1984, Wood D. 1985, MacEachren A.M. 1987, Simkin D.K. 1987, Kennie T.J.M. & McLaren R.A. 1988, Buttenfield B.P. & Ganter J.H. 1990, Freeman S. 1991, Hartmann J.L. 1991, McAbee J.L. 1991). This goes against many of the embryonic tenets of the field, but it is questionable how much they are guided by what is possible, rather than what is desirable. Why use the illusion of three dimensions if it adds so little information to an image while causing so much confusion? Perspective views are pretty and still somewhat novel, but not especially useful unless it is three-dimensional geometry in which you are particularly interested.

Animation, like perspective viewing, is also not as invaluable as has been claimed. You cannot hold a moving picture in your mind as well as you can hold a static image, and comparison of two dynamics is difficult. Animation can tell a story. Visualization, more often, allows you to find a story to tell (Prints XXIII, XXIV & XXV). Much more importantly, with both animation and perspective views, you are limited to producing very simple pictures if you are to be able to understand them. Both ideas are included in this work and they produce nice illustrations, but until the viewer can easily control what is viewed, through interactive visualization, their utility is limited.

The use of colour greatly augments what can be seen in a two-dimensional image. However, use of colour is expensive, and duplicating these prints was not easy21. Colour can also add difficulty, and invoke unintended ideas (good and bad, hot and cold, near and distant hues — often used, intentionally, as such in conventional atlases). In this dissertation colour is not included to make the pictures prettier. I have used colour to include extra information in the image and to show how to display more complex data sets. Often it is not used (as it can be) to simplify understanding of the images, but is added as a final embellishment to elaborate on how the complexity continues as other facets of the social structure are connected (Print XXVI).

1.7 From Mind to Mind


The argument in this chapter has developed from the initial desire to allow people to convey what is in their mind, in a form others can see, to the point where individuals are able to see and paint their own information22. It is as if we had all been mute, and suddenly were able, with the aid of a machine, to make sounds — what sounds should we make? In the past we made sounds by knocking sticks together, so we get the machines to imitate those noises. But surely, we think, there is more (Huggins W. 1973, Evans D. 1973, Mills M.I. 1981, Kosslyn S.M. 1983, Farah M.J. 1988)?

Our vision has a much higher bandwidth than our hearing, far greater scope for communication. We can see thousands of stars, watch sunsets, view landscapes, survey half a million people in a crowd. Naturally we begin to paint things by getting them to look like recognisable objects, chaotic functions to look like mountain ranges or an island archipelago, flowing energy to appear as running water. In this work pictures are often based on natural things which have two dimensional structure, from honeycombs and cobwebs to crowds of upturned faces and flocks of arrows (Prints XXVII & XXVIII).

Here visualization is used to make millions of figures understandable without massacring their meaning, without reducing them to tables, graphs, crude maps or models. If we are to understand the structure of society we must find ways of envisaging it23. This dissertation demonstrates how large amounts of simple information can be shown, and then goes on to increase the potential of the graphics by conveying increasingly complex information.

I have a collection of thoughts and prejudices about British social history over the last two decades. I am going to convey them to you through the medium of images, rather than trying to persuade you with words, convince you with tables, or confuse you with equations. I am going to use government sample and census figures (Figure 3), rather than newspaper cuttings or extended interviews. What I am doing may appear subjective, but it is no more so than any other method. The images shown should be very different from anything you have seen before and probably not instantly meaningful.

The pictures shown here are of things which cannot be easily (or adequately) described, discussed or modelled, and yet many people who see them expect to understand them in an instant, even when they may fail to understand the long complicated narratives which explain them badly, or the intricate mathematical models which could represent them inadequately . If you want to know the shape of Britain you look at a map. You can then go on to investigate rivers and mountains, lakes and bays. It is shown, through this dissertation, how you can look at the shape of British society with visualization and then it is possible to discuss the implications. However, to know the shape you have to look at the picture (Print XXIX), you cannot just describe it in words.

You learn to see, as you learn everything else, from experience. Here, we gain some experience of what there is to be seen, using the first generation of machines able to paint such pictures. The charcoal stick has come a long way.

Prints

XI Stills from a conventional animation of the computer (Colour).
XII Stills from a ray-traced animation of the computer (Colour).
XIII Ray-traced surfaces of the Mandelbrot and Julia sets.
XIV Visualizing Fourier transforms — the art in the science (Colour).
XV A maze of colour — the detail a low resolution image can show (Colour).
XVI Visualization of the Mandelbrot set — magnification and generalization (Colour).
XVII Travel time from the Tyneside road network (Colour).
XVIII Three alternative colour schemes and keys (Colour).
XIX The concentration of British born place of birth (Colour).
XX The distributions of population, age, gender and children in London (Colour).
XXI The distributions of place of birth in London (Colour).
XXII The distributions of employment, occupation and graduates in London (Colour).
XXIII The distribution of broad industrial groups in Britain, 1987 (Colour).
XXIV The changing distribution of broad industrial groups, 1984-87, increases (Colour).
XXV The changing distribution of broad industrial groups, 1984-87, decreases (Colour).
XXVI The change in employment by industry, status and gender, 1984-1987 (Colour).
XXVII Political swing on the electoral cartograms of Northern Britain (Colour).
XXVIII Political swing on the electoral cartograms of Southern Britain (Colour).
XXIX The distribution of voting in English and Welsh local elections (Colour).

 


Drawfiles are used to create the illustrations in this thesis.

A library of procedures was written specifically to produce these files.

Drawfiles are a sophisticated type of computer record. The record contains a list of objects, which can themselves be lists of objects.

Objects can include relationships (with other objects), information (data from other files) and:
text - of a particular font, size, style and colour;
sprites - a pixelmap image (raster graphics);
paths - lines, curves and shapes (vector graphics).

In the example above the Greater London "object" has been shrunk. In the drawfile it is tagged with its identification as County no.1 and the relevant boundary date (1981). Making up the group is its perimeter, the river Thames, and any islands in the river. All aspects
of scaling, appropriate placement and hyphenation of names and colouring are automated. Any feature of an object or group of objects can then be edited - interactively on the screen - as has been done here.

Once a drawfile representing a particular geography has been created, it can be transformed and additional information incorporated. For example, the places could be represented by faces instead of polygons, re-coloured and then merged with another drawfile.

Figure 1: Creating the Graphics


The device used to print the colour illustrations in this thesis was a Colourview 5912 plotter printer manufactured by Calcomp in 1988.The plotter can produce pixels of eight colours by overlaying sheets of magenta, cyan and yellow film with an A4 resolution of 2048 by 1600 pixels (3200 by 2048 when A3). A greater range of colour is possible by using dithered patterns of the eight colours actually available. Text could be more satisfactorily produced through the use of anti-aliasing techniques built into the computer software.

A driver was specifically written to convert red-green-blue output from the screen, to the magenta-cyan-yellow form suitable for printing.

Each A4 print is made up of over three million individual bits of information, and took half an hour's printing time.

Figure 2: Printing in Colour


All information about places, concise enough to be edited manually, was stored in "Comma Separated Value" (CSV) files. These files can be read by many applications on several computer systems, in particular by spreadsheets, allowing complex manipulation to be accomplished easily. An example of the beginning of a CSV file containing information used to create a drawfile of counties is:

"$.GIS.Area.Ward.County.Sheet",1,64,4
"County Topology and Statistics"
"Number","Name","Residents","Nbours","Neighbour"
"1981"
1,"Greater London",6713130,6,0,29,22,43,26,11
29,"Kent",1467079,5,0,1,21,43,22
22,"Essex",1474126,6,0,12,42,1,26,29
43,"Surrey",1004332,8,21,45,1,29,0,24,10,11
26,"Essex",1474126,6,0,12,42,1,26,29
11,"Buckinghamshire",567979,7,43,26,34,9,38,1,10 . . .

The first line gives the filename, number of tables, number of areal units and fixed variables. The second line describes the file, the third has variable names and the fourth holds temporal information. This header is followed by the relevant numbers and text for each place. Notice that the records can be in any order and of variable length. They can be easily edited as this is a text file.

A library of procedures was written to manipulate these files; allowing any other application to read and write to them, taking advantage of an interpreted language which allowed the procedures themselves to invoke routines from the applications which had called them.


Figure 3: Recording the Places


10 [a] The value of visualization was also appreciated in the past:
Often we must deal with conditions where no known equations will connect our experimental results and where a mere tabulation of figures will not yield the desired information without much tedious study. The well recognized superiority of any graphical representation over an equation or table in conveying a clear impression to the mind of the way in which a set of variables is related will often in itself be a sufficient justification for the use of this type of chart. [Peddle J.B. 1910 p.98]

[b] Visualization is a way of doing research, not just a technique for presenting results:
In conclusion, visualization should not be viewed as the end result of a process of scientific analysis, but rather as the process itself. More than simply the application of techniques for displaying data, visualization can be used as a paradigm for exploring regions of untapped reservoirs of knowledge. The "Knowledge Navigator" discussed by Apple's John Sculley in his book Odyssey, is, in some sense, the perfect model for the visualization process. Jim Blinn has used this process for over a decade in attempting to simulate planetary exploration by modelling Voyager's journey through the solar system. Visualization is not new, but its awareness by the general scientific community is. [Wolff R.S. 1988 p.35]

[c] The ideas visualization are easily applied to mapping:
Scientists confronted with conceptually difficult processes plot numbers on graphs to "see" what they mean, often under the assumption that even bad graphs may provide more meaning than tidy lists of numbers. Normally we need all the insight we can get, and graphics are closely associated with the intuition that lies behind so much creative inquiry. The computer business increasingly uses pictorial output. Graphics are used in basic research in engineering, mathematics, physics, and other fields as a means of visualizing complex formulas and models. The map, as a graphic form of symbolic representation, also serves the primary function of visualization in scientific research (Figure 30).
It appears that maps (or graphics) are not designed, intended, or well suited for precision work. One should not expect detailed statistics from mapping. The impact of the map is more often of greater importance than the information. Maps serve well the need for a general picture of the nature of a distribution or the relationships between several distributions, at least when the patterns are not too large. [Muehrcke P. 1972 p.38]

[d] Most importantly, visualization guides and inspires us to see new questions to ask rather than merely repeat old answers:
This elusiveness is not so much a particularity of perception as it is characteristic of cognition in general. The privilege of observing everything in relation raises understanding to higher levels of complexity and validity, but it exposes the observer at the same time to the infinity of possible connections. It charges him with the task of distinguishing the pertinent relations from the impertinent ones and of warily watching the effects things have upon each other. [Arnheim R. 1970 p.62]

11 [a] To put the argument somewhat more technically:
Visual displays of information encourage a diversity of individual viewer styles and rates of editing, personalizing, reasoning, and understanding. Unlike speech, visual displays are simultaneously a wideband and a perceiver-controllable channel. [Tufte E.R. 1990 p.31]

[b] Why does our visual system work so well?:
Human visual perception is performed by the most complex structure of the known universe, the visual cortex, that contains at least 1010 neurons, where each neuron in average contains 104 synapses (gates). This enigmatic processing network can perform prodigious feats when properly coupled to the visual stimuli. [Papathomas T.V. & Julesz B. 1988 p.355]

[c] And how does it operate so quickly?:
Humans can recognize unexpected objects in around 100 neuron-firing times. [Plantinga W.H. 1988 p.56]

[d] Our vision has evolved over a long time to become this powerful:
Average human beings can be beaten at arithmetic by a one operation per second machine, in logic problems by 100 operations per second, at chess by 10,000 operations per second, in some narrow "expert systems" areas by a million operations. Robotic performances can not yet provide this same standard of comparison, but a calculation based on retinal processes and their computer visual equivalents suggests that a billion (109) operations per second are required to do the job of the retina, and 10 trillion (1013) to match the bulk of the human brain.
Truly expert human performance may depend on mapping a problem into structures originally constructed for perceptual and motor tasks — so it can be internally visualized, felt, heard or perhaps smelled and tasted. Such transformations give the trillion-operation-per-second engine a purchase on the problem. The same perceptual-motor structures may also be the seat of "common sense," since they probably contain a powerful model of the world — developed to solve the merciless life and death problems of rapidly jumping to the right conclusion from the slightest sensory clues. [Moravec H. 1989 p.177]

12 [a] But we may not realise that we have never been taught how to see:
The lack of visual training in the sciences and technology on the one hand and the artist's neglect of, or even contempt for, the beautiful and vital task of making the world of facts visible to the enquiring mind, strikes me, by the way, as a much more serious ailment of our civilization than the "cultural divide" to which C.P. Snow drew so much public attention some time ago. He complained that scientists do not read good literature and writers know nothing about science. Perhaps this is so, but the complaint is superficial. It would seem that a person is "well rounded" not simply when he has a bit of everything but when he applies to everything he does the integrated whole of all his mental powers. [Arnheim R. 1970 p.307]

[b] Visual skills can, however, be enhanced:
Researchers increasingly are becoming aware that people need to be educated graphically in order for them to comprehend often increasingly complex graphics. It frequently has been suggested that graphicy is one skill that is generally not sufficiently developed thoughout our educational system as are numeracy, articulacy and literacy (Balchin, 1976). This research incorporated the concept of learning effects in order to judge its impact. [Halliday S.M. 1987 p.63]

[c] Researchers may need to learn to use graphics more:
Having established this high-bandwith communication link from the computer's vast computation power to the human brain, we are ready to look at ways of translating scientific data into pictures. We also need to educate scientists to the use of computer graphics. I have known many scientists who did not believe that mere pictures could help them understand their research. So they continued to burn up hours of supercomputer time (with over a hundred million calculations per second) and assumed that they had absorbed the complete result by studying output numbers. But once scientists begin to use computer graphics, they wonder how they ever got along without them. They find those "mere pictures" not only give them a firmer understanding of problems and provide a means of more easily explaining their work to colleagues but quite often open up whole new areas of research through observation of some subtle feature in an image. [Prueitt M.L. 1987 p.4]

[d] The advantages, once visualization is accepted as a method, are numerous:
Visualization is often opportunist; that is, an interpreter will not always have a good idea of exactly which attributes are of principal interest and indeed may often specify conflicting aims. It may also be advantageous to generate initial representations for large quantities of data automatically and quite independently of analyst interaction. [Robertson P.K. 1990 p.121]

13 [a] The subject matter of the earliest maps is interesting:
Chinese literature tells us that maps were being used in the East as early as the 7th Century BC, while the earliest surviving examples of maps are clay tablets found at Nuzi, in northern Iraq. Believed to be from the period circa 2,300 BC, they show rivers, settlements, land-holdings and hills. [Brannon G. 1989 p.38]

[b] Other forms of graphical display of information are much younger than cartography:
Maps have been used for more than 5,000 years whereas most other forms of graphic information date from the eighteenth century — graphs are a surprisingly modern discovery (Tufte, 1983). The earliest use of pictures is, of course, long before the first map, but perhaps we should exclude pictures from our definition of graphic information: pictures do not share the geometrical or conceptual structure of maps and graphs. [Phillips R.J. 1989 p.24]

[c] It is the increased availability of information which necessitates new visual solutions:
The early problem of spatial organization grew with the amount of data to be analysed. Multiple measurements proliferated with the Industrial Revolution in Europe, which brought a spate of new measuring devices: the air and water thermometer (c. 1590), micrometer (1656), weather-clock (c. 1660), mercury thermometer (1714), etc. Spatial organization of multiple measurements was achieved in two competing forms, coordinate systems and tables, which dominated quantitative graphics in the 17th and 18th centuries ... [Beniger J.R. & Robyn D.L. 1978 p.2]

[d] The popularity of visualization has been cyclic:
In mathematics, it is considered the most flagrant gauchery to use a diagram. "Graphics" is thought to be an inflated title for "mechanical drawing". In fact, all the intrinsically visible subjects; geography, graphics, and geometry, are suspected of being really grade school subjects, fit only for brains that are still undergoing biological maturation and whose harmfully misleading approach will have to be undone later. [Bunge W. 1968 pp.31-32]

14 [a] The history of computer graphics is short, but eventful:
Computer graphics started with the display of data on hardcopy plotters and cathode ray tube (CRT) screens soon after the introduction of computers themselves. It has grown to include the creation, storage, and manipulation of models and images of objects. These models come from a diverse and expanding set of fields, and include physical, mathematical, engineering, architectural, and even conceptual (abstract) structures, natural phenomena, and so on. [Foley J.D., Dam A. van, Feiner S.K. & Hughes J.F. 1990 p.1]

[b] Many milestones mark the way:
The scientific visualization going on today, Rosebush shows us, has been going on for a long time. In 1964 Ed Zajak of Bell Labs, who was a programmer animator, did a satellite orbiting in space... [Neal M. 1988 p.9]

[c] The discipline is now reconstructing its history:
The concept of scientific visualization reaches back into prehistoric times when a caveman drew a map of his local environment on his cave wall. In antiquity, legend tells us that Archimedes was slain by a Roman soldier while visualizing figures sketched in the sand. In this century, chemists began to understand the structure of matter and satisfied the need to visualize molecules with wooden and plastic models. Visualizing data and concepts is not new, nor is it computer dependent.
In the computer age, we have progressed through line-printer output, contour plots, etc., to more sophisticated techniques. Yet scientific visualization has only emerged as a technology in the last two or three years. [Rosenblum L.J. 1990 p.209]

[d] And beginning to realise where the future lies:
Structure, however, has been left behind in the race to create more and more realistic images. While photo-realism is eye catching, it is not necessarily informative. One of the great potentials of computer graphics is to provide a vision of what we might not otherwise be able to see in a photograph or real life. [Dooley D. & Cohen M.F. 1990 p.307]

15 [a] The technical term visualization was sprung upon the scientific community in 1987:
Visualization is a method of computing. It transforms the symbolic into the geometric, enabling researchers to observe their simulations and computations. Visualization offers a method of seeing the unseen. It enriches the process of scientific discovery and fosters profound and unexpected insights. In many fields it is already revolutionizing the way scientists do science. [McCormick B.H. et al. 1987 p.3]

[b] Grand claims have been made of the philosophy:
Computer graphics and image processing are technologies. Visualization, a term used in the industry since the 1987 publication of the National Science Foundation report Visualization in Scientific Computing, represents much more than that. Visualization is a form of communication that transcends application and technological boundaries. [DeFanti T.A., Brown M.D. & McCormick B.H. 1989 p.12]

[c] Things have changed very quickly:
Images and animations are no longer merely illustrations in science and engineering — they have become part of the content of science and engineering and are influencing how scientists and engineers conduct their daily work. [Foley J.D., Dam A. van, Feiner S.K. & Hughes J.F. 1990 p.22]

[d] Recognition of this revolution is increasing:
Computing imaging is not new, but the term, "scientific visualization", is justified as an indicator of an important new phase of development and a novel alignment of several computational technologies. [Haber R.B. & McNabb D.A. 1990 p.74]

[e] Its value to geography was recognised ten years ago:
Visualization is important, if not essential, in human thought. Visual thinking is not exclusively an artistic talent, but is constantly used by everyone. It pervades all human activity, from abstract and theoretical to everyday and down-to-earth. Yet development of visual thinking has been immobilized by society and education. Even geographers, in spite of their historic association with maps and mapping, fare little better. They seem to have given up on maps at the very time that other disciplines were discovering the power of graphics and documenting the physiological basis of visualization. This is particularly ironical since the geographic map may well be the most highly developed of the various graphic media that have been conceived in response to the need for visualization. Geographers are fortunate to be so closely associated with such a powerful, sophisticated tool of thought (something practitioners of other disciplines point out repeatedly). Yet incredible as it seems, geographers have not taken anything near to full advantage of their traditional relationship with maps. [Muehrcke P. 1981 pp.37-38]

16 [a] We need to be careful in deciding what is forming the patterns we see:
The boundaries between shadings on a choroplethic map tend to dominate the visual impact of the representation, because sharp visual contrasts occur along these lines. Map-readers tend to assign significance to these boundaries and, as a result, often assume that they designate breaks in the configuration of the statistical surface. Since this seems to be the normal reaction among map-users, the map-maker is obliged to use generalizations in which there is a concurrence of boundaries and surface breaks. [Jenks G.F. & Caspall F.C. 1971 p.229]

[b] We have to decide how we want to visualize what we are studying:
Yet the problem of devising a standard set of eight shadings for the maps was most troublesome. There were three initial requirements: first, that the shadings should be smoothly graded; secondly, that the class status of any area should be readily identifiable on inspection; and thirdly, that the shading specification could be applied by draughtsmen working independently in different cartographic departments without too much loss of comparability. The first two requirements proved to be almost incompatible. Smooth grading could best be achieved by using either a graded series of dots, or a graded series of lines with a constant direction. Neither of these produced visually acceptable results and the recognition of class status proved to be extremely difficult with both systems. [Hunt A.J. 1968 p.7]

[c] The debate over the simple shading of choropleth [patch or block] maps continues:
Nowhere is this debate between technical constraint and effective communication clearer than in the discussion over continuous shading. Tobler (1973) pointed out that digital techniques could remove the need to establish a finite number of levels of shading in making a choropleth map, as it is technically possible to crosshatch each area with a density of lines directly proportional to the value of the mapped attribute for that area. Evans (1977) took the opposing position that while the need for a finite number of levels can certainly be regarded as the consequence of technical constraints in manual cartography, it also has a distinct and legitimate function in communication. [Goodchild M.F. 1988 p.313]

[d] Use of colour is not necessarily always an added advantage:
Observer performance experiments are conducted to study the merits of the proposed color methods. The results of the study show that observers performed better with a linearized gray scale than with the newly-developed LOCS [Linearly Optimised Colour Scale] at a statistically-significant level of confidence. They also show that observers performed better with the LOCS than with another colour scale (the heated-object scale), but at a non-significant level of confidence. [Levkowitz H. 1988 p.v]

17 [a] The general appearance:
When it comes to the more important "gestalt" effects of the various decisions all considered together in a final map presentation, our ignorance is staggering. Before map data, map elements (symbols), and map users can be functionally and most effectively integrated into the geographic information system, it will be necessary to satisfy the need for research in which the simultaneous interaction of all mapping variables is determined. [Muehrcke P. 1972 p.49]

[b] Our ignorance is being realised again in visualization:
In the excitement over the obvious benefits of scientific visualization, few questions have been asked about the nature of perceived information and how well the human visual system actually performs. Because visualization is a new, emerging discipline, the lack of structure is not surprising, but their development is necessary and offers significant research opportunities. [Rosenblum L.J. 1990 p.211]

[c] We know a little of the mechanics of vision:
Sharpness falls off so rapidly that at a deviation of ten degrees from the axis of fixation, where it is at a maximum, it is already reduced to one fifth. Because retinal sensitivity is so restricted, the eye can and must single out some particular spot, which becomes isolated, dominant, central. [Arnheim R. 1970 p.24]

[d] Some aspects can be mimicked on a computer screen , for example using an ...:
Optical fish-eye window. Information in the window is compressed like the image of a convex mirror. [Card S.K., Pavel M. and Farrell J.E. 1985 p.240]

[e] Other research has provided explanations for some of the mechanisms through which vision may operate:
Our conclusions about the medium follow the research findings, but they also make sense for other reasons: for one thing, the medium's limited spatial extent fits in neatly with how we might expect visual perception and imagery to have developed over the course of human evolution. That is, the medium presumably evolved to process information from the sense organs, which means that it only needs to be large enough to handle the arc subtended by the eyes. And because the eyes are spread apart horizontally — as is, presumably, the spatial medium they feed — they have a greater horizontal scope. [Kosslyn S.M. 1983 p.71]

[f] Yet only the most general understanding has been gained:
Vision is therefore, first and foremost, an information-processing task, but we cannot think of it just as a process. For if we are capable of knowing what is where in the world, our brains must somehow be capable of representing this information - in all its profusion of color and form, beauty, motion and detail. [Marr D. 1982 p.3]

18 [a] Colour is most useful, after position, to show information in our pictures:
A little reflection convinces us that in static displays nothing does even nearly as well as right-left position and up-down position for (a) producing impact, (b) facilitating synthesis of impressions from groups of points, and (c) making fine distinctions both possible and easy. Nothing is comparable, but what comes closest?
In order, from stronger to weaker, we feel that the prominent representatives are:
(1) colour (where the establishment of synthesis-prone sequences deserves attention, as emphasized to us by W.J. Dixon);
(2) shape (of characters or symbols);
(3) size (as in pseudo-perspective displays);
(4) contrast (e.g. more gray corresponds to more distant).
Colour deserves more attention than the others, especially in view of the hope for synthesis. [Tukey P.A. & Tukey J.W. 1981 p.193]

[b] The use of combinations of colour scales has been well studied:
All three studies [Olson 1981, Mersey 1980, Carstensen 1981], in addition to this research, indicate that bivariate mapping is a viable technique with numerous graphic design possibilities and considerable flexibility. The main criteria is that students have to be able to comprehend the mapping technique [Halliday S.M. 1987 p.69]

[c] A theory behind two-colour mapping developed simultaneously:
If the relationships, as far as geographic location was concerned, were essentially random, the resulting map would show no particular tendency toward an areal concentration of similar colors but, instead, would exhibit a patchwork of small contrasting color blocks throughout the country. [Bureau of the Census 1970]

[d] Experiments have been conducted to confirm many assumptions:
In Olson's (1981) concluding experiment, experiment IV, it was discovered that subjects found the spectrally-encoded two-variable maps to be aesthetically appealing. While the univariate maps were given a high rank for readability, the spectrally-encoded bivariate maps were seen as more innovative and interesting to work with. [Halliday S.M. 1987 p.15]

19 [a] The combination of three colour scales is contentious:
It is far more difficult to distinguish the amounts of the three primary colors painted simultaneously onto a point in space, but it is possible (barely possible) to do so. Therefore a crude, but effective, way exists for displaying three functions of three independent variables. [Staudhammer J. 1975 p.183]

[b] Most people are taught to distinguish the primary colours as yellow, red and blue:
In the psychological realm, color vision is based on a few pure, elementary qualities, by no means necessarily or simply related to the physiological types of receptor. Just as perceived shapes are more or less complex elaborations of simple shapes, so color patterns are seen as elaborations of the elementary, pure qualities of yellow, red, blue. Here and there, these qualities are encountered in their purity, but most of the time there are mixtures, which are understood perceptually as combinations of the underlying primaries. Some of these combinations are sufficiently precise in themselves to function as visual concepts in their own right, e.g., orange, green, or purple. In the system of colors, as we find it applied, for example, in painting, these secondary concepts serve as transitional links between the primaries, which are the fundamentals of the system. It is a heirarchic system, similar to that of traditional logic, in which a multitude of more particular concepts derives from a basic few, thereby creating an order, which defines the nature of each element through its place in the whole. [Arnheim R. 1970 pp.30-31]

[c] Some have used the default red, green and blue guns of the computer monitor:
Clearly, since a total of three dimensions are available in color space it is possible to construct and transform much more complex models. It is possible to express a trivariate distribution by mapping each variable onto one of the dimensions of color space. [Sibert J.L. 1980 p.214]

[d] Others have suggested the printer's screens of cyan, yellow and magenta:
The collection of maps does not answer the question "what is there at a given place?" But maps with the same scale can be superimposed three by three. It is sufficient to transcribe them on three different color films: cyan-blue, yellow, magenta-red. [Bertin J. 1981 p.163]

20 [a] We form a generalize image of a picture:
Generalization, if you wish to call it that, occurs spontaneously in all perception. Complex though a map may be, the mind derives from it a simplified pattern. [Arnheim R. 1976 p.9]

[b] Often it is better not to generalize in cartography:
The collection of comprehensive maps does not involve problems of generalization. Indeed, the eye immediately sees a shape, whatever its complexity. Each map can thus carry an impressive amount of data, as with the twenty-five million buildings in Poland on a scale of 1/2 M (F. UHORCZAK). But at the same time, the eye is free to focus on any level of an ordered or quantitative variable and is thus free to "generalize", that is to regionalize, as it pleases. [Bertin J. 1981 p.161-163]

[c] Some automated generalization can be useful:
The space smoothing techniques can be employed to investigate the existence of geographical patterns of property values. The generalization facilitates the recognition of patterns because it appears to be true, as Holloway (1958, p.386) suggests, that we do ".... high-pass filtering in our 'mind's eye'." [Tobler W.R. 1989 p.19]

21 [a] Cost was a problem in producing the colour prints for this dissertation:
Another practical issue in displaying the data is cost. Using color provides an important extra dimension in displaying the complex data sets obtained. Currently, however, the cost of publishing two-color plates in some scientific journals represents more than half the cost of the laboratory computer that controls the experiment, stores the data, and displays the results. Therefore, the use of color figures (which can best present the results) might be hard to justify. [Long M.B., Lyons K. & Lam J.K. 1990 p.138]

[b] Colour printing is still technically difficult as well:
Showing complexity is hard work. Detailed micro/macro designs are difficult to produce, imposing substantial costs for data collection, illustration, custom computing, image processing, production, and fine printing — expenses similar to that of first-class cartography (which, in the main, can be financed only by governments). [Tufte E.R. 1990 p.50]

[c] Using colour with computers can be frustrating:
It is nearly impossible to get the same colour on two different devices so do not produce systems that depend on that. [Hopgood F.R.A. 1991 p.9]

[d] However, we no longer require expensive computers:
A personal computer with an appropriate display system can be just about as effective as a larger system for our visualization techniques and interactive when outfitted with a suitable computation accelerator, such as the one we described. [Wolfe R.H. & Liu C.N. 1988 p.29]

[e] Cheaper computers are often more useful nowadays:
Workstations, minicomputers and image computers are significantly more powerful and effective visualization tools than supercomputers. It is a waste of supercomputer cycles to use them to convert model data into new pictures. Specialized graphics processors are more cost-effective than supercomputers for specialized picture processing and/or generation. Workstations should be placed on the desks of each and every researcher to give them immediate access to local graphics capabilities. [McCormick B.H. et al 1987 p.9]

[f] Many leading researchers predict the demise of monolithic machines:
McCormick says, "The visualization problem is not peculiar to supercomputers in any sense. In fact, I don't expect the bulk of [the visualization initiative] to be tied to them. Supercomputers are in many ways dinosaurs, a dying breed." [Frenkel K.A. 1988 p.113]

[g] A much cheaper future lies on the horizon:
Visualization of anything will become the norm in computer uses. Specialized data display managers will migrate to chip sets for both faster response and ease of user interaction. [Staudhammer J. 1991 pp.42-43]

22 [a] The next major hurdle to cross involves allowing people to paint images as fast as they can see them:
The unhappy thing about all this, of course, is that whereas I have the ability (and we all have the ability if we're sighted) to take images in at a fantastic rate, I have no ability to create images with the same facility. This is a one-way street. On the other hand, I can create language and symbols at about the same rate I can take them in, which means I can create speech at about the same pace that I can listen to it. So it is not at all unexpected that for most of us language seems to be the main carrier of our thoughts because that is the thing we can hear ourselves saying and were conscious of its use. [Huggins W. 1973 p.37]

[b] We have always, of course, been able to create images in our own minds:
Aristotle was one early commentator who gave a crucial role to mental images, claiming that "thought is impossible without an image" and "memory, even the memory of concepts, does not take place without an image." [Kosslyn S.M. 1983 p.5]

[c] And have for a long time sought ways of showing what we can see to others:
The coordinate approach grew out of the analytic geometry developed by Descartes, Fermat, and other French mathematicians in the first half of the 17th century. Descartes himself was convinced that "imagination or visualization, and in particular the use of diagrams, has a crucial part to play in scientific investigation" [103, p.28]. [Beniger J.R. & Robyn D.L. 1978 p.2]

[d] And now can use images to view social structure:
The general consensus in the scientific visualization field is that a broad commonality exists among the visual needs of all numerically intensive sciences. While users have applied this computational environment to fields as diverse as computational fluid dynamics, molecular modelling, geophysics, and meteorology, we are keenly awaiting its application to fields with a shorter history in numerical computing, such as econometrics and the social sciences. Will users from these fields find this environment appropriate for their needs? [Upson C., Faulhaber T., Kamins D., Laidlaw D., Schlegel D., Vroom J., Gurwitz R. & Dam A. van 1989 p.41]

23 [a] We have to be able to visualize spatial social structure if we are to begin to comprehend it:
The observed variation pattern in spatial data is often extremely complicated. Geographical maps have traditionally served the need for visualization in attempts at description and explanation of spatial patterns. A logical goal of the information display (mapping) process is to produce, as efficiently as possible, the most effective graphic communicator of distributional information. Image processing entails the assimilation, manipulation, and analysis of information which is given in spatial pattern form such as a map. [Muehrcke P. 1972 p.53]

[b] To see the structure we must first decide how to draw it:
Recently we have witnessed a heightened interest on the part of specialists of various branches of science, in investigations connected with the problem of the graphic representation of social, economic, and other spaces. Various methods of image transformation under the graphic constructs of anamorphozy [transformed images] [3,4,7,9,12] are used in these cases. Such images, in our opinion, are promising in city-building analysis, especially during the use of "time" scales for them. [Tikunov V.S. & Yudin S.A. 1987 p.203]

[c] And we must use machines if we are to cope with the mass of information that has been collected:
In the GIS environment, visualisation techniques are recognised as an invaluable system component, aiding in the interpretation of spatially related phenomena and complex data analysis that takes the GIS a step beyond two dimensional polygonal overlay analyses. Many of the GIS vendors are including this capability in their systems to help cope in our understanding of the "fire hose" of data being produced by contemporary sources such as satellites. [Kennie T.J.M. & McLaren R.A. 1988 p.737]

[d] We must look outside of our own disciplines for inspiration:
Perhaps then, the differences between maps and other forms of graphic information are not as great as they appear. All types of graphic information are different solutions to a common problem: our limited capacity to remember unprocessed information. By removing the limitations of short term memory, graphic information allows us to do kinds of thinking which are difficult or impossible in other ways. [Phillips R.J. 1989 p.25]

[e] And recognise what has already been achieved:
That very ancient merger of Geography, Geometry and Graphics still exists and, if anything, with increasing vitality. Many breakthroughs still lie ahead. The map is the geographer's laboratory. [Warntz W.W. 1973 p.85]