To understand map design, and how maps work, it is useful to see how map design concepts play out on a real map.

One of the significant updates to the 2nd edition of Making Maps was the inclusion of a map of the 1986 trans-global flight of the experimental aircraft called Voyager. This map, originally designed and created by David DiBiase and I back in 1987 for David Woodward’s map design course (and in the University of Wisconsin Cartographic Lab), is repeated thirteen times in seven of the chapters, annotated to show how the concepts and ideas in these chapters play out on the map.

The repeatedly annotated Voyager map serves as an example of map design in practice, but also a guide to “reading” a map from the perspective of map design.

Martin Dodge suggested the annotated maps be available together, for instructional purposes. A good idea! So here they are. Each individual map on this page is a 800k PNG file (click for full size).

A PDF file (8.8mb) with all thirteen higher resolution TIFF images is available here.

The Voyager map project was quite a bit of fun to create back in the day. It won a map design award, was published in the book about the Voyager flight, and printed on paper as part of a promotion for the Waukesha County, Wisconsin Airport. I cannot imagine anything much more exciting than that.

••••••••••

The second map in the series prefaces the initial chapter in Making Maps 2nd edition, and poses a series of questions that will be addressed in future chapters (and annotated Voyager maps):

Chapter 2, What’s Your Map For?, sets the context of the map (and of its re-creation for the book) and how such context shapes the design of the map:

Chapter 6, The Big Picture of Map Design, repeats the Voyager map five times, focusing on the key map design concepts covered in the chapter. This is where the map flips orientation south up (a controversial design choice, according to some reviewers: good! Think about why such a choice is controversial, and if it should be [or ask your students to do so]):

South is now up (so the story reads from left to right):

Annotations about map pieces including title, scale, explanatory text, legend, directional indicator, border, sources, credits, and insets & locator maps:

Annotations about visual arrangement including path, visual center, balance, symmetry, sight-lines, and grids:

Annotations about graphical excellence, based on Edward Tufte’s ideas, including complexity, detail, design variation & data variation, context, revision, non-data ink, data-ink ratio, explanatory text, editing, chartjunk & map crap, redundancy, and multivariate data:

Chapter 7, The Inner Workings of Map Design, reveals the Voyager map with no visual differences (a confusing mess of lines and type):

The map with visual differences is then annotated, with regard to key methods for establishing visual differences, including detail, edges, texture, layering, shape, size, closure, proximity, simplicity, direction, familiarity, and color:

Chapter 8, Map Generalization and Classification, annotates the Voyager map in terms of the generalization concepts of selection, dimension change, simplification, smoothing, displacement, and enhancement:

Chapter 9, Map Symbolization, annotates the map in terms of the visual variables: shape, size, color hue, color value, color intensity, and texture:

Chapter 10, Words on Maps, annotates the Voyager map with regards to typographic variables including typeface, type form, type weight, and type size:

I have a few ideas for additional annotated Voyager maps (such as a full color map) which I hope to cobble together in the future. If you have any other ideas for variations that might be useful or interesting, let me know.

Cover, Making Maps, 2nd edition (Amazon | Guilford)

Krygier and Wood’s book should be used by anyone interested in the way the world looks, the way the world works, or the way the world should be. It remains the most accessible yet comprehensive guide of its kind. The second edition meets the needs and expectations of the “Google generation” of map users while remaining true to the guiding principles that govern how maps look, work, and function. The very accessible, extensively illustrated format makes the book easily usable by students at all levels, as well as those taking steps to develop expertise in cartographic design. Paul Longley, Department of Geography, University College London, United Kingdom.

Building on their solid first edition, Krygier and Wood have created a new and much richer follow-up. The second edition represents a serious reworking of subject matter and graphics. The book uses extraordinary map exemplars to address the full range of basic cartographic concepts and to demonstrate many subtle and advanced design techniques as well. Making Maps is appropriate for beginning to intermediate college cartography students and others who want to tap into the power of map creation. Addressing current social issues including map agendas, ethics, and democracy, it is the kind of book that will inspire readers and cultivate admiration for the field. James E. Meacham, Senior Research Associate and InfoGraphics Lab Director, Department of Geography, University of Oregon.

More than two years in the making, the second edition of the book Making Maps is set for printing. Copies should be available in February or March of 2011. A Korean translation (?!) is planned for 2012.

This is no weenie update: Denis and I ruthlessly reorganized and rethought every bit of content in the book. I then redesigned the entire book and spent the better part of eight months producing it. We both think it’s a much better book.

Denis and I were careful to keep the spirit of the first edition of Making Maps intact while sharpening the overall look, content, and usability of the book. The goal from the beginning was to create a map design text that was different from other map design texts – more visual, creative, critical, engaging, and focused on making maps as well as really understanding how they work. It is a synthesis of what we like most about the academic study of maps and the actual design and production of maps. It is difficult to express how complex and challenging achieving this goal has been. When I look at this new edition, it feels so easy. Why couldn’t we have just done this 8 years ago when I started on the initial edition of the book?

The 2nd edition is larger in size (now 7″ x 10″) allowing more content on each page. In a Tuftean fit of non-data-ink removal, gone are a bunch of pages that didn’t have much content (such as the overview pages near the beginning of each chapter). We did retain ample white space, since absence makes the heart fonder.

We also added new material, including many real mapped examples, yet we are dozens of pages shorter than the first edition. Our goal was a lean book – “the greatest number of ideas in the shortest time with the least ink in the smallest space” – as Tufte put it.

The cover initiates an expanded version of the “road connector controversy” which sets up the point of the book – you make things happen by making maps.

There is a completely new first chapter setting the context for the entire book. It introduces The Flight of Voyager map, which is annotated a dozen times over throughout the book showing how map design concepts in the text play out on an actual map:

The chapters in the book are about the same, with a new first chapter and some recast chapter names:

Introduction

1: How to Make a Map
2: What’s Your Map For?
3: Mappable Data
4: Map Making Tools
5: Geographic Framework
6: The Big Picture of Map Design
7: The Inner Workings of Map Design
8: Map Generalization and Classification
9: Map Symbolization
10: Words on Maps
11: Color on Maps

While some chapters retain a significant amount of the original edition’s material, chapters 6 and 7 were extensively revised.

A makingmaps.net blog posting “How Useful is Tufte for Making Maps?” led me to incorporate Tufte’s ideas in the book in a much more explicit manner than in the 1st edition. See, for example, the Tufte-influenced annotated Flight of Voyager map (2 page spread, chapter 6) below:

Chapter 7 was revised as “The Inner Workings of Map Design” including figure ground:

Chapter 9 on map symbols also underwent significant renovations:

•••••••••••••••

Making Making Maps … Second Edition

I am but slightly embarrassed to admit that, once again, I produced the entire book in a 6-year-old version of the now defunct Freehand MX software. My original plan was to shift to InDesign since I was redesigning the entire book, but in the end I just wanted to make the damn book rather than futzing with transferring the maps and graphics from Freehand to InDesign and learning the ins and outs of InDesign. So my plan is to eventually shift the entire book to InDesign assuming a 3rd edition sometime in the future.

The book was produced on my 4-year-old MacBook Pro, which allowed me to work on it at home on the dining room table, at home on the table on our front porch (where Denis and I had earlier sat and pounded through the plan for the 2nd edition), at CupOJoe coffee at the end of the block, at Panera while waiting to pick up Annabelle after her morning pre-school, at soccer practice at some god-forsaken indoor soccer warehouse in the hellish outer suburbs of Columbus, in Raleigh NC whilst visiting Denis to work on the book, at the OSU recreation center with the climbing wall, at the OSU recreation center with the pool (both while waiting for kids to finish various climbey or splashy activities), at my parents house in Waukesha (Wisconsin), the Caribou Coffee in Waukesha, my in-laws in River Hills Wisconsin, and in my office at Ohio Wesleyan.

•••••••••••••••

This is really a labor of love – given the time and brain power expended on the text – and we both hope this new edition lives up to the expectations of the kind and usually enthusiastic readers of the first edition.

Map of New York City, Showing the Distribution of the Principal Nationalities by Sanitary Districts published in Harper’s Weekly (June 1, 1894) using 1890 U.S. Census data.

This map looks great, revealing a substantial amount of information with its intense, juxtaposed patterns.

The textures on the map show the relative amounts of different nationalities (qualitative data) in each of the areas (sanitary districts) on the map:

The map shows if a district has more or less diversity (more or fewer lines of different textures), the relative proportions of different nationalities, the nationalities themselves, and, at a broader scale, the districts that are similar or differ in their nationality constitution. Because of the careful rotation of the lines of textures, the different sanitary districts can also be distinguished from each other.

To quote the text which accompanies the maps (and explains the methodology of the map):

The census of 1890 obtained the nationality of the residents of each sanitary district by descent from the mother. The table in which this appears was made the basis of the nationality map. As a basis it will appear fair enough when it is considered that at the time of the census over seventy-six percent of the white population in the city had foreign-born mothers, and over forty per cent. were foreign-born themselves. So the latter certainly, and probably a majority of the thirty-six percent. of native-born of foreign mothers, would show the traits of their maternal nationality. All the nationalities given in the table are not plotted. The Scotch, English, Welsh, Scandinavian, and Canadians have not collected in colonies, but are scattered over the city. These, being in small numbers, and perhaps less foreign than the others, were disregarded. They appear in the unclassified [category] in the diagram at the foot of the map. Of the nationalities represented only those making up two-thirds of the population of any district have been plotted. This rule was adopted to bring out in clearer contrast those that do exist to a greater extent. The nationalities are represented by bands conventionally marked. The breadth of a band in any district bears the same relation to the sum of the breadths of the different bands in that district as the number of the nationality it represents bears to the two-thirds of the population in that district. Sanitary district S of the Twelfth Ward and the Twenty-third Ward are not touched. These were left blank because the method of representing nationalities gives an erroneous idea in regard to the density of population. These thinly populated districts where natives preponderate slightly would appear as native settlements. These, of course, they are, but not like other parts of the city, for they are suburban.

A map of population density (shown below) accompanies the Nationalities map, using increasingly dense textures to represent quantitative population differences:

The Nationalities map illustrates (in 1894) Edward Tufte’s demand for maps (and other information graphics) that reveal the multivariate and complex. This is a map to spend time with, not because it is poorly designed, but because it contains a substantial amount of information.

The New York City Principal Nationalities Map in its entirety (1890 data; rotated to fit your screen):

The New York City Population Density Map (1890 data):

The map pair, side by side for comparison (as originally published):

The map was created by Frederick E. Pierce for the Tenement-House Committee, one of the progressive organizations working against urban slums and blight in the late 19th and early 20th centuries. The text notes that the original map was in color, but was redrawn for publication in monochrome.

•••••

Texture is usually included as one of the cartographic visual variables. The visual variables, which provide a guide for matching visual marks to data characteristics (such as qualitative vs. quantitative data), are attributed to French semiotician Jacques Bertin.

I discuss and illustrate, with contrasting good/poor maps, the basic visual variables (shape, size, color hue, color value, color intensity, and texture) in chapter 9 (excerpt here) of Making Maps.

Texture can be difficult to work with as it can imply either qualitative or quantitative differences, and has a tendency to vibrate.

Back in the day, Zip-a-Tone (a brand of screentone) was used to create texture and gray-scale patterns on maps (and comics). Printed on a thin, clear material with adhesive backing, many a cartographer spent hours searching through sheets of Zip-a-Tone, cutting it to size, and adhering it to different areas on the map. Now diverse pattern fills are available in GIS and graphic design software (Illustrator, Corel).

The problem is that while generic textures – Zip-a-Tone or digital – could be used as “area fills” for a contemporary map of nationalities (like the 1890 Nationalities Map), one would still have to engage in the painstaking task of calculating ratios and creating appropriate areas for the different textures: the same task that engaged Mr. Pierce more than 100 years ago.

When I come across maps such as the 1890 Nationalities Map it reminds me that, despite the millions of maps generated by sophisticated software every day, few are information rich, complex, and multivariate – Tuftian. At least in this context – the high art and science of multivariate, data-intense mapping – we may not be much better off than we were in 1890.

Making maps is rife with rules. But following rules does not necessarily produce a great (or even good) map. It may be the implementation of broader design principles that leads to a successful map.

Principles are an intellectual generalization of a broad field of knowledge: a kind of map, in the broadest sense of the word.

They are useful for guiding map makers and helping map users understand how maps work.

There are numerous sets of cartographic design principles. My previous post on Edward Tufte distilled six map design principles (or commandments as I called them) from Tufte’s first book, The Visual Display of Quantitative Information.

In 1999 the British Cartographic Society’s Design Group proposed “Five Principles of Cartographic Design.” When I first came across this set of principles I thought them interesting – even a bit passionate – a rare state of affairs in the often stoic world of cartography. I added a few maps and my own comments (in italics).

More on these map design principles below: Concept before Compilation, Hierarchy with Harmony, Simplicity from Sacrifice, Maximum Information at Minimum Cost, and Engage the Emotion to Engage the Mind.

Cool maps below include: Geo-Smiley Terror Spree Map, The Continents and Islands of Mankind, Hate Groups and Hate Crimes Map, and Where Commuters Run Over Black Children, Detroit 1968.

Five Principles of Map Design

Concept before Compilation
Without a grasp of concept, the whole of the design process is negated. The parts embarrass the whole. Once concept is understood, no design or content feature will be included which does not fit it. Design the whole before the part. Design comes in two stages, concept and parameters, and detail in execution. Design once, devise, design again. User first, user last. What does the user want from this map? What can the user get from this map? Is that what they want? If a map were a building, it shouldn’t fall over.

…or, why are you making your map, who is the audience, and what do they want from the map?

•••••

Hierarchy with Harmony
Important things must look important, and the most important thing should look the most important. “They also serve who only stand and wait.” Lesser things have their place and should serve to complement the important. From the whole to the part, and all the parts, contributing to the whole. Associated items must have associated treatment. Harmony is to do with the whole map being happy with itself. Successful harmony leads to repose. Perfect harmony of elements leads to a neutral bloom. Harmony is subliminal.

…or, what’s important? Make it visually jump out. What’s less important, but necessary in a supporting role? Make it fall back…

(Geo-Smiley Terror Spree Map. Reproduced from Making Maps, p. 144)

•••••

Simplicity from Sacrifice
Great design tends towards simplicity (Bertin). Its not what you put in that makes a great map but what you take out. The map design stage is complete when you can take nothing else out. Running the film of an explosion backwards, all possibilities rush to one point. They become the right point. This is the designer’s skill. Content may determine scale or scale may determine content, and each determines the level of generalization (sacrifice).

…or, less is more…

(Redrawn from William Bunge, The Continents and Islands of Mankind. Areas in black have more than 30 people per square mile. Reproduced from Making Maps, p. 160-161)

•••••

Maximum Information at Minimum Cost (after Ziff)
How much information can be gained from this map, at a glance. Functionality not utility. Design makes utility functional. All designs are a compromise, just as a new born baby is a compromise between its father and mother. The spark which makes a map special often only comes when the map is complete.

…or, carefully select the content and marks on the map (symbols) to maximize the map’s information content and communication capabilities…

(Hate Groups and Hate Crimes Map. Apparently more hate groups in a state means fewer hate crimes. Reproduced from Making Maps, p. 208)

•••••

Engage the Emotion to Engage the Understanding
Design with emotion to engage the emotion. Only by feeling what the user feels can we see what the user sees. Good designers use Cartographic fictions, Cartographic impressions, Cartographic illusions to make a map. All of these have emotive contents. The image is the message. Good design is a result of the tension between the environment (the facts) and the designer. Only when the reader engages the emotion, the desire, will they be receptive to the map’s message. Design uses aesthetics but the principles of aesthetics are not those of design. We are not just prettying maps up. The philosophy is simple, beauty (aesthetics) focuses the attention. Focusing the attention is the purpose of map design!

…or, embed a bit of passion…

(Where Commuters Run Over Black Children, Detroit 1968. Detroit Geographical Expedition. The title says it all.)

What if there was a gap between mapped data and our perception of it?

Buried in the ArcGIS symbolization options for proportional symbol maps is a puzzling check box labeled Appearance Compensation (Flannery) that addresses one gap between perception and data symbolized on maps.

This check box is a vestige of academic cartography’s extensive engagement with psychophysics beginning in the 1950s. Psychophysics relates “matter to the mind, by describing the relationship between the world and the way it is perceived.” Psychophysical studies select specific sensory stimuli and evaluate human perception of the stimuli. Cartographers studied thresholds (what is the smallest type size the average viewer can read?), discrimination (what is the minimum difference between two gray tones required for the average viewer to perceive a difference?), and scaling (how to scale a map symbol so the average user correctly judges the symbol’s value?).

The most studied map symbol was the proportionally scaled circle.

The scaling of proportional map symbols was the primary research focus of James Flannery, a student of Arthur Robinson, one of the founders of American academic cartography. Flannery’s research on map symbols was based on 1920s research on the human perception of circles and other symbols on statistical graphics. Flannery’s dissertation, completed in 1956, was among the first in Geography using psychophysical methodology.

A typical proportional circle map (below, right) scales circle area to the value (usually a total) for a geographic area on a map. Such a map is an alternative to the more common choropleth map (below, left) that shades geographic areas to their value (usually derived data, such as a density).

While not as common as the choropleth map, proportional symbol maps can be found in the public eye, as on the map accompanying a recent New York Times article on poppy cultivation in Afghanistan.

The absolute scaling of circles is common on proportional circle maps: a county with a value of 100 has a circle with an area of 2 square cm, a county with a value of 200 has a circle with an area of 4 square cm, etc. (In 1801 William Playfair first scaled circle areas to represent quantities; the use of area rather than diameter persists to this day).

But psychophysical research revealed that while people tend to correctly estimate lengths, they tend to underestimate areas and volumes. In other words, when asked to pick a circle that is two times the size of another in a range of different circle sizes, most people would pick a circle that was about 1.8 times the size. This tendency gets worse with larger areas, and is worse in general for estimations of volumes.

The graph below shows three apparent-magnitude curves for estimations of symbol length, area, and volume.

The implication of this perceptual underestimation was that absolute scaling on proportional symbol maps led to inaccurate perception of the values: while the circles were scaled accurately, the perception of the areas, and thus the values the user got from the map, were wrong.

The solution was to devise a method of perceptual (or apparent) scaling of graduated symbols, or appearance compensation in ArcGIS-ese.

Thousands of perceptual tests led Flannery to develop a method for scaling circles that compensated for the underestimation. When you check the Appearance Compensation check box in ArcGIS the method scales up proportional circles, the larger the circle, the more the scaling.

There are problems with perceptual scaling that suggest avoiding its use on map symbols.

Edward Tufte, in The Visual Display of Quantitative Information (1998, 2nd ed. 2001) stands opposed to anything but absolute scaling: “The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the numerical quantities represented” (see my previous post on Tufte here). If one actually measures a perceptually scaled circle on a map (or graph) one would get the wrong actual value. Tufte’s demand to “tell the truth about data” excludes compensation for human perceptual failings. In one fell swoop Tufte wiped out a broad swath of psychophysical research in cartography.

The fact that perceptual (or apparent) scaling was based on an average subject ignored the fact that a broad range of user reactions varied greatly from the average. Perceptual adjustments for the average subject didn’t solve the perceptual problem for a considerable number of potential map users, and led to problems for those who were able to correctly judge areas.

Finally, once added to the complexity of a map, the value of apparent scaling diminished. Research revealed the perceptual problem was not as evident on maps with a smaller range of circle sizes; that good legend design could eliminate the perceptual problem; and that circles on complex maps may have other problems, including optical illusions, which are impossible to correct and lead to even worse perceptual problems. For example, the middle circle in the two groups below is the same size (showing how the perceived area of a circle is shaped by the circles that surround it):

Some vestiges of psychophysical studies are embedded in map design guidelines like those illustrated in the Making Maps book. However, psychophysical map studies assumed map-readers, maps, and map symbols were quite a bit simpler than they are in reality. This limited the utility of such research, much of which was seen by practicing cartographers as not useful for making maps or, at best, merely confirming common sense rules of map design.

A great overview of the history of psychological research on maps including psychophysical research is Daniel Montello’s “Cognitive Map-Design Research in the Twentieth Century: Theoretical and Empirical Approaches” (Cartography and Geographic Information Science, Vol. 29, No. 3, 2002, pp. 283-304). Borden Dent’s text Cartography: Thematic Map Design (4th ed., WC Brown, 1996) has a good discussion of absolute and perceptual scaling of map symbols, including equations for absolute and perceptual scaling, in chapter 8 (this chapter was the source for some of the redrawn illustrations above).

Postscript: The implementation of Flannery’s perceptual scaling in ArcGIS has a software-specific problem: perceptual scaling should only be applied to circles. If the symbol is changed in ArcGIS, to a square or some other shape, the perceptual scaling can still be applied. Research has shown that square symbols, on a proportional symbol map, don’t have the same perceptual problem that circles do (we can correctly estimate the areas of squares). Little research on map symbols besides circles and squares exists. Bottom line: if you use the Appearance Compensation in ArcGIS, only apply it to proportionally scaled circles.

Edward Tufte’s The Visual Display of Quantitative Information (1998, 2nd edition 2001) is a classic book, arguably his best, and certainly a key text in the field of information graphics (which encompasses cartography). I know some cartography courses use the book as a text.

I recall being inspired by the book as a neophyte cartographer back in the late 1990s.

The book looked great: its design communicated the importance of design (when most other cartography and information graphics books were clunky and poorly designed). The tone was serious and high-minded: I was designing information graphics. And I think I absorbed Tufte’s minimalist design philosophy, although cartographic design, at least the way I learned it, was largely minimalist, with no allowance for flourish, fake 3D embellishment, or other chartjunk (or “map-crap” as I call it in the Making Maps book).

While I won’t impugn the importance of lofty inspiration, I did wonder what kind of practical guidelines I could derive from Tufte’s book. You know, specific stuff that would help me to design and make better maps. I sat down one day and made a list of Tufteisms from the book: that list is below.

20 Tufteisms from The Visual Display of Quantitative Information

1. Graphical excellence is the well-designed presentation of interesting data – a matter of substance, of statistics, and of design.
2. Graphical excellence consists of complex ideas communicated with clarity, precision, and efficiency.
3. Graphical excellence is that which gives to the viewer the greatest number of ideas in the shortest time with the least ink in the smallest space.
4. Graphical excellence is nearly always multivariate.
5. Graphical excellence requires telling the truth about the data.
6. The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the numerical quantities represented.
7. Clear, detailed, and thorough labeling should be used to defeat graphical distortion and ambiguity.
8. Write out explanations of the data on the graphic itself. Label important events in the data.
9. Show data variation, not design variation.
10. In time-series displays of money, deflated and standardized units of monetary measurement are nearly always better than nominal units.
11. The number of information-carrying (variable) dimensions depicted should not exceed the number of dimensions in the data.
12. Graphics must not quote data out of context.
13. Above all else, show the data.
14. Maximize the data-ink ratio.
15. Erase non-data-ink.
16. Erase redundant data-ink.
17. Revise and edit.
18. Forgo chartjunk
19. If the nature of the data suggests the shape of the graphic, follow that suggestion. Otherwise, move toward horizontal graphics about 50 percent wider than tall.
20. The revelation of the complex.

I recall being somewhat underwhelmed at my practical how-to list. But I pondered how they related to my map making and cartography in general, and reduced them into fewer categories, the Six Commandments.

Commandment 1: Map Substantial Information, including (1), (2), (3), (4), and (20). Unfortunately, such choices were not up to me. For example, I might be told to make a map that showed the location of a dozen cities and a study area in Bolivia, and there was no way to make that interesting, multivariate, or complex. Bottom line, many maps are of boring data, chosen and assigned by someone else, and there is not much the map maker can do about it. But these five related Tufteisms did make me understand the potential for maps of non-boring data… and maybe if the clients, the people who decided what data to map, read Tufte they would do a better job at selecting interesting, multivariate, complex data to map.

Commandment 2: Don’t Lie with Maps, including (5), (6), (9), (10), (12), and (13). This evokes two classics, Huff’s How to Lie with Statistics (1954) and Monmonier’s How to Lie with Maps (1991, 2nd edition 1996). The idea of “lies” resonates deeply with some people, but it is a complicated issue. In the Making Maps book I mapped out poverty rates by county in the US, classified with different classification schemes (unclassified, quantiles, equal interval, natural breaks, and a unique scheme). This produces, of course, five quite different looking maps that, as no classification scheme is illegal or inherently “lies,” are all potentially viable. Maps are always made with a purpose, and purpose will drive the choice of classification scheme. Each scheme has advantages and disadvantages and each obscures and emphasizes different aspects of the data. Beyond the falsification of data, I find the concepts of “lies” and “truth” a bit too simplistic: it is a lot tougher than that. It’s about being smart and critical and understanding the inherent trade-offs among diverse map design options.

Commandment 3: Effectively Label Maps, including (7) & (8). This was not a revelation to me, as cartographers spent a serious amount of effort working out guidelines for the effective labeling of maps. Tufte did encourage me to include explanatory text on maps: tell people what you believe the map is showing and why it’s important. You cannot communicate everything with single words and non-text map symbols, and the map reader will see where you are coming from (and that will help them be critical of your map design choices and what you are communicating with the map).

Commandment 4: Minimize Map Crap, includes (11), (14), (15), (16), and (18). Map-crap, chart-junk, all the same: graphic dross that encrusts a map. Big, honkin’ north arrows (you don’t even need a north arrow if the orientation is obvious), fancy borders, fake 3-D effects, etc. More often map crap is the result of poor design choices for elements that need to be on the map. For example, a graticule consisting of black lines: the graticule may have to be on the map, but it should not stand out. Light gray lines, or even white lines (reversed out of a gray – as with the map that heads this posting – or colored background) would be preferable. Just like Tufte’s graph reduction exercise on pp. 126-7 in my 1st edition of The Visual Display of Quantitative Information. This minimalist design approach is stern and serious, and may be inappropriate when designing maps for advertising, promotional purposes, or fun (does anyone do that?). Chartjunk and map-crap have a place.

Commandment 5: Map Layout Matters, includes (19). Layout is a bigger issue than this one point from Tufte and it is an issue that is not stressed in map design texts (although I do devote a decent chunk of a chapter in Making Maps to map layout). It is difficult to talk about map layout (or the layout of information graphics in general) in the abstract. But layout strongly effects the look and feel of the map, and can make a map easy or difficult to read and interpret.

Commandment 6: Evaluate your Map, includes (17). Evaluation is really important and, for whatever odd reason, is typically not part of information graphics or cartography texts. There are different kinds of evaluation, from documentation of your design and production process, to formative evaluation (where you or others critique and revise your map as it is produced), and, finally, impact evaluation where formal methods are used to assess the effectiveness of the map among a subset of its intended audience. I devote parts of two chapters in Making Maps to these different kinds of evaluation.

The 20 Tufteisms and the Six Commandments are superficially less than I expected, from the perspective of practical guidelines. But, upon reflection, they do touch on many of the fundamental issues that determine if a map design is going to work or not, and what could more practical than that?

Maybe someday I will review Tufte’s subsequent books: Envisioning Information (1990), Visual Explanations (1997), and Beautiful Evidence (2006).

Lets look at a site that has been around awhile: The Modern Language Association’s Language Map. The site allows you map language data collected in the 2000 U.S. Census. A nice focused site with interesting data (I use it in my classes and the students enjoy pondering the patterns): here is the default map of the total number of language speakers in each county:

The basic language map allows you to view 33 different languages, mapped by county in the U.S. The total number of people who speak a particular language (above) can be mapped, but mapping totals can be deceptive, as the sizes of the counties vary. Thus a county may have more speakers of a particular language just because it covers more area than a smaller county. To account for these variations in county size, map the data as a percentage (the percent of people in a county that speak a particular language, see below). But you can map totals and there are sometimes good reasons to do so. Just realize the potential limitations of what you are seeing.

The basic map also allows you to map the data by state or zip code, and to add additional information (cities, roads, county & state names, etc.). The basic map does not allow you to change the classification scheme, and the arbitrary classification schemes the map uses are weird (but probably chosen so different maps can be compared to each other). Different classification schemes will produce different patterns, so take the patterns produced on these maps with a grain of salt: they will change if you (could) change the classification scheme. It would be also good to flip the legend, putting higher numbers (high =more) at the top of the legend, and lower numbers (low=less) at the bottom.

Here is the % by county map of the same data:

The symbolization of the county (or state or zip) boundaries on the map is annoyingly prominent. One of the Six Fancy Ideas (future post!) I took from Edward Tufte’s first book (The Visual Display of Quantitative Information, 2001) is to remove or minimize non-data ink, or, in this case, less important data-ink. This is a version of the old cartographic adage about visual hierarchies: stuff on your map that is more important should jump out, and stuff that is less important should fall back. The county boundaries (in gray) are so prominent that they obscure the data in smaller counties on the MLA map (at a national scale) and are certainly overly noticeable at all scales. Why not just make the boundaries white, and as fine as possible, so we can see the most important part of the map, the language data?

The map projection (the manner in which the surface of our 3D earth is flattened to 2D) is the plate carree, a typical default for web mapping software. Unfortunately, the plate carree distorts area (as you move north) thus this map projection is distorting the data. Dammit! Counties and states in the north are bigger than they should be in comparison to the counties and states in the south, which may distort your interpretation of the area patterns on the map.

Finally: map crap alert! Why include a north arrow when the vast majority of map viewers will recognize the U.S. and that north is up is? Why a scale bar showing 838 (?!) miles? Zoom in and you get 419 or 209 miles. Is anyone ever going to need to make measurements on this map? And the plate carree projection distorts distances anyway.

So: a decent, focused and useful site that could use a few tweaks. The MLA site also allows you to compare two language maps and play a bit with the data (in the Map Data Center).