All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Envisioning Information
Visual Explanations
Beautiful Evidence
Paper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
All 4 clothbound books, autographed by the author $150
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer:
Visual and Statistical Thinking $2
The Cognitive Style of Powerpoint $2
Seeing Around + Feynman Diagrams $2
Data Analysis for Politics and Policy $2
catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
Atlanta, February 23, 24
Miami, February 27
Tampa, March 1
Boston, March 14, 15, 16
Mapped pictures: image annotation

Diagramatic pictures allow more focused, tighter commentary on pictures, compared to text or captions outside the picture frame. I've long been concerned at how the format of this board seems to enforce a separation of text and image. I'll be trying out more mapped pictures here and would appreciate good examples of mapped pictures (for thinking about the issue in Beautiful Evidence).

Instead of a mapped picture with direct labels, here's an impediment visual-verbal encoding:

                     A' = Anna                                                    Z = Zerlina                      A'' = Alex                           A''' = Abby

-- Edward Tufte

Response to Mapped Pictures

Here's a couple of diagrams from the pre- and nearly pre-photographic era. Both are from a nifty book, "You Are Here — Personal Geographies and Other Maps of the Imagination," by Katherine Harmon, published this year by Princeton University; perhaps you've seen it. It's a presentation of "personal maps" and has mostly map-like images of personal histories, literary places, artworks in a cartographic manner and such. But one section contains a number of very fine images of the human body in one form or another, from sources a little more off the beaten track: phrenology charts, Asian medical manuals, acupuncture diagrams, a variety of contemporary art. I'm not sure if these two qualify as mapped pictures but thought they might be of interest.

Kunisada Utagawa, Rules of a Dietary Life, ca. 1850 ... "It shows the functions of the respective internal organs and instructs the reader about proper nutrition."

Sangye Gyamtso, Interconnecting Blood Vessels: Anterior View (detail), ca. 1680s ... "The chart depicts three or four major "channels" of Tibetan Buddhist medicine, with information on which blood vessels are acceptable for bloodletting."

-- Steve Sprague (email)

Response to Mapped Pictures

The idea of mapped pictures is presented in this opening of a chapter in Beautiful Evidence.

-- Edward Tufte

Response to Mapped Pictures

Ah insight! So those pictures appearing in various sign language dictionaries to teach me the language are "mapped". I don't know if the ASL/English Dictionary edited by William Stokoe used mapped pictures but certainly the Dictionary of British Sign Language/English has them—indeed I look at them regularly so they are the ones I know best. Almost every dictionary/gloss-list of signs published here in the UK since 1992, when the BSL/English dictionary was published, uses mapped pictures in a similar style to those featured in the Scientific American article about a sign language created by Nicaraguan children:

sign language in Nicaragua

True languages get much of their power by breaking complex ideas into small pieces, such as words, which can then be rearranged to form innumerable new ideas. "It's like building with bricks rather than building with clay," Senghas explains. To look for this feature in NSL, the researchers played a cartoon showing a cat wobbling down a hill after eating a bowling ball. When asked to describe the motion, Spanish speakers who can hear often augment their verbal description with a gesture that combines the ideas of "wobbling" and "down" in a single motion (top). Deaf students from the early years of the school use a similar gesture, which is a direct analog of the distinctive motion. In contrast, later generations of students separate the ideas of "wobbling" and "down" into distinct signs (bottom). Although the separation actually weakens the description, it is an important feature of abstract language, the researchers note.

The direction (sometimes width) of the arrows encodes semantic information that is otherwise difficult to describe.

-- Trevor Jenkins (email)

Response to Mapped Pictures

The sciences, and anatomy in particular, do seem to lead the way when it comes to this sort of picture mapping. I can remember marveling at those transparent overlays that were mentioned earlier, in the Encyclopedia Britannica when I was a boy. Take Henry Gray's Anatomy of the Human Body as another case in point.

Notice the different typefaces and colors used to distinguish between various tissues, such as muscle and bone (although they do not seem to be completely consistent in this example), and the hyphenation of the digastricus as it passes beneath the stylo-hyoideus. Note also the inconsistent type orientation, how some of the vertical muscles are labeled with the text running up-to-down, while others run down-to-up.

I assume that the drawings in the early editions of the book were quite a bit larger than the one shown here. The illustrations in my commercial paperback edition from the 1970s are a little bit bigger, but not much, and the type in the diagrams is still fairly difficult to read. (Indeed, some of the diagrams are much more detailed than the one shown here and are nearly illegible at their currently published size.)

One wonders if these particular labeling techniques weren't in part a deliberate means to promote more careful study of the very rich illustrations in Dr. Gray's book. After all, the medical student is presumed to have seen all of these structures in the flesh (so to speak), and can study their names more clearly in the text. The object of the diagram is to associate the names of the body parts with the actual elements the students have already seen during a dissection.

Many of the other diagrams in the book use callouts with thin, arrowless lines pointing to the subject of the label. Still others use a combination of callouts and subject labels directly on the structures. In fact, it's interesting to think about why Gray chooses to use callouts instead of direct labeling in various situations. Sometimes it is simply a matter of available space, caused by the need to specify a long name for a small body part. In other cases, the use of callouts seems also to be related to the overall shape of the subject he is diagramming. Cranial diagrams, for example, seem to use direct labeling more often, and avoid callouts except for structures near the outer edges of the picture. A diagram of an arm or leg, by contrast, might use callouts exclusively. Gray seems to try to avoid callouts with long lines wherever possible, which would obviously be more difficult to parse.

This book has fascinated me for years. I'm glad for this discussion that has urged me to pull it out and pore through it again. Sometimes, you have to be reminded that there is as much detail looking inward as there is looking outward.

-- Craig Zacker (email)

Response to Mapped Pictures

What a thoughtful contribution, an interesting example right on point from Kindly Contributor Craig Zacker.

Let us all now locate our sterno-cleido-mastoideus.

-- Edward Tufte

Response to Mapped Pictures

A hidden gem in Craig Zacker's post is the source of the picture: With its host of free, classic references and a reading list to aspire to, this site alone motivated me to learn how to set bookmarks in my browser and make hyperlinks in documents.

-- Niels Olson (email)

Response to Mapped Pictures

Netter's image that compares to Gray's above.

Netter's anatomy

Netter's callout style contrasts well with Gray's cartographic style. To see the callouts, find plate 23 in Netter's Atlas of Human Anatomy, third edition:

Netter's anatomy with callouts

It is worth noting that Gray names 28 structures, while, by using callouts, Netter can name 38. Netter's decompressed style also minimizes the need for line-breaking hyphens (Sterno-cleidomastoideus). Gray's cartographic style may be a nice bit of visual confection, but Netter ties more information to his image (if only I could find the image with callouts out on the web!).

-- Niels Olson (email)

Response to Mapped Pictures

Gray's drawings exaggerate the figure-ground contrast, the thick black lines on the lighter ground. Color is an after-thought.

Netter's drawings avoid the black stripes, resulting in a more subtle, 3-dimensional effect, closer to reality.

-- Edward Tufte

Response to Mapped Pictures

At what point does a "mapped picture" simply become a "map"?

Aerial photography, arguably consisting of images/pictures, is commonly overlayed with annotations and then used as basemapping in land use planning and natural resource management (as well as obvious military applications and other fields).

Not knowing if it is an accepted term or not, I refer to these as "photomaps".

Some examples include NRCS county soil survey mapping showing various soil series designations:

soil survey map

or FEMA floodzone/flood insurance rate maps showing estimated flood risk areas:

FEMA flood zone map

For a recent open space exhibit, we displayed maps and photomaps of identical areas, at the same scale, side-by-side. The public seemed to extract information from both, in a parallel yet combined form, as both versions have strengths and weaknesses.

-- Mark Kasinskas (email)

Response to Mapped Pictures

For Niels Olson's request for Frank Netter's work online, Google image search for "CIBA Netter" turns up a few examples. Nice stuff.

-- Peter Lauterbach (email)

Response to Mapped Pictures

Some additional Brodel images.

These are available at the National Library of Medicine's profile of Florence R. Sabin, but aren't laid out together there. The style is very similar to Wolf-Heidiger's. Interestingly, Powell's was selling Kelly and Brodel's two volume Gynecology and Abdominal Surgery for $45, while another seller on Amazon was selling it for $270. I bought the copy from Powell's; I hope it's still readable.

Between Google and the continued exponential growth of on-line content, it is becoming remarkable easy to add information (and hopefully content) to these threads. I wonder if initiatives like The Digital Information Infrastructure and Preservation Program will help further the trend or if more content will be put behind firewalls, only accessible to intranet computers and subscribers.

-- Niels Olson (email)

Response to Mapped Pictures

I should mention that the department of Art as Applied to Medicine at John's Hopkins University has an large trove (probably most of the finished work of his long career) of original Brödel pen and ink and carbon dust illustrations in their archives. Unfortunately, they are not usually on public display. But, for the students and faculty of the program he founded at JHU, they provide extraordinary inspiration.

-- Nick Woolridge (email)

Response to Mapped Pictures

The rock 'n roll timelines at the Rock and Roll Hall of Fame:

Rock and Roll timeline

-- mike (email)

Response to Mapped Pictures

Tufte's text asks: "Should not every explanatory image of DNA - from journalism to science - show the molecule with a scale of measurement?"

I think not. Not every image, and particularly not every scientific image. As discussed in other threads here, biochemical diagrams often are used summarize information about diverse genetic and chemical interactions. Single diagrams often incorporate graphical representations of objects ranging from a dozen base pairs (a footprint that can be recognized by a transcription-regulating protein) to cells (>10^3 larger), tissues (another 10^3-10^7), or even organs. In such diagrams, which are extremely common, scale is not the point - informational or chemical topology is. In a diagram of the topology of a computer network that incorporates both local and large-area information, scale bars showing the diameters of Cat5 or T1 cables would be chartjunk, and nothing more. The same is true in many diagrams that include pictures of DNA.

Moreover, for biologists who focus on molecular structure, a detailed scale drawing that includes a DNA helix (2 nm diameter, 10 base pairs per turn, 3.4 nm per 10 base pars), or an alpha-carbon backbone in a protein (0.5 nm backbone diameter, ~3.5 amino acid residues per turn, spaced about 0.15 nm apart), those elements provide implicit scale bars. These are scale indicators as recognizable to a molecular biologist or biochemist as a person standing next to Tufte's Spring Arcs sculpture would be to most people. It is for this reason that one rarely sees scale bars in specialist papers about protein and DNA structure.

Certainly, scale bars should be present much more often in diagrams of DNA and other macromolecules. But scale bars are certainly not desirable in every illustration of this kind.

-- Alex Merz (email)

Response to Mapped Pictures

Watson and Crick's 1953 proposal in Nature for the structure of DNA included a graphic but didn't include a scale bar, although scale could be interpreted from data in the article.

original figure from 1953 Nature article

As a student taking Biochemistry right now I concur with Prof. Merz's opinion. In addition, I would add that scale bars are of minimal utility when trying to figure out mechanisms (is this carboxyl oxygen close enough to that amine to abstract a hydrogen? Can a hyrdrogen bond form there?) when compared to the 3-d ruler functions in the visualization programs. Like engineering and architecture visualizations, visualization of macromolecules is generally done with specialized software. PyMol and RasMol are examples. If they're known, the coordinates of each atom in a biomolecule in a particular conformation are available from various sources like the Protein Data Bank, which are constantly growing and provide the raw data for the visualation programs. The image below is from A. Geva, et al's report in this week's New England Journal of Medicine on a novel hemoglobin mutant, hemoglobin Jamaica Plain. The red lines represent 3-d ruler measurements, the lengths of which are reported in another frame of the visualization program this image was taken from. In this case the authors merely used the lines to indicate steric interactions created when the bulky amino acid phenylalanine is coded for where a smaller leucine should be. One (small) reason for these authors to not provide a scale bar would be that, as they state in the article, the mutant's structure has not yet been determined by x-ray crystallography (the same technique Rosalind Franklin used in her study of DNA), they just took the next closest structure, sickle cell hemoglobin, and stuck a phenylalanine in where a leucine should be.

DNA structure of Jamaica Plain

The helices named and illustrated above are examples of the alpha helices that Prof. Merz describes.

Microscopy and images of planetary surfaces are mainly flat, so scale bars are appropriate. In molecular modeling, a scale bar may in fact be misleading because the models generally accommodate neither small changes in bond length and orientation that are inherent in macromolecules, nor significant perspective challenges posed in imaging macromolecules.

To the original question: in Geva's image, there's a trade off between abbreviation and putting the label close to the object. Compared to "F68", writing out "The 68th amino acid in the peptide sequence, a phenyalanine" would be more democratic, but would also require leader lines.

-- Niels Olson (email)

Protein Anatomy

One other addition to Niels's post: "ribbon" diagrams protein structure, as shown in the picture attached to Niels's post above, use idealized shapes to depict the path of the polypeptide backbone through space: helical coils or even simple cylinders for alpha helices, arrows for the strands that make up beta sheets. The peptide backbone itself follows a more angular, zigzag path (see here for an example from the original Pauling/Corey paper). These idealized representations, pioneered by Jane Richardson at Duke University, are now the standard method for depicting the global 3-D fold of a polypeptide. They are sometimes called Richardson diagrams.


In fact, the three-dimensional structures of macromolecules are sufficiently complex that their representation has posed problems for structural biologists that are in many ways comparable to those confronted by classical anatomists. Such models, of course, usually peel away the molecule's partially-ordered hydration shell and VanDerWaals "surface" to reveal the underlying bonding topology and fold, just as anatomists leave out or peel away skin, muscle, and connective tissue to show us the skeletal structure. It is no accident that we refer to a protein's "backbone" conformation.

This reminds me of an additional inaccuracy: macromoecules are not, in nature, alone. They are in a dense sea of other macromolecules, membranes, and solutes. Often they are assembled in solid-state machines. Most often, we have little idea of what these larger machines look like. This density is rarely illustrated, both to reduce visual clutter, and because depicting the milieu poses serious challenges in illustration.

One illustrator who has confronted these problems head-on is David Goodsell, a structural biologist at Scripps. His work is so good, and in my opinion so important, that I implore Dr. Tufte not to put Beautiful Evidence to bed until Goodsell's work has been examined (if it has not yet been). Goodsell's now got three books. Of the first two, I recommend The Machinery of Life. It is already a classic. Goodsell has a third book out now, Bionanotechnology: Lessons from Nature, that I've not yet read; I look forward to it as I look forward to reading Dr. Tufte's own work.

-- Alex Merz (email)

Response to Mapped Pictures

I would like to second Prof. Merz's endorsement of David Goodsell's work (he also authors and illustrates the enormously informative Molecule of the Month at The Protein Data Bank).

He works in two distinct yet complementary modes:

1. He creates unique linear renderings of molecules based on depth maps produced by a molecular modelling program (Rasmol). For some techniques to reproduce this approach, I have made some hasty notes.

2. He creates pen and ink and watercolour paintings of the molecular " mesoscale", where the shapes of biomolecules are simplified, and represented, as much as is presently known, in their true scale, numbers and interrelationships in a shallow cellular landscape. These latter are especially beautiful, and remind me of an exquisite handmade quilt.

-- Nick Woolridge (email)

Response to Mapped Pictures

A fine pair of mapped pictures comparing image methods for coronary arteries. From the New York Times, November 17, 2004, p. A24. Original source: Dr. Mario J. Garcia, Cleveland Clinic.

-- Edward Tufte

Response to Mapped Pictures

Photographic images and drawings are different, and need separate discussion. There seems to be several direction in how photos can be used:

(1) adding a layer of text labels and lines to photo;

(2) text labels and lines + some processing of the photo (reducing brightness, etc);

(3) drawing a diagram from the photo and showing it in pair with the unchanged photo;

(4) series—using a series of photos in evidentiary function;

(5) collage—using pieces of photos to build a compilation.

The first direction appears very limited and often produces unsatisfactory results. The last direction (montage) probably cannot be used as evidence in strict sense.

-- Priit P. (email)

Abbreviations and labels

This is a rather belated comment on Niels Olson's remark above:

Specifically, he said: in Geva's image, there's a trade off between abbreviation and putting the label close to the object. Compared to "F68", writing out "The 68th amino acid in the peptide sequence, a phenylalanine" would be more democratic, but would also require leader lines.

The tradeoff that he mentions is indeed real, but I'm surprised at how often illustrators prefer to leave a lot of space rather than expand their abbreviations. This isn't a fault in the example given, as most of the space is used, but it's quite common to find illustrations where the elements are labelled too cryptically, or not at all. However, even though there is certainly not room to write "The 68th amino acid in the peptide sequence, a phenylalanine" there would be room to write "Phe-68", with a gain in intelligibility. Everyone who works with protein sequences knows that F is phenylalanine, but the potential audience is wider than that, and for everyone who knows that F is phenylalanine there are probably 20 people who know that Phe is phenylalanine, and at least some of the extra 19 probably care (whereas the countless millions who don't know that Phe is phenylalanine probably don't care either).

The example of a protein structure illustrates another point (especially if contrasted with ET's dogs on 9th September). In order to be easily legible white text against a dark background needs to be either bold or at least set in a font that is naturally bold. Whereas the names of the dogs stand out nicely, the labels of the aminoacid residues look distinctly spidery.

-- Athel Cornish-Bowden (email)

Response to Mapped Pictures

I think the molecular graphics example in this thread raises an interesting point that is while not exactly on the topic of this thread is certainly related.

Of the scientific illustrations presented in this thread, the molecular structure of haemoglobin is the only illustration on a dark background—I discount the photograph of Crick and Watsons structure as just that, a photograph.

In Envisioning Information, Prof Tufte moved a similarly computer generated graphic (A storm simulation rather than a molecular model) from a dark background to a light background with a substantial increase in clarity. I think a similar effect could be obtained here with an associated increase in ease of labelling.

I believe this is a common problem in structural biology. The researchers use computer generated images for evaluation of the model, and of the supporting data, using a dark background to reduce display glare, but then only minimally modify this image for publication purposes despite the different needs of the print medium.

Better ways to depict molecules in print exist and have already been mentioned. Jane Richardson's work on cartoon representations of molecules has been translated into software by Per Kraulis (Molscript). This program, commonly used, produces excellent illustrations in postscript mode using many illustrators tricks such as careful use of think and thin lines(c.f. Understanding Comics by Scott McCloud). Notably, in postscript output mode the program assumes a white background and the output has dark outlines around the ribbons to aid in distinguishing one helix or chain from another. Annotation may be done within the program of by importing this postscript file into an illustration package.

Molscript did, however, take a step backward in its most recent version providing only dark background "screenshots" in it's Encapsulated PostScript output (Actually a TIFF image in an EPS wrapper). This step was presumably taken so that images could be incorporated into MS Word documents and powerpoint presentations (Neither application handles postscript well). Using the postscript output now requires a conversion to EPS by a separate program before it is able to be included in a document. This is relatively easy and well worth the effort.

I think that many structural biologists are trying to produce "photorealistic" images when no such reality exists to be photographed. The illustrations are of models, of carefully selected information to show what the author wants to you to see. The photorealistic approach may be an attempt to be more convincing. To draw attention from the fact that what is shown is a model, to give the reader the impression that this is "reality".

The analogy is that of a satellite photograph of a region and a map—the flawed molecular model image tries to be the photograph of what "it" looks like when it should concentrate on being the map of what we know.

A related question: then is how does the mapping of an illustration differ from the mapping of a photograph? How does the production process affect the subsequent mapping and is this a basis for choosing one over the other?

-- John Walker (email)

Response to Mapped Pictures

Jim Bumgardner's recent use of Flickr to plot the time of sunset during a year is a counter example to Priit P.'s assertion "(montage) probably cannot be used as evidence in strict sense." See the thread Images Used as Data Points.

mapping sunsets

About 15000 photos tagged "sunset" taken within the last year.

Their horizontal positions represent the day of the year the photo was taken. January is on the left, December is on the right. The vertical bars are the boundaries between months.

The vertical position represents the time of day the photo was taken, according to the EXIF data. The horizontal lines are hours, with the thick line in the middle representing 12 noon.

The deepest "dip" in the wave formed by the images is the Summer Solstice.

-- Niels Olson (email)

Response to Mapped Pictures

Gloria Hopkins' essay on photographic composition includes some good examples of mapped pictures. In these cases, the purpose of the mapping is to explain in objective terms the attributes of successful compositions. The "maps" displayed are useful in fulfilling their purpose, and Gloria implements the technique well.

mapping photographic composition

mapping photographic composition

The diagrams are notable for another reason, though. "Escaping Flatland" is central to the ideas behind much of ET's work. These composition maps represent a way to help people enter Flatland—they act as tools to help photographers visualize a three-dimensional (or more: consider that a photograph is not an instantaneous event, but has duration; there are numerous other dimensions in a photograph) image in two-dimensional terms.

-- Scott Zetlan (email)

Response to Mapped Pictures

Another mapped picture of the sternocleidomastoid. Style is lacking, but it's dripping with interesting content and highlights the need for the calmer anatomical drawings above. Picture from the Clinical Cases Central Line Placement Procedure.

-- Niels Olson (email)

Mapping New Orleans during Hurricane Katrina, from the New York Times.

Hurricane Katrina
Hurricane Katrina
Hurricane Katrina
Hurricane Katrina
Hurricane Katrina
Hurricane Katrina

-- Andrew Nicholls (email)

Hurricane Katrina

The white background in the boxes should be somewhat transparent; the black outlines around the boxes should be omitted. The annotation is thin, the equivalent of a couple of paragraphs in a story. And why the map? It is not very helpful and is covered up by the boxes anyway.

A much better way to do all this, at least on a newspaper page with more space and better resolution than a television computer screen, is to place much more annotation surrounding the map, with light pointer lines then crossing over the map to the relevant area. With more annotation, perhaps some of dynamics of what happened could be indicated. This method would also make use of a newspaper's great advantage—lots of high-resolution space within the eyespan.

-- Edward Tufte

Magerius mosaic

This Roman mosaic from a villa in North Africa commemorates a venatio—a show of simulated hunting—paid for by the villa's owner. The performers are professional touring beast-fighters called venatores, who are shown killing captured North African leopards. There appear to be labels naming not just the venatores but also the leopards they dispatched. The figure in the centre carries a tray bearing bags of money, and the text surrounding him describes the performers' request for payment. (more details here)

Magerius mosaic detail

-- Derek Cotter (email)

Excellent, especially the full image.

-- Edward Tufte

Fascinating movie from NASA of descent of Huygens probe. Among mapping mappings, note the very clever scaling, a footprint on the Earth's Moon, near the end. Read the text first carefully before viewing.

-- Edward Tufte

Here is an example of annotations in different cultures and a free tool allowing annotate and share digital images. I hope it can be interesting to you and the community. I'd be glad to hear your feedback, especially if you consider the tool useful or have any comments or improvement suggestions.

Sumerian seal
Greek vase
Bayeux tapestry
Palermo mosaic
Last Supper
Mughal portrait

-- Alex (email)

Notice that none of the examples above have Post-It-like color fields containing the type. Most examples in the proposed tool, however, are demonstrations of Post-It labeling, word boxes covering images, not examples of words working with images. Use then, at a minimum, more subtle transparent fields to hold the type on the image. Overall there needs to be much greater design and typographic subtlety in the proposed tool. Putting Post-It notes on existing images maintains the unfortunate segregation of text and image.

Post-It tool

More generally, the tool is driven by a programming solution rather than by ideas about text and image together. Substantial theoretical statements, with many examples, about words and images together are:

Meyer Schaprio, Words and Pictures: On the Literal and Symbolic in the Illustration of a Text (1973).

Meyer Schapiro, Words, Script, and Pictures: Semiotics of Visual Language (1996).

Edward Tufte, Visual Explanations: Images and Quantities, Evidence and Narrative (1997), 120-151.

Edward Tufte, Beautiful Evidence (2006), 82-121.

That's your reading assignment before continuing further.

-- Edward Tufte

We wrestled with the label/mapping problem when creating a new series of meridian charts for use by students, teachers and practitioners of TCM based modalities such as acupuncture and shiatsu. Such charts are characterised by complex meandering linework which, in accord with the esoteric nature of TCM, is coloured according to the Five Element theory. Most other charts used artistic renderings which allow easier control of the interplay between scale, depiction of anatomical features and labelling. In our pursuit of realism (accuracy and veracity) we chose to markup a student and photograph him. Issues with scale, skin colour, mixed contrast lighting, skin reflection and stubble became real stumbling blocks.

Most often the labelling of specific points and features in drawings is accomplished both within and without the image outline. In our case we were cognizant of the 'lifecycle' that such charts have in the hands of a typical student who progresses from close study to a working practitioner. By the end of that cycle, at best the chart is hanging up in the waiting room, shocking the patients. By placing all labelling inside the body outline we were able to preserve the aesthetics of the imagery when viewed from a distance and hopefully have a calming, if not pleasing, effect on the viewer.

In the three images below you can see firstly the simple linework of the TCM, next the 'heretical' "zen extensions" of Shizuto Masunaga, and then the TCM and Zen linework combined. The majority of information is carried by the colouring, opacity, thickness, layering and, in some cases, patterning of the linework. On the combined chart it must also show where Masunaga agrees or disagrees with TCM orthodoxy. Sometimes it must also explain why. Labelling per se is restricted to the points, some of which have special roles, and are colour coded accordingly.


Masunaga's Zen extensions

Combined TCM and Zen extensions

Click through to a closeup of the legends.

-- Lee Rydstrand (email)

Here are some sketches for mapped images as an overall design strategy for books and computer screens, at least for highly visual material that is first revealed and then explained. Below, in Bouquet, some good edge-play in the light cast onto the element at left. Note the Necker Illusion in the vertical fold, which reads as both folded-in and folded-out in relation to the viewer.

Multiple edges cast light to the left of the vertical; those edges themselves appear folded, this way and that way, depending up the reading convention applied by the viewer to the image.

Immediately above is an experiment in mapping images with detailed text. The idea is to pair an image without annotation followed by the same image annotated. Let the viewer explore, then point the viewer to particular elements in the image.

Could a book (or computer screens) be written in this format of direct captioning? Words that discuss the overall image would be in the usual format of text—sentences and paragraphs—but words that discuss local detail would be placed right on the image itself. This would finesse the problem of finding the detail and would illustrate directly what a close look at the image can reveal.

Perhaps the sequence of the unannotated image preceding the annotated image accommodates both unguided exploration followed by guided analysis. This is similar to the usual teaching technique: show something and ask "what do you see," then encourage a closer look ("use the little gray cells" as Agatha Chritie's Hercule Peroit probably said), finally followed by showing what a close (annotated) look can in fact reveal.

-- Edward Tufte

How would it work in the book? Would the annotations be added using a transparency? Would they be facing pages? Would they be in the same location on different two page spreads? Each scenario might influence the way that the reader scans the two images. I am not sure that I could look at the former without glancing at the latter unless it was hidden from view.

Another interesting element of this image is use of arrowheads and the angles of the lines.

-- Tchad (email)

Annotated (but not scaled) astronomical picture

Today's Astronomy Picture of the Day is from last week's lunar eclipse. I've seen multiple exposures across the sky many times, but always against a fixed azimuth and altitude. Because these take many hours of the night, the picture shows a very long sequence in terms of moon diameters. This one is a multiple exposure against the fixed star background, and the moon travels a much smaller distance across that (approximately 1/28 as far, i.e. in the ratio of about one day to one month).

The map of the sky is nicely labelled, with the zodiacal constellation of Leo marked up and named, and the ecliptic drawn in, with a circle where the Earth's shadow is, showing the moon entering and leaving it. Saturn is also pointed out, in its current position on the ecliptic. I wish the ecliptic and the shadow circle had not been such a dark blue, as it's very hard to see on my CRT monitor when it's at a comfortable setting. I'm also prompted by the Ask E.T. discussion on the subject, to feel more and more annoyed by the absence of a scale in degrees for these pictures. I happen to know the Moon is about half a degree in diameter, but still.

-- Derek Cotter (email)

Here's a version I altered myself, which I hope shows the blue lines more clearly.

-- Derek Cotter (email)

The lens of the human eye yellows with age. Yellow opposes blue, with the consequence that seeing blue on black is increasingly difficult with age. Out of sympathy for those of us with aged eyes, may we never annotate with blue on black (or black on blue).

-- Dan Dill (email)

There are many other such claims made by small subgroups of users about color design:
the 5 (or is it 7?) types of color deficiency, various needs for certain contrast ratios, and so
on. Then, in practice, these colors pass through all sorts of computer monitors (some
adjusted properly, and many not) with all sorts of varying color renditions and contrast
ratios. Accommodating some special interest claims may require, for example, glaring
colors and high figure-ground contrast ratios, which have substantial costs (high contrast
ratios lower the resolution of the information and also may result in eye fatigue during
extended use).

Unaccommodated special interests seek to regulate design by means of legally mandated
standards, which may result in lowest common denominator design or in blocking
innovative designs. Sometimes those standards accommodate both deserving and non-
deserving claims. Probably most professional GUI designers can provide horror stories
about how mandated special interest standards have corrupted good design. One such
story, possibly apocryphal, was that the brilliant NeXT computer could not be sold in
Europe because it was black (rather than computer-plastic tan) and thereby in failed to
meet ISO standards (European GUI standards, some years ago, were written by a German
veterinarian who was the low bidder for the standards-writing contract). Big computer
companies sometimes dominate the standards-writing committees and use their
domination to raise entry costs for small innovators. In turn, a standards consulting
industry and standards departments proliferate—as a special interest for special interests
becomes a bureaucratic career—adding one more thumbprint on the design and one more
approval. And bureaucratic regulation has failed to solve a notorious color design issue:
the red/green traffic light.

The design issues are quite likely to be more subtle than the special interest and
regulatory standards imagine. This is in part because design elements interact and are
intensely multivariate, but special interests and standards are univariate. Excellent design
has a hard time surviving interest group politics and centralized bureaucratic bloat.

What to do?

First, consumers can often come up with strategies to manage their special difficulty on
their own. Thus wearing glasses by some avoids forcing giant type on the many. Also
much can be done by user adjustments of their own computer monitor. Furthermore, the
individual user is already likely to be an expert in making her or his necessary
adjustments. Now and then, the designer might provide tips to various special interests to
help them better use the design.

Second, market rather than regulatory solutions may yield some mutual adjustment
between the designer and the special interest.

Third, designers should carefully distinguish between deserving and undeserving special
interest claims.

Fourth, designers should of course try to accommodate the deserving special-interest
claims without compromising the design for the many and without imposing costs on the
many. The spirit of the designer should be a generous respect for deserving special
interest claims accompanied by an evaluation of the opportunity costs (and the transfer
payments from one class of users to another) of satisfying those claims. Realizing this
spirit in practice may be difficult.

(Re the example in the contribution above: I like the blue line, although it might be one
small step lighter. Perhaps some clever anti-aliasing can help here. Little contrast between
figure and ground is necessary to indicate a line. If the line becomes at all light in color it
will glare and flare. Since little contrast is need to specify a straight line, and more contrast
to specify type, the standards committee is already over its head, or at least has gone to
the dogs.)

-- Edward Tufte

To improve conformity across different displays, embed an ICC color profile in images exported from Photoshop. It's not perfect, but it helps.

Unfortunately I don't know a good reference for further information.

-- Rob Simmon (email)

Came across a wonderful resource showing World War Two cutaway diagrams and explanations of various devices from Parachute to Submarine.

mapping WWII

mapping WWII

mapping WWII

mapping WWII

mapping WWII

-- Bill Paton (email)

A mapped photograph of the geologic boundary between the Neoproterozoic and the Cambrian:

mapping geologic boundaries

The shale is younger than the limestone, and the contact between the two occurred at or near the beginning of the Cambrian, when complex, multi-cellular life became commonplace.

-- Robert Simmon (email)

Kernel Summit 2007

A number of fairly famous contributors to the Linux kernels are identified here. The photo is from

mapping Linux contributors

-- Niels Olson (email)

A very famous series of hand drawn walking guides to over 200 individual walks in the "fells" and hills of the English Lake district (Beatrix Potter country). These were made by Alfred Wainwright who lived in the lake district and was a passionate walker.

Wainwright walking guide

They are superb examples of data integration. They include (all hand drawn) text descriptions of the walk, views from the start, views from the summits including angular arrangement of key features visible and distances, topographic information, maps, personal commentary and humour etc. His book series (A Pictorial Guide to the Lakeland Fells: being an illustrated account of a study and exploration of the mountains in the English Lake District) is still in print and over 2 million copies have been sold. The books were made by him to be used as the walker progressed through the walk and are therefore very nice examples also of "Instructions at the point of Need".

A nice collection of Wainwrights own favourite walks has been published recently.

-- Matt R (email)

dead trees mouseover

By printing the labels on something like waxed paper, the ingredients in this cookbook can be viewed with or without mapping.

Source: Ma, Nancy Chi. 1960. Mrs. Ma's Chinese Cookbook. Rutland, Vermont, and Tokyo: Charles E. Tuttle Publishers. (28th printing, 1972)

-- Nathan Vander Wilt (email)

-- Edward Tufte

Here's a lamp, a giant lamp, at a recent auction:

Source: Sotheby's Important Postwar & Contemporary Design, New York, November 16, 2007, 132-133.

-- Edward Tufte

For a few versions now, Adobe Acrobat has supported layers and linking. My team has taken to using this feature to make user-selectable overlays available in portable electronic documents, specifically for images of small mechanical assemblies, extracted from larger parts, shown in cross-section. These small parts are mounted in epoxy, then sectioned and polished, and finally imaged on an optical microscope.

We first present a macro image of the entire cross section, then place overlays with links to higher magnification areas of interest. The "button" layer can be toggled on or off by the user. The layers typically contain information about damage and quantitative measures of the constituent materials.

With the advent of Acrobat 3D, we can now extend this approach to a wireframe or other solid model, allowing the viewer to see layers of information which compares the actual (photograph) to the ideal (3D model). The 3D model is selectively overlain to the micrographs, and can be rotated and even disassembled in the Acrobat reader. This makes conformance and deviations of the part quite clear. It also allows the viewer to gain better perspective of the assemblies, and how they fit into the larger part.

This is not strictly an example of a diagrammed or mapped picture, but instead a moment beyond which allows the viewer to interact with an arbitrary combination of data in a single screen.

-- S. Valentine (email)

Here is an extensively mapped image database of Trajan's column. The site maps the continuous spiral layout of the column's artwork with cartoons that have annotations and links to high resolution images. The navigation allows you to move up and down the column and see either individual panels or a whole spiral piece around the column at a particular height. The image below shows the navigation mechanism.

Trajan's Column

mapping Trajan's Column

-- Matt R (email)

I really like the use of technology in this chart from the NYTimes to further illustrate the point by providing more information about each movie (click through to interact).

mapping box office success

-- Oliver Passemard (email)

Readers may like to know about the Isotype revisited project.

This interesting research project traces the history of the Isotype. It draws on the Otto and Marie Neurath Isotype Collection at the University of Reading; which, judging by the images available on the project website, offers a rich and valuable source for understanding modernist approaches to visual communication, graphic design, and information visualization.

-- Catherine Howell (email)

Threads relevant to analytic design:

Seeing Around: New ET essay published