The Graphic Truth

We have all been taught to be critical of the written word, but we tend to let the omnipresent graphics of our era pass without close scrutiny. This we do at our own loss and peril. Graphics are rich stores of information, but often they lie about quantitative information and are unnerving and confusing when they could be aesthetically pleasing. In his recent book, The Visual Display of Quantitative Information, Edward R. Tufte tells us what we should treasure, reject, and censor in graphics—and much more. It is a thoroughly engaging and handsome volume that should inform and entertain everyone from scientists and engineers to economists, as well as every kind of historian.

The book deals with the design of statistical graphics—a formidable subject until Tufte shows us that it is “often the most effective way to describe, explore, and summarize a set of numbers” and that it is, in fact, a device for looking “at pictures of these numbers.” Tufte is well qualified to write this elegant and thoughtful introduction to the subject—he is a professor of political science and statistics at Yale University, a consultant on graphics for the media, and also the founder of Graphics Press, of Cheshire, Connecticut, which published the book.

Showing numbers with abstract, nonrepresentational pictures is, Tufte informs us, a surprisingly recent invention. J. H. Lambert (1728–77), a Swiss-German scientist and mathematician for whom the “lambert” unit for measuring the brightness of light is named, and William Playfair (1759–1823), an English political economist, were prominent among the innovators. Playfair’s imaginative and craftsmanlike graphics have stood the test of time so well that Tufte uses them to make points about good contemporary graphics design. Playfair’s The Commercial and Political Atlas (London, 1786) has forty-four charts, all but one of which are time series that show, for instance, imports and exports and the balance of trade in England over almost a century. The one exception is the first known bar chart, which he invented to portray one-year data.

Tufte illustrates with classic graphics, some of which dramatically and deftly convey the essence of momentous historical events. With brutal eloquence, Charles Joseph Minard (1781–1870), a French engineer, shows the tragic course of the invasion of Russia by Napoleon’s army from 1812 to 1813. Minard extends across Russia a band diminishing in width as it approaches Moscow and shrinking to a thin line as it retreats to Poland. The full width represents the invading army (422,000) and the pathetically thin one the remnant straggling into Poland (10,000). This multivariate graphic plots army size and location on a map, direction of movement over time, and the temperature on various dates during the deadly winter of retreat. Tufte comments that “it may well be the best statistical graphic ever drawn.”

A superb diagram by E. J. Marey (1830 – 1904) also exemplifies the clear visual presentation of complex relationships. Marey shows graphically a schedule for Paris-to-Lyon trains in the 1880s. The intervening train stations are named along the vertical axis and separated by intervals proportional to the actual distance between them. The horizontal marks the twenty-four hours. The slope of the lines that represent the various trains gives, therefore, the speed of the trains—the steeper the slope, the faster the train. (A vertical line would mean velocity at the speed of light, an achievement still eluding even the new Paris-to-Lyon express.) The intersection of upward lines with the downward gives the time when Paris-to-Lyon trains meet those heading for Paris. Every railroad buff should have the Marey graph among his collection of schedules.

There are other gems. Dr. John Snow used dots on a map of London to locate cholera deaths in London for September 1854. He also showed water pumps. From the plot emerged a pattern of deaths around the Broad Street water pump; when the handle of that pump was removed, the neighborhood epidemic that had taken more than five hundred lives ended. Some historians still use statistical graphics in the detective work of finding pattern and cause in history—graphs often provide clues that a detailed narrative can investigate and explicate. For instance, Alfred D. Chandler, Jr., the doyen of business historians, saw a pattern of revolutionary industrial development in a time-series plot of iron and anthracite-coal production in northeastern Pennsylvania during the middle decades of the nineteenth century. Driven by the scent of truth, he then wrote a seminal article, in 1971, to document and narrate the Industrial Revolution in America. Fernand Braudel, the distinguished representative of the Annales school of French historians, employed no fewer than sixty-eight figures in his two-volume history of the Mediterranean, most of which are ingenious visual displays of quantitative information. It is notable that reviewers of these highly praised volumes rarely mention the role graphics play in the Braudel style.

Historians have used the visual display of quantitative information but, with notable exceptions, not often or well enough. Historians writing about technology are especially at fault, for their sources—patents, laboratory notebooks, and technical articles—are treasure troves of the visual thoughts of their authors. Anyone who has attended a conference of engineers or of managers of high technology has found visual aids omnipresent. These professionals think and communicate graphically, and business and engineering archives abound with superb examples of their use of the art.

If Tufte publishes more on graphics, he might turn to these sources, taking, for example, the papers of Samuel Insull (1859–1938), the early Chicago utility magnate, who learned to think visually when he was Thomas Edison’s personal secretary. Insull’s speeches are rich with accompanying visual displays such as a multiple-variable plot relating output of electricity per capita, income per unit of output, and utilization of capacity for the world’s leading electrical utilities.

The failure of historians to use such right-side-of-the-brain thinking as Insull’s may be due to ignorance of, and even hostility toward, graphics in a profession that works primarily on word processors. Tufte observes that many people think of data graphics as “devices for showing the obvious to the ignorant.” Others are so biased as to believe that graphics exist only to decorate numbers, a use that Tufte abhors. He shows what he judges “may well be the worst graphic ever to find its way into print”—an example of egregious data decoration in which five colors forming two three-dimensional patterns serve to illustrate, confusingly, just five bits of data.

The sins against graphic integrity identified by Tufte are committed by members of every profession. Simple ideas are often made complex instead of complex ones being clarified. Too many statistical graphics mislead by employing too few variables, showing, for instance, a single variable over time and thus implying that time is the causal factor. Graphs should have multiple variables showing relationships, including causal ones. They should not take numbers out of context, as is often done by truncation.

Tufte gives us good examples of what he does not like. Using an analogy with what the architects Robert Venturi, Denise Scott Brown, and Steven Izenour call architectural “ducks”—that is, constructed decoration in contrast to decorated construction—Tufte singles out some graphic “ducks” for disapprobation. With their emphasis on style rather than information, graphical ducks are like that classic architectural eccentricity, the poultry store designed to resemble a giant duck. Tufte also discourses on “chartjunk,” which includes moiré vibration, unnecessarily heavy grid lines on a graph, and superfluous grid ticks (and extraneous frames) or borders. The junk clutters the graphic and lowers the economy of communication.

Those who decline to use or pay attention to information conveyed “by means of the combined use of points, lines, a coordinate system, numbers, symbols, words, shading, and color” obviously overlook a rich palette of ways of communicating. In his pithy and witty book, Tufte eases the transition from a word-bound world to a more variegated one by subtle but supportive bridges of transition. He tells us that insight into graphical design follows from theories about excellence in prose; he insists that graphics must be edited using techniques much like those in traditional editing—ink that fails effectively to convey information must, like useless words, be pruned. Revisions of graphics, like critical revisions of text, draw out the metal from the dross. And he reminds us that a graphic is a paragraph of information comparable to a paragraph of words, and that it should be chosen appropriately and intermixed sensitively with text.

Often inspired in his selection of examples to make his points, Tufte uses one of Leonardo da Vinci’s wondrous notebook pages to show how the imaginative thinker will move astutely back and forth between the verbal and the visual to best convey ideas. Leonardo avoided the practice of modern authors who merely illustrate a text after having written it; he did not even separate text and illustration, as do some of the most prestigious book publishers of today. With an innate and refined sense of appropriateness, Leonardo thought visually or verbally according to circumstances and subject.

As should we all.

Thomas P. Hughes is a professor of the history of science and technology at the University of Pennsylvania.