January 6, 2017
Discovering Charles Minard: Information Design, Numerical Magnitudes, and a New Understanding of an Old Technique
January 6, 2017
The collections of the Map Division of The New York Public Library contain a rare gem of information graphics and cartographic design: a copy of Charles Minard's self-published folio of 1861, Des tableaux graphiques et des cartes figuratives. WorldCat lists only three other copies of this work in the world's libraries; the copy in the NYPL is the only one in North America.
I had the opportunity to see this folio while visiting the Map Room on a Short-Term Research Fellowship in September. The folio consists of an introductory text printed in French (a translation of which was made by Dawn Finley and posted online), followed by four graphic plates with reprints of Minard's drawings: two graphs (plates I and III), one map (Plate II), and one compilation of other authors' classic statistical data techniques redrawn. On the front cover is an inscription to John Bigelow and signed (with initials) by Minard.
The folio was a gift to Bigelow while he was serving as American Consul in Paris in the 1860s, subsequently making its way to the collections of the NYPL along with Bigelow's many books and papers. Hoping to learn more about Bigelow's interest in Minard, I spent a few days looking through Bigelow's appointment books and diaries in the archives, to see if I could find any reference to appointments or meetings with Minard (or engineers generally), but found nothing.
Geography students, statisticians, graphic designers, and afficionados of the works of Edward Tufte will recognize Charles Minard as the author of the famous Carte Figurative illustrating Napoleon's 1812 march to Moscow, drawn and published in 1869, the year before Minard's death in Bordeaux at the age of 89. That map, considered by some to be a masterpiece of information design, came at the end of over two decades of flow line map productivity, during which time Minard expanded the language of flow line technique to new levels of statistical expression.
The first flow line map that he drew, the map that launched those years of innovation, appears as Plate II in this folio.
Like the other plates in the publication, the map is a hand-colored lithograph. Minard originally drew it in 1845, to depict the movement of passengers by public conveyances between Dijon and Mulhouse. The map is designed with a simplified base of city and town locations and rivers, as a supporting structure for the "zones," or flow lines, which are offset in blue watercolor fill. Minard uses curving routes to approximate the actual routes in a general way, what we would today call a "realistic" (as opposed to "abstract") flow line technique. Cities and towns anchoring each segment are sketched in with a circle sized to give a general sense of relative town sizes.
No numerical figures are included with the flow lines, as Minard (inspired by the works and writings of William Playfair and Alexander von Humboldt) preferred the reader observe magnitude with eyes first and measure later. "It is by sight alone," wrote Minard, "that this map, which was found to be eloquent, made visible the relationship between the numbers of travelers, because it will be noticed that it does not carry a single numeral" (Minard 1861:3).
I was excited to spend time with this historic work, as it has many qualities of interest to scholars of the history of information graphics. For example, in addition to documenting his early preference for realistic, curving line routes, the map also displays his preference for a highly stripped back base geography, an aesthetic he would carry into most of his subsequent work. As I explored the map more closely, however, I suddenly perceived another design feature which I had never before noticed, and that was the way Minard defined the numerical magnitudes of his flow lines.
In his map, Minard provided two bar scales at the lower right. One is a conventional bar scale intended for the reader to measure geographical distances, with increments in kilometers. Below it, however, is a second bar scale with increments marked in 10,000 passengers, intended for the readers to measure the widths of the flow lines. In essence, that is, to measure passenger data distance. Intrigued, I revisited Minard's other, later flow line maps (many of which can be found online at the Bibliothèque Numérique Patrimoine des Ponts), where I found the same technique repeated.
For me, this was a radically different way of conceptualizing flow line technique, one that suddenly opened my eyes to mapping habits so ingrained in me as to be invisible. Until now, I've thought of flow lines as individual symbols whose widths are proportionally sized to match the data they represent. Minard showed me that, in fact, the flow line map consists of two overlapping planes of different data spaces: physical geographical space and passenger space. Though they cannot occupy the same spatial plane (in the same way as, for example, miles and kilometers both occupy the same plane of physical geographical space), they do intersect at the locations where passengers' journeys begin and end.
By the same logic, a map that depicts data with both flow lines and proportional circles would be three overlapping planes, with three different distance scales, which intersect only at the locations where flow lines begin and end, and at the center points of the proportional circles if those center points are positioned at the physical locations of the data they represent.
If I were to tell someone that I was making one map depicting three different data spaces, each occupying separate spatial planes which only occasionally intersected, the person might think I was making something unconventional, impractical, perhaps unusable. And yet, this is simply a different way of looking at what I've been making all this time, when I make a flow line or proportional circle map. Indeed, it is a very basic and obvious truth about statistical cartography; nevertheless it is one that I never saw until Minard gave me a device to measure it.
Years after his death, Minard's students would continue and then extend the visual grammar of their mentor's flow line technique in the maps of their masterful, annual Album de Statistique Graphique. I'm disappointed to observe, however, that they did not also continue his practice of indicating the measurability of those flow lines with a bar scale. Instead, they explain with words the relationship of the flow line widths to their corresponding values. It is a small difference. But explaining to a reader how to use a ruler to measure and convert the values of individual lines does not encourage that person to perceive that those lines coinhabit a data space forming its own plane. Our sense of the map as a representation of multiple data space planes, the contemplation of which will lead to creative insights about how those data relate, is gone.
For my part, I'll never look at or make a statistical map the same way again. The second bar scale in this reprinted map from the nineteenth century was, for me, a revelation, for which I am grateful to Monsieur Minard.
Margaret Pearce is a cartographer based in Rockland, Maine. She is writing the entry on "Thematic Mapping Techniques" for History of Cartography, volume 5: Cartography in the Nineteenth Century, edited by Roger Kain and forthcoming from University of Chicago Press.