The Birth of Statistical Graphics and Their European Childhood

On the historical development of W.E.B. Du Bois’s graphical narrative of a people

Visual Revelations

In 1786, when the Scot William Playfair published his Commercial and Political Atlas, the clock of scientific progress, which has only two positions, moved to “after” (Berlinski, 1995).

There were graphs before Playfair; Michael Florent van Langren drew a linear plot in 1644 to show the inaccuracies in the measurement of longitude; in 1669, Christiaan Huygens sent a graph of life expectancies to his brother Lodewijk, based on the data in John Graunt’s Bills of Mortality; and the eponymous Robert Plot sent a graphic “History of the Weather” to Martin Lister in 1685 (Wainer, 2005).

But these were all essentially unique events.

None of Playfair’s predecessors used the tool of graphic representation coherently toward a goal of broad communication of quantitative phenomena. And no one before him constructed displays as richly filled with information as did the remarkable Scot, nor designed and executed them as beautifully.

Playfair’s plots were narrowly focused on financial topics—primarily imports and exports, with rare side trips to topics like the national debt and the size of the sinking fund. He also produced one plot comparing the sizes of various countries (both physical size and population), including what continents the countries were in (it was for this aspect that he invented the pie chart) (Playfair, 1801/Wainer, 2007—two publications).

Data and graphs enjoy a symbiotic relationship; without data, there can be no empirical graphs, and without graphs, the broad dissemination of information is hampered by the usually insuperably difficult task of extracting information from tables. There was an explosion of data collection in the 19th century and a concomitant increase in the production and dissemination of graphics, but the topic of the data, and hence the graphs reporting them, differed by location.

In France, there was an abundance of crime data, which, in 1829, led André-Michel Guerry, a young French lawyer working in the Ministry of Justice, and the Venetian geographer Adriano Balbi to produce the first statistical map of crime data showing crimes against people and property. They also included a literacy measure to suggest a causal connection. Four years later, Guerry extended this to include other social variables. Thus began social explanations of criminal behavior.

Across the Channel, the English lawyer Joseph Fletcher plotted data pointed toward social welfare rather than crime; his plots of the frequency of such variables as ignorance, bastardy, and improvident marriages influenced both society and social policy (Cook and Wainer, 2016; Friendly, 2007; Friendly, Valero-Mora, and Ularqui, 2010; Friendly and Wainer, in preparation; Wainer, 1997).

Arguably the most-important outcome of graphing data at that time was associated with medical information. The data for what we would now call epidemiological studies grew directly from the 1836 Act of Parliament that established the General Registration Office, under whose mandate data were gathered on births and deaths for the entire population of England. These data were used serially by two young physicians, William Farr (1807–1883) and John Snow (1813–1858), to try to track down the cause of cholera and thence to mitigate its spread.

Farr assumed that the disease spread through the air and developed a theory of susceptibility to the disease based upon the elevation above the Thames at which people lived. It proved incorrect. Snow felt that the disease was water-borne, and his map showing the location of cholera deaths, as well as the placement of the various water pumps, was so spectacularly successful that it has achieved iconic status as marking the beginning of modern epidemiology (Friendy and Wainer, in preparation; Wainer, 2014).

Graphs Cross the Atlantic

The use of statistical graphs was not limited to Europe. Thomas Jefferson became aware of them during his years in Paris and carried the idea with him when he returned to America. His bar chart showing the availability of vegetables in the Washington market is well-known and parallels Joseph Priestley’s equally well-known 1765 “Chart of History.”

The use of graphics in the United States gained an invaluable proponent when, in 1869, Francis Amasa Walker (1840–1897) was appointed chief of the Bureau of Statistics of the United States and thus superintendent of the 1870 census. This particular census has two important characteristics that set it apart from all that preceded it: (i) Coming, as it did, just five years after the Civil War, it was the first that was required to fully count the newly emancipated African-American citizens, and (ii) its results were published in an award-winning Statistical Atlas that, for the first time, visualized the census data.

After a law was passed that allowed him to appoint and train census enumerators who would be free from political influence, Walker accepted the appointment as superintendent of the 1880 census. One result of that census was a substantial increase in the population of southern states due, as it turned out, to an undercount of some sub-populations in the 1870 census.

Walker emphasized the importance of having trained professional census staff so the enumeration could be done completely and the results disseminated accurately, unaffected by political considerations. This led to the establishment of the Bureau of the Census in 1902.

At the beginning of the 20th century, Paris was arguably one of the most-important locations in the entire western world for both the arts and the sciences. The Sorbonne, in the fifth arrondissement, was the site of the fourth International Congress of Mathematicians, attended by more than 200 of the world’s leading mathematicians. The iconic German mathematician David Hilbert laid out 10 of his famous 23 unsolved “mathematical problems” on August 8, 1900, and set career goals for all aspiring 20th-century mathematicians.

However, as hard as it might be to believe, that was not the biggest event hosted by Paris at that time.

W.E.B. Du Bois’s Statistical Graphs and Their Audience of 50 Million

The Exposition Universalle was a World’s Fair held in Paris that ran from April 14–November 12, 1900. Its dual goals were to celebrate achievements of the past century and encourage future developments. During those seven months, nearly 50 million people visited it. Among the innovations displayed were the diesel engine (running on peanut oil), escalators, talking films, and the largest refracting telescope ever constructed.

Visitors were delighted with a 100 m Ferris wheel, an enlargement of the original 80 m wheel that George Ferris debuted at Chicago’s Columbian Exposition seven years earlier. It also included “The Exhibit of American Negroes,” organized by Booker T. Washington and W.E.B. Du Bois, whose stated goal was to show African Americans’ positive contributions to American society.

The exhibit included 400 patents by African Americans, as well as 200 books by African-American authors. There were also a large number of facts about African-Americans, woven into a narrative by transforming them into 58 statistical graphics.

These graphs spanned a wide range of characteristics of African Americans and their lives, but much of Du Bois’s story would have been more difficult and less accurate without the 1870 expansion of the U.S. Census to include African Americans (and its more-rigorous enumerations in 1880 and 1890). Francis Walker’s work provided Du Bois with the grist from which to mill his hand-drawn displays.

Du Bois began his story with a graph dramatically showing the profound effect of Lincoln’s Emancipation Proclamation (Figure 1). It shows the balance of enslaved versus free African Americans from 1790 until 1870. From this, we learn that the proportion of free African Americans during slavery ranges around 12% of the African-American population until January 1863, when emancipation was the key to the dramatic change shown.

Figure 1. Proportion of slaves and free Negroes from 1790–1870.

Having established the status of African Americans within the United States, he then used a simple bar chart (Figure 2) to show both that there were 7.5 million African Americans in 1890 and the exponential growth of that population over the previous 150 years.

Figure 2. Increase of the Negro population of the United States from 1750–1890.

To put the size of the U.S. population of African Americans at that time into a global context, Du Bois presented a plot showing the outlines of 10 other countries in which their size was proportional to their total population; in the center was a map of the United States, whose size was proportional to the African-American population (Figure 3). Study of this plot quickly reveals the number of African-Americans as greater than the entire populations of Australia, Norway, Sweden, Holland, Belgium, Switzerland, and Bayern.

Figure 3. Negro population of the United States with the total population of other countries.

Du Bois reused this same kind of graphical form to compare the size of the African-American population with the total population of the U.S. (Figure 4), augmenting the plot with proportional figures, showing that the African-American portion of the U.S. population had been diminishing over time.

Figure 4. Proportion of Negroes in the total population of the United States.

He makes the differential rates of growth explicit in a well-designed line plot (Figure 5), showing that even though the African-American population had been growing exponentially, so too was the rest of the U.S., and the coefficients of the latter’s growth were larger.

Figure 5. Comparative rate of increase of the White and Negro elements of the population of the United States.

Du Bois was not done with making international comparisons. He next used a unilateral version of what has come to be called a “population pyramid,” in which he compared the age distribution of African Americans in Georgia to that of the French (Figure 6), in which we immediately see that the Georgia African-American population is much younger.

Figure 6. Age distribution of Georgia Negroes compared with France.

This figure provides a mystery: Why use data from Georgia, rather than from the entire United States? The national data were available to him (Figure 28 of Gannett’s Statistical Atlas).

1. Was it because almost half of Georgia’s population was African American? Unlikely; if that were a motivation, he would have chosen South Carolina, where 60% of the population was African American. Of all the states, it did have the largest number of African Americans, and many states had few or none (see Figure 7).

Figure 7. Distribution of Negroes in the United States in 1890.

2. He lived in Georgia and almost surely had a great deal of information about the state available to him, some of which might not have been easy to get for the entire nation (Figure 9).

3. Atlanta, in particular, was a place where African Americans had an established middle and upper-middle class, including a successful business and higher education community; Atlanta and Georgia would have been a place where there were not only a lot of black citizens, but ones who had climbed the socioeconomic ladder—despite the Jim Crow laws and the endemic racism.

4. People like Alonzo Herndon were there—someone who was already a millionaire at that time and the owner of an influential company, who subsidized Du Bois’s work.

Du Bois abandoned both Georgia and the dual-bar chart approach in favor of the more modern-age pyramid when he showed the marital status by age for African Americans nationwide (Figure 8—an only slightly modified version of Figure 136 in the Census Statistical Atlas). In it, he shows that African-American women seemed to marry about five years younger than men, who, on the other hand, seemed to die much sooner.

Figure 8. Conjugal condition of American Negroes according to age periods.

Gathering the information to construct Figure 6 for the entire nation would not, in hindsight, be difficult, since it contains data on two variables (age, race) that Du Bois surely knew were gathered by the census. Other sorts of variables—for example, those related to wealth—might have been harder for Du Bois to obtain, so he was forced to rely on state data (Figure 9) that Georgia provided (I assume that Georgia might have gathered such data associated with some sort of taxing scheme), showing the status of one measure of the wealth of African Americans over a 25-year period and how it had changed. In his “snail-chart,” he found an imaginative way to show bars of very different lengths on a single page.

Figure 9. Assessed value of household and kitchen furniture owned by Georgia Negroes.

Coda

William Edward Burghardt (W.E.B.) Du Bois was born in Great Barrington, Massachusetts, on February 23,1868, and died in Accra, Ghana, on August 27, 1963.

During the course of his 95 years, Du Bois accumulated an extraordinary list of accomplishments. He earned a PhD from Harvard—the first African American to do so. He was a sociologist, historian, civil rights activist, and author. Among his better-known books were The Souls of Black Folk, The Suppression of the African Slave Trade, Dusk of Dawn, Darkwater, and The Talented Tenth.

After examining his hand-drawn data displays, we should certainly add that he was also a talented data scientist, who contributed to the development of the modern approach to constructing a narrative argument based on evidence and using a series of graphic displays. In this way, he is a direct lineal descendant of Playfair, Guerry, Farr, Snow, Fletcher, and Walker. The nine graphs reproduced here are but a 15% sample of all of those he prepared for the 1890 Exposition.

To see the entire corpus of Du Bois’s Paris graphs, go to the Library of Congress website.

Du Bois’s plots lacked the breadth, richness, and beauty of Playfair’s—but so did everyone else’s. Walker’s graphical work was an immense improvement over the endless tables that were the output of Census before him and Du Bois’s plots more then held their own in that fast company.
The notion of using a sequence of graphs to show the evolution of a complex statistical trend did not originate with Du Bois; the great Minard showed the effect of the north’s blockade of southern ports had on the cotton trade with Europe, using a series of the sort of flow maps that have since become his signature.

But such a topic, doubtless of great interest to economists and historians, pales in comparison to the task that Du Bois had shouldered. He wanted to show the story of the emergence of an entire people from the yoke of slavery; how they grew in number, in wealth, in influence. Du Bois’s contribution was not in the quality of his plots, but in his unified vision of the grand story he told: the story of a modern Exodus.

Du Bois had something that Moses lacked, though: data. The straw for the bricks out of which Du Bois constructed his story were the data from the 1870 census and beyond. The impressive contributions that were highlighted in the 1900 Paris Exposition are magnified substantially if we remind ourselves that until 1863, 89% of African Americans were enslaved. Thus, all of the accomplishments described took place in only 37 years and the starting point for those contributions included poverty and illiteracy for that 89%.

Last, instead of some sort of Marshall Plan to jump-start the lives of former slaves, they were faced with racism and Jim Crow. The story told in Paris was a remarkable beginning.

Further Reading

Berlinski, D. 1995. A Tour of the Calculus. New York: Pantheon Books.

Cook, R., and Wainer, H. 2016. Joseph Fletcher, thematic maps, slavery and the worst places to live in the UK and the US, in Visible Numbers, the History of Statistical Graphics, C. Kostelnick and M. Kimball (eds.), pp. 83–106. Farnham, UK: Ashgate Publishing.

Friendly, M. 2007. A.-M. Guerry’s Moral Statistics of France: Challenges for multivariable spatial analysis (PDF download). Statistical Science, 22(3), pp. 368–399.

Friendly, M., Valero-Mora, P., and Ulargui, J.I. 2010. The first (known) statistical graph: Michael Florent van Langren and the “Secret” of Longitude (PDF download). The American Statistician 64(2), pp. 185–191.

Friendly, M., and Wainer, H. (in preparation). The Origin of Graphic Species. Cambridge: Harvard University Press.

Gannett, H. 1898. Statistical Atlas of the United States: Based on the Results of the 11th Census. Washington, D.C.: Government Printing Office.

Munroe, J.P. 1923. A Life of Francis Amasa Walker. New York, N.Y.: Henry Holt & Company.

Playfair, W. 1801/2007. The Commercial and Political Atlas, Representing, by means of Stained Copper-Plate Charts, The Progress of the Commerce, Revenues, Expenditure, and Debts of England, during the whole of the Eighteenth Century (edited and introduced by Howard Wainer and Ian Spence). New York: Cambridge University Press.

Playfair, W. 1801/2007. The Statistical Breviary; Shewing on a Principle entirely new, the resources of every state and kingdom in Europe; illustrated with Stained Copper-Plate Charts, representing the physical powers of each distinct nation with ease and perspicuity< by William Playfair (edited and introduced by Howard Wainer and Ian Spence). New York: Cambridge University Press.

Wainer, H. 1997. Tom’s veggies and the American way. CHANCE 10(3), pp. 40–42.

Wainer, H. 2005. Graphic Discovery: A Trout in the Milk and Other Visual Adventures. Princeton, NJ: Princeton University Press.

Wainer, H. 2012. Moral statistics and the Thematic Maps of Joseph Fletcher. CHANCE 25(1), pp. 43–47.

Wainer, H. 2014. Medical Illuminations: Using Evidence, Visualization & Statistical thinking to Improve Healthcare. London: Oxford University Press.

About the Author

Howard Wainer is a statistician and author. His most recent book, Truth or Truthiness: Distinguishing Fact from Fiction by Learning to Think like a Data Scientist, was published by Cambridge University Press and was named by the Financial Times of London to its list of “Top six books of 2016.” He is currently collaborating with Michael Friendly on a history of graphics, On the Origin of Graphic Species, which will be published by Harvard University Press. He has written this column in CHANCE continuously since 1990.

howard wainer
Visual Revelations covers many topics, but generally focuses on two principal themes: graphical display and history. Howard Wainer, column editor, encourages using this column as an outlet for popular statistical discourse. If you have questions or comments about the column, contact Wainer at hwainer@nbme.org.

Back to Top

Tagged as: , ,

Leave a Response

Please note: comment moderation is enabled and may delay your comment. There is no need to resubmit your comment.