On January 13, 1888, a group of 33 gentlemen founded the National Geographic Society in the Cosmos Club, a select private club in Washington, DC. The group included members of the city’s scientific and intellectual circles and a few prominent explorers. The gathering was vividly portrayed, years later, in a painting by Stanley Meltzoff (below). The round mahogany table can still be found in our headquarters. The central characters in the scene are Major John Wesley Powell, hero of the battle of Shiloh and explorer of the Grand Canyon, and Gardiner G. Hubbard, a prominent Boston lawyer with an aristocratic white beard. Hubbard would become the Society’s first President and his son-in-law, Alexander Graham Bell (better known as the inventor of the telephone) would succeed him in 1897 after his death.
Click on any image in the post for an enlarged view.
Graphics and maps at National Geographic go back to that very first issue. With the concept of “data visualization” becoming a trendy one in recent years, sometimes it’s good to remember that not only the concept is centuries old (going all the way back to Scottish engineer and economist William Playfair, who invented the pie, bar and fever line charts in the late 1700′s). For one thing, “data visualization” in the early days had something that is so often lacking today: clear, simple, intelligible storytelling.
Five color maps illustrated an article entitled “The Great Storm of March 11-14, 1888″. Following them, a remarkably elegant chart shows the changes in atmospheric pressure registered by different vessels and land stations over the duration of the three-day storm, and compares them with the normal value. Technical but still a good chart even by today’s standards:
The maps show pressure, temperature and wind at different moments during the storm:
In addition to numerous maps (which will be featured in future posts) black and white charts continued to appear in National Geographic at the turn of the century. A March 1893 article entitled “The movements of our population”, by then editor Henry Gannet, marks the first extensive use of charts in the magazine. It was not uncommon for an article to take the entire issue of the magazine, as shown in this restrained cover:
The story featured multiple charts and choropleth maps, starting with this comparison of urban and rural population at each census since the first American census, which took place in 1790 (the country’s population was then enumerated to be 3,929,214).
This very original graph shows the evolution of the U.S. population and its ethnic breakdown from the first census to the latest at the time (1890). While “native” and “colored” population increases steadily, immigrant numbers grow exponentially, with a clear majority of German and Irish newcomers. It’s a beautiful, clean design.
Another graph shows a breakdown of the population of large cities, this time less successfully. The heavy, dense grid and the patterns chosen make the chart appear to move from right to left. The perversely literal gradient moves from “white – native-born of native parents” (in white) to “white – native-born of foreign parents”, then “white – foreign-born” and finally “colored” (in black).
The next chart (below) looks at the rate of increase of white and colored populations, a smart analysis and the best way to compare the growth or decrease of variables with very different absolute numbers. Quite remarkable for something done in 1893. I would have done the same thing today.
The sharp decrease of both groups in the 1860-1870 is no doubt consequence of the terrible losses during the Civil War.
The choropleth map at the bottom show the proportion of African-Americans to the total population, by state.
Another chart in the same story (below) offers an interesting comparison, paired with a graph showing immigration in each decade. Despite the up and downs, even the lowest data point signals a 10 percent increase over the previous decade, as the United States continued to attract immigrants.
Overall, the entire story is a great demographic analysis by Henry Gannet (1846-1914), an American geographer and founding member of the National Geographic Society and the United States Geological Survey. Gannet has been described as the “Father of the Quadrangle”, which is the basis for topographical maps in the United States.
The chart below appears in the May, 1898 issue. Population increase is plotted against different measures of progress. Using both wealth and wealth per capita, and area versus population are indicative of the editors’ efforts to provide good context to the story of population growth. However, the methodology or type of units in the y-axis are not explained. The series appears to have been indexed to show the percentage change since the first data point (the only way such different variables could be compared) but it’s unclear how it all is supposed to work since the different lines don’t always start at the same baseline.
Another interesting chart appears in a May 1900 article about the most recent Cuban census data. The first chart is a nice percent bar. Some of the other follow an odd spacial distribution (which doesn’t make them wrong).
The wavy chart below is my favorite from the early days of National Geographic. A January, 1905 article on immigration included this beautiful work showing the immigrants coming to the United Stated every year from 1820 to 1904, by country of origin. Over a thousand data points are represented visually, and all numbers are included as well. The work is credited to F. P. Sargent, the Commissioner General of Immigration, but unfortunately the draftsman producing the chart is not mentioned.
Looking at old Nat Geo issues is always a source of inspiration. We had the immigration graph in mind when we used the same visual approach for this 2007 chart showing sales of different music recording formats over time.
I think our immigration chart was very likely inspired by the work of Francis Amasa Walker (1840-1897) who was the Chief of the Bureau of Statistic in the 1870′s and created the Statistical Atlas of the United States (1874), a remarkable visual work based on that census. The statistician used beautiful, elegant charts. The one in the middle in the example below will surely remind you of ours.
Another story in the same issue, entitled “The character of our immigration” by Z. F. McSweeny, shows the characteristic heavy grid of most charts at the turn of the century. It’s a pity because they are otherwise beautiful, elegant and clear.
The story ends with this little oddity about inmates of different ethnic origins. I love the hand-lettering and the curious grouping of racial divisions.
The samples above are just a quick, random sampling from the early 1900′s and even earlier. Today, our access to new technology and formidable amounts of information has spurred the many forms of the too prolific “data viz” field. It’s good that such experimentation continues. But do a quick Google Images search for “data visualization” and you’ll quickly see the state of affairs. Clarity in the transmission of information, which is the essential mandate in journalism, is not a priority for so many of today’s data visualization designers. A simple bar chart is just not cool enough. But if you think of it, since William Playfair first plotted values on two axis with ink and paper there is really not a lot that has been invented in terms of accurately displaying numerical information with precision, clarity and familiarity (which breeds understanding) for readers.