Looking back over 30 years of making charts and maps for The Economist
If you attend a conference about data visualisation in 2018 you will probably hear at least one speaker refer to William Playfair’s early line charts of prices and wages published in 1786. Failing that, you can count on someone to refer to Florence Nightingale’s rose diagram of battlefield sanitary conditions or Charles Minard’s depiction of Napoleon’s disastrous losses during the Russian campaign of 1812.
Conversation rapidly turns from these early examples to the present-day tools: D3, R and mobile scrollytelling. But skipping from historical examples to the present day means there is little consideration of the time spent and tools used in making charts and maps before computers came along.
Jump forward 200 years
I joined The Economist newspaper in 1987 as a cartographer from Chevron Oil, having earlier learned my craft as a civilian cartographer at Britain’s Ministry of Defence. Back in the 1980s we weren’t called data visualisers. We were a rag-tag collection of classically trained cartographers, graphic designers, typographers and technical draughtsmen and women. We knew where to place labels, we cared about small details and we could draw.
Before computers, creating charts was a lot more like art than data science. A mouse was a small rodent and a McIntosh was a popular variety of apple. The charts we produced were limited to the sort of thing you might get in a school textbook. A single, simple measure was plotted over time as a line chart or a column chart. It gave context, broke up the monotony of a page and added credibility. Occasionally we would take a deep breath and generate semi-circular pie charts to represent the political landscape of a new government.
But even these seemingly simple charts had their challenges and took a lot of time to make. Data were found in books by a research department skilled in the art of extracting obscure economic figures and statistics, which were copied to scraps of paper. We would use rulers, dividers, protractors and geometry (Thales’s theorem) to divide axis lines into equal parts to draw the scale ticks. We would plot the data manually in pencil on a special drawing board and sketch out the wording and title for approval before we inked the whole thing in. Text was added last using stencilling, or later, Letraset dry-transfer lettering. Making a spelling mistake was distressing. Areas were filled with sticky-back plastic pre-printed film cut out with a scalpel.
Maps were far more time consuming and required a trip to the local map shop to buy a suitable starting point. The map was placed in a Grant projector, a huge machine that projected a scaled image onto a glass plate. Using it involved putting one’s head under a shroud like an early photographer. Details were copied onto semi-transparent tracing paper from the projected image and later transferred onto the drawing board by means of carbon paper. Again, all the labels would be added in pencil for approval before the time-consuming (and virtually irreversible) inking took place. No pressure! I recall that country names have a disproportionate number of “A” characters; certainly more than they provided for on the Letraset sheets. Finding sheets with enough “A’s” on deadline was sometimes frustrating. Here is the evidence, in fact:
All this carefully constructed artwork was drawn at twice final size. We didn’t use colour, so differences in tone or line styles were limited. Lines would be pecked (dashed) by scraping away the ink with a scalpel. A white correcting fluid, Tipp-ex, that dried fast and could be overwritten was our friend. The Tipp-ex applicator brush was invariably broken or splayed. With the inevitable mistakes and amendments the artwork could look like a teenager’s homework by the end of the night. Fortunately the camera does lie. All camera-ready artwork would be scaled down to half size and projected onto bromide paper for a clean, crisp image that could be pasted — literally — onto the page layout. The paste, or Cow Gum, with its uniquely addictive smell, would allow repositioning without slipping.
As technology crept into the newsroom in the late 1980s, we began to create artwork that had additional layers (masks). These layers were given colour, when available, or shaded areas such as the sea on maps and chart backgrounds. Each layer’s registration had to be adjusted to fit the base as the camera scaling and distortions were never 100% accurate.
Eventually Steve Jobs and Apple came to our rescue from all this manual crafting. A Macintosh Plus (spelt differently from the apple variety for legal reasons) cost around a quarter of a cartographer’s salary at the time. It was quite an investment. The pressure was on to make it productive. MacPaint was hardly the Adobe Illustrator of today. At first we used it to escape from the tyranny of Letraset’s lack of “A”s, by using the cute little machine in the corner as a typesetter. We would key in all our text and titles and print them onto adhesive labels. Our spelling deteriorated almost overnight as the pressure of making mistakes was removed. Or was it just keyboard incompetence? Before long we had figured out more uses and the Apple Mac became a firm favourite. Adobe Illustrator ’88 changed our lives. The Rotring pen became a thing of the past.
Those of us who survived those heady days are now looking at multiple screens full of code, and data files that software can render as a world map in seconds. You want rivers with that? And railways? No problem. My Tipp-ex has long since dried up. Now we worry about how to squeeze what looks great on a big screen onto a smartphone. Time, however, seems to be in just as short supply.
Graham Douglas is a data visualisation specialist at The Economist.