A good map is geographically accurate and to scale, right?
Not always. Transit maps are one exception. They are intentionally distorted in order to be information dense, yet clean, spacious, and organised.
Many of the design decisions that go into a transit map also apply to scientific figures. There’s a lot for us scientists to learn from a careful look at transit maps.
A typical transit map: Singapore
Singapore’s transit map is typical of many around the world. Each rail line is a bright colour. Stations are evenly spaced throughout the map. Most lines are vertical, horizontal, or at a 45° angle (with the exception of the circle route). Overall it’s a tidy depiction compared to the mess of lines in the geographically accurate diagram:
Transit maps and scientific figures share the same goals and constraints
Transit maps and scientific figures both aim to condense a lot of detailed information into something easy to grasp conceptually. This is achieved by abstracting away certain details like geographical accuracy in order to enhance clarity.
Transit maps might be designed for a brochure or to fit on a wall. In either case the size is fixed once the design is completed—just like a scientific figure in a paper. No zooming in and out. No live updates. No interaction with the user. (I’m ignoring interactive transit map apps which exist, but create as many problems for the designers as they solve.)
Space constraints force the designer or scientist to decide what to put in and leave out, what typeface and font size to use (usually a condensed, sans serif for transit maps), and how to maximise use of the printable area while maintaining ample whitespace.
Transit map designers take their constraints seriously. There’s pride to be had in finding a good solution to this vexing puzzle that has no definitive answer. Some designers attack manually. Others see the problem as something best addressed by computers. Either way, it’s a hard problem. Algorithms that create the skeleton of the transit map and place its labels may take hours to find a half-decent solution. It doesn’t help that designers insist on breaking as few design rules as possible. Non-horizontal text and line-breaking, hyphenated names are to be avoided as much as possible. And call-out boxes (think speech bubbles) are both ugly and kind of cheating.
Scientists should also take these constraints seriously. Don’t, for example, blindly accept the default layout produced by your favourite graphing software. Any struggle to label a scientific figure is nothing compared to what transit map designers routinely face.
Distortion in scientific figures
A defining feature of most transit maps is the use of a fish-eye lens, zooming in on the downtown region where many stops exist and zooming out (but still including) the outer regions with more scattered stops.
For scientific figures, distortion often isn’t necessary. Many scientific concepts are already abstract. Unlike a geographical map, there’s no concrete depiction as a starting point. There’s nothing to distort in the first place. In these cases, designing a figure with transit-map traits can arise naturally. Biochemical pathways, for example:
But what about distortion in cases when there is something tangible or geographic being displayed? In these cases, there’s two options. One is to stick to reality. The other is to stylise to such an extent that the distortion is obvious. For example:
No one will misconstrue this figure as anything but a schematic circulation. By comparison, many maps of Arctic Ocean overlay simplified circulation arrows on geographic maps. The result is an awkward grey area between realistic and stylised in which the arrows may or may not be meaningful.
The good, the bad, and the ugly of transit maps
Colour and geometry. Get those two right and you’re on your way to an attractive transit map or scientific figure.
A designer mapping a transit system with, say, ten rail lines needs a map where all those lines are easily distinguished. A scientist making a line plot might face the same requirement. Distinguishable isn’t too hard; it’s only 10 colours. What’s harder is picking 10 or so colours that look nice together. Here’s two colour schemes that work well, the first being the one used by the Singapore map at the beginning:
By comparison, two less appealing combinations:
Do you agree that the top two schemes are better than the bottom two, but aren’t sure why? If so, this 10-minute read is the best primer I’ve come across on colour use in data visualisation.
Octolinearity is the gold standard in transit maps. You know, octolinearity…? It’s the fancy word for the layout I mentioned earlier using lines that are horizontal, vertical, or at a 45°. It’s not a fixed rule, though. Done well, a 30/60° layout can work, as can multiples of 18° if you really want. Ultimately, what matters is symmetry, consistency, and order.
Straightening anything that doesn’t need to wiggle is the first step. A mess of cables, for example, can be easily tidied up:
Even something as simple as labelling a diagram can benefit from from judicious use of 45° angles:
Constraining yourself to two or three angles can also help when deciding how to stylise something realistic. If I asked you to draw a river or an estuary any way you like, where would you start? What if I asked you the same question but impose the constrain that you must base you design on horizontal and vertical lines?
The bad and the ugly
Over at transitmap.net, graphic designer Cameron Booth maintains a Hall of Shame of bad transit maps. Things to avoid when designing a transit map should also be avoided when designing a scientific figure. I’ve alluded to most already. As a reminder, avoid
- words oriented at various angles
- call-out boxes
- poor colour combinations
- an ugly font (or worse, many)
- a lack of contrast when it’s needed
- a general lack of attention to detail.
Embracing the transit map metaphor
Layout, colour, and labelling all matter, but the heart of a transit map is the metaphor: removing as much information as possible so as to emphasise the origins, journeys, and destinations. Two examples that I’ve come across (from the same scientific paper) that truly embody the metaphor show the pathways of water masses through the global ocean.
The first transit map in mass use was created by Londoner Harry Beck in 1931. As described in Hello World, Beck “concluded that a geographical map was no longer feasible”. His “diagrammatic approach would seem clearer and more logical to passengers even if it risked distorting geographical reality.” This positive trade-off is exactly what these final two figures achieve.