Harry Beck was an engineering draftsman, not a cartographer. In 1931 he submitted a redesign of the London Underground map to the London Passenger Transport Board, and it was initially rejected as too radical a departure from geographic convention.
The previous maps had shown the tube lines overlaid on a roughly accurate rendering of London’s surface geography. Streets were recognizable. Distances were proportional. Someone with knowledge of the city could look at the map and locate themselves in relation to actual neighborhoods.
Beck’s map removed all of this. He replaced geographic space with topological space: stations spaced evenly regardless of actual distance, all lines running at 45° or 90° angles, the Thames rendered as a gentle stylized curve. Hammersmith appears geographically close to several central stations; on Beck’s map the distance looks roughly equal to any other interchange. Angel, far north of its underground position, sits placidly in its diagrammatic place. The map does not look like London.
It works because of what underground travelers actually need. Once you descend into the tube, geography becomes irrelevant — you cannot see it, you cannot use it, you cannot navigate by it. What you need to know is the sequence of stops and which lines connect where. Beck’s map preserves exactly this: topological relationships. It discards exactly what would have added noise: geographic accuracy.
The design principle is explicit. Beck chose a constraint — prioritize connection structure over spatial fidelity — and the constraint determined everything else. The resulting diagram is not a simplified map of London. It is a different kind of thing entirely: a map of a specific relationship structure that exists underground, rendered visible by the decision to suppress everything else.
Gerardus Mercator made a different decision in 1569.
Navigation at sea required a specific capability: the ability to plot a course by constant compass bearing. A ship sailing at 45° northeast for three days needs to follow a rhumb line — a path that crosses all meridians at the same angle. On a globe, rhumb lines are curves. Mercator’s projection straightened them out: on his map, constant compass bearings appear as straight lines.
The cost was area. To preserve bearing accuracy, Mercator stretched the poles. Greenland, in reality about 836,000 square miles, appears roughly the same size as Africa, which covers 11.7 million square miles — fourteen times larger. Distortion increases as you move from the equator. Polar regions become enormous, tropical regions accurate.
Mercator’s projection is not deceptive. It is a specific answer to a specific question: what transformation preserves rhumb lines? The mathematics determine the distortion as a consequence. No sleight of hand is involved. But the map encodes a constraint — and the constraint shapes what the map can and cannot do. It is the best navigation map ever designed, and the worst area-comparison map ever designed, and these two facts are the same fact.
The Tissot indicatrix makes this visible. Draw small circles at regular intervals across a globe. Project them using any mapping method. On the Mercator projection, these circles remain circular near the equator and become enormous ellipses near the poles — a direct visual measure of what the projection is doing to the underlying geometry. Every projection has its own indicatrix pattern. Conformality (angle preservation), equivalence (area preservation), equidistance (distance preservation): no flat map can achieve more than one of these simultaneously. The indicatrix shows which one was chosen.
Gottfried Wilhelm Leibniz and Isaac Newton developed calculus independently and essentially simultaneously. Newton’s notation used dots over letters to denote derivatives: ẋ for velocity, ẍ for acceleration. Leibniz used fractions: dx/dt, d²x/dt².
For the century that followed, British mathematicians used Newton’s notation. Continental mathematicians used Leibniz’s. Continental mathematics advanced faster.
The reason is visible in the notation itself. Leibniz’s fractions behave algebraically — you can manipulate them like fractions, and the manipulations are valid. The chain rule, which connects derivatives of composed functions, becomes obvious: dy/dx = (dy/du)(du/dx). The du terms cancel like fraction terms. In Newton’s notation, the chain rule requires a separate statement with no algebraic motivation. The computation is possible; the structure is hidden.
Leibniz’s notation is a map of calculus. It was designed to make certain relationships visible — the algebraic structure of differentiation, the connection between rates and compositions. Newton’s notation was designed for something else: a specific physical intuition about fluxions, quantities in motion. Both notations map the same underlying mathematics. They make different things easy to see.
Continental mathematicians using Leibniz’s map had better topological access to the territory. Not because they were more intelligent. Because the map they carried preserved the relationships that mattered for the work they were doing.
Every map is a map for something. This sounds obvious and most of the time it is. What makes it less obvious is the tendency to treat a map as a neutral representation — a rendering of the territory — when it is always a rendering of a specific projection, chosen by a specific constraint, preserving specific relationships at the cost of others.
Beck’s map preserved connectivity. Mercator’s map preserved compass bearings. Leibniz’s notation preserved algebraic structure. Each one is honest in the sense that the selection it makes is knowable, auditable, visible to someone who knows how to read it. The Tissot indicatrix is the indicatrix for maps in general: the tool that makes selection explicit.
Language works this way too. A vocabulary with a dedicated word for a concept makes that concept easier to handle — easier to hold, reason with, pass from one sentence to the next. Schadenfreude is a convenience but also a cognitive affordance: the concept becomes a unit, and units can be combined, compared, placed in logical relations. The same concept without a word remains available but at higher cost, like navigating with a Mercator projection when what you needed was area accuracy.
This is not a claim that language determines thought or that vocabulary limits what can be perceived. It is a more modest claim about access: the maps we carry shape the paths we can take easily. Better maps do not change the territory. They make specific features of the territory visible, at the cost of making other features harder to see.
Every map embeds its indicatrix. The question worth asking of any map — cartographic, mathematical, linguistic — is not does it accurately represent reality but what constraint does it carry, and what does that constraint make easy to see.