Borders

Why do some borders follow rivers and mountains while others run in straight lines?

Unfold a world map and two handwritings are visible at once. Some borders wander — they follow a river's bends, or trace the crest of a range, and no two stretches look alike. Others run dead straight for hundreds of miles, indifferent to everything beneath them.

The tempting reading is that the wandering ones are natural and the straight ones artificial. That is not quite it, and the "not quite" is the interesting part. A river is not a boundary until two parties agree it is one. So the question is not what the land offers, but what a line has to do for the people drawing it — and who those people are.

A border has to be agreed — both sides pointing at the same thing and meaning the same thing — findable on the ground, and durable. A line nobody can locate or hold is not a border, it is a claim.

So suppose you are settling a frontier in the fifteenth century, with no way to fix longitude and no survey party to send. What can you agree on? Only what both parties can see. A river is visible, continuous, named, already in everyone's mental map; a watershed is nearly as good, since water drains this way or that and the divide sorts territory without anyone measuring anything. Such lines are cheap to agree, cheap to write into a treaty, cheap to verify. Defence adds a second reason for the ridge: armies move through valleys, and a crest is expensive to cross and cheap to watch. Mountain borders survive not because mountains are boundaries in some deep sense, but because holding them costs less.

Then the technology changes. Once you can fix latitude, and later longitude, a parallel becomes agreeable in exactly the way a river was — both parties can point at it, because both can compute it. And a computed line has an advantage a river never had: you can agree on it without going there. That is when straight borders appear in quantity, and where — across territory the signatories had not walked. The 1785 Land Ordinance laid a survey grid over land the United States was still acquiring. The Berlin Conference of 1884-85 did not itself carve Africa — it fixed the rules by which European powers would recognise each other's claims — but the bilateral treaties that followed drew boundaries by coordinate, on maps, in European capitals.

That explains how each line gets drawn, not why it is still there. A border, once fixed, becomes what everything else is built on — land titles, tax rolls, courts, army postings, the map in the classroom. Moving it means renegotiating all of that, and everyone who would lose has reason to refuse, so the cost of keeping a line stays low while the cost of changing it grows, whatever its merits. When African states became independent, the Organisation of African Unity resolved at Cairo in 1964 to respect the boundaries existing on the day of independence — not because the lines were good, but because reopening them was judged worse. That is the mechanism: not natural fitness, but path dependence.

And a sting in the tail: the "natural" borders are the less stable kind. Rivers meander, and occasionally jump their channel outright, so a boundary agreed on the water can end up with farms and even towns on the wrong side. The Rio Grande did this between El Paso and Ciudad Juarez, and the resulting Chamizal dispute ran for a century before a settlement in 1963. Treaties now anticipate it, fixing the line at the thalweg, the deepest channel, or freezing the coordinates so the border stops following the water at all. A meridian never moves. It may just run through the middle of a people.

The refinement that matters is that "natural" versus "artificial" is the wrong axis. Both kinds of line are agreements; they differ in what made agreement cheap at the moment of drawing. Visibility made rivers and ridges cheap when nobody could survey; coordinates made straight lines cheap once anyone could compute. Either way the line records the state of the negotiation, not the shape of the land.

Nor is it true that river borders sit comfortably with settlement while straight ones cut through it. River valleys are where people concentrate, so a river frequently divides a single community rather than separating two — the Rhine, the Mekong, the Ganges plain.

One honest caution. That colonial-era boundaries have lasting effects is well supported, including work matching ethnic groups against the lines that split them; how much of any given country's present difficulty is attributable to its borders, as against colonial administration, extraction, or later politics, is argued rather than settled. Path dependence explains why the lines persist. It does not by itself measure what they cost.

• How maritime boundaries are drawn, where there is no terrain to point at and no settlement to cut.
• Why enclaves and exclaves survive at all, given how expensive they are to administer.

Every term the collection defines is gathered in the glossary.