Where a projection is chosen before the ground is known

The series comes first

A national mapping programme does not choose a projection sheet by sheet. It chooses once, for the whole country, and then divides the chosen surface into tiles. The Universal Transverse Mercator system, used by military and civilian agencies across much of the world, divides the globe into sixty six-degree-wide zones, each projected on a cylinder wrapped around a narrow strip of longitude. Within a zone the distortion is kept small — scale error at the central meridian is held to 0.9996, and it rises to around one part in a thousand at the zone edges. For a sheet covering the middle of a zone, that is a reasonable bargain. For a sheet that happens to fall at the boundary, the ground arrives already carrying inherited distortion before a single benchmark is visited.

The projection is not wrong. It is simply indifferent to what the ground turns out to contain. A zone boundary drawn to balance distortion across thousands of kilometres of ocean and farmland may run straight through an urban area, a river delta, or a mountain range — exactly the terrain where precise angular and distance relationships matter most to the people actually using the sheet.

What the edges cost

At the edge of a UTM zone, the scale factor departs from unity by roughly one part in a thousand. That sounds modest, and across most applications it is. But one part in a thousand over ten kilometres is ten metres — enough to matter to a surveyor closing a traverse, to a planner calculating setbacks, or to anyone trying to reconcile a field measurement with a coordinate on the sheet. The distortion is also not uniform in direction: the Mercator cylinder preserves angles locally but stretches distances away from the central meridian, so a line running east–west accumulates more error than one running north–south over the same distance.

The seam between two zones is the hardest case. A feature that straddles a zone boundary — a river, a road, a cadastral boundary — must be represented in two different coordinate systems, and the two representations do not simply join. A surveyor stitching work from both sheets has to reproject one of them, and reprojection is not lossless. Small mismatches propagate into precisely the documents — subdivision plans, infrastructure alignments, flood-risk models — where mismatches are least welcome.

Indifference as policy

The decision to fix a projection for an entire series is not negligence; it is the only practical option for a programme that must maintain consistent, joinable coverage. Changing the projection mid-series would produce a seam of a different kind — one internal to the mapping, where the geometry shifts without any physical boundary to explain it. Series coherence is the goal, and distortion at the edges is the price paid to achieve it.

What a user of the sheet needs to know is which part of the tile they are working on. The centre is well-behaved. The corners, particularly in the east–west direction, carry the cost of the bargain struck for the series as a whole. Large-scale engineering surveys in those zones often apply a grid-to-ground correction — a local scale factor — to recover the real-world distances the sheet coordinates imply but do not quite render. That correction is not printed on the sheet. It has to be calculated, case by case, from a projection formula that was settled long before anyone arrived to measure the ground.

The sheet is a compromise. The compromise was made at the series level, in the abstract, for good reason. Knowing which edge of it you are working on is the first step to working with it honestly.