Triangulation

Part 1

One eye is not enough

Suppose you are standing on a shore, and somewhere out on the water there is a ship. You happen to carry a wonderful instrument — a theodolite, essentially a telescope bolted to a protractor — which tells you the direction of anything you point it at, down to a fraction of a degree.

What it absolutely cannot tell you is how far away the ship is.

A single angle measurement pins the ship not to a point, but to a ray. Every ship along that ray looks identical from A: same bearing, same reading, no way to tell them apart. The one thing we want — distance — is precisely the thing this measurement is blind to.

The whole game of triangulation is to recover distance while measuring nothing but angles. And the trick is almost embarrassingly simple.

Part 2

The second viewpoint

Walk along the shore. Not toward the ship — you can’t — just sideways, some distance you can measure with a tape or by pacing. Call it the baseline, \(L\). Now measure the direction of the ship again, from the new spot.

You now own two rays. Two distinct, non-parallel rays cross at exactly one point — and the ship has nowhere left to hide.

Notice what the demo is quietly claiming: the two angles α and β, together with the baseline L, determine the ship’s position completely. In geometry-class language this is angle-side-angle congruence: a side and its two adjacent angles admit exactly one triangle. Not “one, up to some ambiguity.” One.

So the existence of the answer is settled. But surveyors don’t want a picture — they want a number.

Part 3

Turning angles into distance

The third angle comes for free, since angles of a triangle sum to a straight angle:

$$\gamma \;=\; 180^\circ - \alpha - \beta$$

and then the law of sines converts angles into lengths — each side of a triangle is proportional to the sine of the angle facing it:

$$\frac{L}{\sin\gamma} \;=\; \frac{d_A}{\sin\beta} \;=\; \frac{d_B}{\sin\alpha}$$

A short shuffle of the same identity gives the number people usually care about — the perpendicular distance \(d\) from the baseline to the target:

$$d \;=\; L\,\frac{\sin\alpha \,\sin\beta}{\sin(\alpha+\beta)}$$

This is the punchline of triangulation as a measuring technology: one honest length, measured once, plus angles — which are cheap, fast, and astonishingly precise — buy you every other length in the picture.

Part 4

The enemy: skinny triangles

Real instruments wobble. Say every angle you measure is only trusted to within ±δ. Each measurement then pins the ship not to a ray but to a thin wedge, and the ship lives somewhere in the little quadrilateral where the two wedges overlap.

Here is where the geometry gets opinionated. Push the ship far away, or squeeze the baseline, and the two rays become nearly parallel — and nearly-parallel lines are terrible at deciding where they cross. The overlap stretches into an enormous sliver.

For a fixed instrument error, the blur of the answer grows roughly like

$$\text{uncertainty} \;\sim\; \frac{\delta \, d^{2}}{L}$$

— quadratic in the distance, and inversely proportional to the baseline. This single fact shapes all real triangulation: surveyors chase “fat” triangles with intersection angles near \(90^\circ\), and anyone measuring something very far away needs a very long baseline. Keep that thought; astronomy will return to it at the end.

Part 5

The cousin: trilateration

Flip the setup. What if you could measure distances but not angles? A radio signal that stamps its departure time gives you exactly that: distance = travel time × speed of light, with no directional information whatsoever.

One known distance pins you to a circle. Two circles meet in two points — one real, one impostor. A third circle settles the argument.

This is what GPS actually does — in 3D, with spheres instead of circles, plus a fourth satellite to forgive your cheap wristwatch clock. It is almost universally called triangulation, and is almost never triangulation. Angles: triangulation. Distances: trilateration. You are now entitled to be insufferable about this at parties.

Part 6

How to triangulate a continent

The real power move is chaining. Measure one baseline by hand, with excruciating care. Triangulate a new point from it. But look — each solved triangle hands you new sides of exactly known length, and any of them can serve as the baseline for the next triangle. Angles alone carry you forward, triangle after triangle, indefinitely.

The Great Trigonometrical Survey of India did precisely this, starting in 1802: one ~11 km baseline near Madras, then a lattice of giant triangles marched some 2,400 km up the subcontinent for decades.

At the far end of that chain, in 1852, computers (the human kind) reduced sightings of a distant Himalayan summit labelled “Peak XV” and got 8,840 m — within about a tenth of a percent of the modern value for Mount Everest, measured from stations over a hundred kilometres away from a mountain no surveyor had ever touched.

That is the promise from Part 1, kept at continental scale: measuring the unreachable.