Satellite positioning tells a vessel where it is to within a metre or two. That is superb for knowing your position on a chart and considerably less superb for knowing your position relative to a channel that was dredged last spring, silts unevenly, and is marked on a survey that may be eighteen months old. The precision of the fix is not the limiting factor. The currency of the chart is.

A range does not have that problem, because it is not describing the channel. It is the channel. If the authority moves the line, it moves the towers or relights them, and the information reaching the bridge is correct the moment the work is done.

There is a second argument, less often made out loud. A range fails visibly. If a lamp is out, you can see that a lamp is out. An electronic system that is confidently wrong looks exactly like an electronic system that is right, and the history of groundings in well-charted water is substantially a history of people trusting a screen.

So ranges persist in commercial ports, on river approaches, and anywhere large vessels move through tight water. Many have been rebuilt in steel lattice, unlovely and entirely functional. A reasonable number of the original masonry towers are still doing the job they were built for, which is a rarer thing in this field than it sounds.

The real reason is sensitivity. What the mariner is judging is the vertical gap between the two lights. On the centreline that gap is zero, or as close to zero as the surveyor could manage. A boat's length off to one side, the gap opens by some amount. How large that amount is, for a given error, is what decides whether the range is any use.

Two things control it: how far apart the towers are, and how much taller the rear one is. Increase either and a small drift produces a visibly larger separation. So a range built for a narrow dredged channel, where being fifteen metres off matters, will have its towers set well apart and a pronounced height difference. A range marking a broad approach can be much more relaxed about both.

This is why range towers so often look wrong in photographs. They are not composed. Their proportions are the output of an equation about how tightly the channel needs to be held, and the equation does not care what the result looks like from the shore.

It also explains the awkward siting. A rear range is frequently standing in somebody's field, half a mile inland, apparently unrelated to the water. It is exactly where the geometry required it to be, and nowhere else would have done.

Two lights solve it, provided they are in the right relationship. Put one low and near, put the second higher and further back, and align both on the centreline of a safe channel. A mariner steers until the two appear stacked, one directly above the other. At that moment they are on the line joining the towers, and the line has been surveyed to sit in deep water.

Off to one side, the lights separate. Which side they separate towards tells you which way you have drifted, and by how much. There is no instrument involved, no calculation, and nothing to fail. The information is delivered by geometry straight into the eye.

That is why the arrangement survived every technological shift that killed off its contemporaries. It is not a signal that has to be interpreted. It is a picture that is either correct or visibly not, and a helmsman can act on it in the time it takes to look up.

The formal name is a leading line, and the two towers are the leading lights, or in American usage the range. The near one is the front range, the far one the rear range. Everything interesting about how they perform comes down to the distance between them and the difference in their heights.