The water beneath the hill

AIn dry country the hard part is rarely finding water. It is keeping it. An open channel cut across a plain loses a great deal of what it carries before the water arrives anywhere useful: some of it sinks into the bed, and much more is simply lifted away by hot dry air. Any society that wanted to move water across such a landscape had either to accept the loss or to put the channel out of reach of the sun. The second answer, worked out on the Iranian plateau more than two thousand years ago, was to bury the whole thing.

BThe method begins at the edge of the mountains, where rain and melting snow have soaked into loose rock and collected underground. A deep shaft is dug there until it reaches that water. From the settlement, often many kilometres away, a tunnel is then driven back towards the shaft, rising very slightly all the way. The slope is the whole craft. Make it too steep and the moving water scours the floor until the roof falls in; make it too gentle and the water stops and silts up. Neither mistake can be corrected afterwards without abandoning the work.

CBecause a tunnel of that length cannot be dug from one end alone, vertical shafts are sunk every twenty or thirty metres along the route, so that spoil can be lifted out and air can get in. From the surface the result is unmistakable: a line of holes, evenly spaced, running from the hills towards the fields, each ringed by the earth brought out of it. The shafts are not abandoned once the tunnel is finished. They are the only way back in, and clearing silt from a tunnel is work that never ends.

DWhat made the system last was partly that nothing about it can break. There is no pump, no fuel and no moving part; the water is driven by gravity and arrives at a steady rate. It also cannot be overdrawn. A tunnel takes only the water that happens to reach it, so in a dry year the flow falls and in a wet year it rises, but no one can make it deliver more by trying harder. That limit, which looks like a weakness, is the reason many of these systems ran for centuries without emptying the ground beneath them.

EA steady flow still has to be divided, and here the arrangement was unusual: the water was shared out by time rather than by quantity. A household did not own a volume of water; it owned a number of minutes of the whole stream. At the head of the channel sat a timekeeper with a bowl, floated on water and pierced with a small hole, which sank at a known rate; each time it sank, one share had passed and the next household opened its ditch. Disputes were therefore about minutes, and the written records that survive read less like engineering than like accounts.

FThe decline came from an unexpected direction. From the middle of the twentieth century, diesel and then electric pumps became cheap enough for an individual farmer to buy. A pump reaches deeper than a tunnel and lifts water far faster, and a few pumps near an old system will pull the water table down below the level of its tunnel. The tunnel then runs dry, not because it was worn out but because the water it depended on had been taken from underneath it. The new technology did not compete with the old one; it removed the thing the old one lived on.

GSome are now being repaired, for two quite different reasons. One is heritage: a working example is a rare thing and attracts visitors. The other is practical, and has become more persuasive as fuel has become less reliable, since a restored tunnel delivers water for years and needs no electricity at all. The obstacle is not money but people. The judgement of slope and soil was carried by a small number of master diggers, was taught by working alongside them, and was almost never written down.