A. Trevor Hodge
For the Roman engineer, the ideal aqueduct was one running just below the ground, and the arches and bridges that so excite our admiration were something to fall back on only when it could not be helped. Naturally, that does not mean that, once the arches had to be there anyway, the Romans were insensible to their spectacular appeal; but it does mean that the arches represented only a small fragment of the whole aqueduct, while they are the first thing we think of, and sometimes the only one. This is like trying to form a picture of a complete railway system only on the basis of its bridges and viaducts. Though such works might be accepted only as an unavoidable necessity, they did, of course, exist, and still do. Such engineering works fall into three categories. The first, carrying the conduit through inconveniently located mountains, are tunnels. The second, carrying across valleys and gorges, are bridges. The third, carrying it across broad open plains, where the object is simply to keep the water level high enough to provide proper service at the point of delivery, are arcades.
It is not strictly accurate to speak of tunnelling under a mountain. The object being to dig as short a tunnel as possible, if at all, the best way to deal with a mountain was to go round it. But a ridge, a col, or a long projecting spur, would have to be pierced, there being no way round. Once the surveying had been done and the line of the tunnel marked out, there were two ways of digging it. One was to start simultaneously at each end and meet in the middle. This, as we have seen in the sixth century tunnel of Eupalinos on Samos (above, pp. 42–3), was a technique that goes remarkably far back in history. Though it sounds extremely sophisticated, in fact it has several disadvantages. One is that, there being only two workfaces, only two crews can be working at once; and, given the restricted