Second, a cable ship has two winches on board. But this does not do justice to them, as they are so enormous, so powerful, and yet so nimble that it would almost be more accurate to say that a cable ship is two floating winches. Nearly everything that a cable ship does reduces, eventually, to winching. Laying a cable is a matter of paying cable out of a winch, and repairing it, as already described, involves a much more complicated series of winch-related activities.

As Kelvin figured out the hard way, whenever you are reeling in a long line, you must first relieve all tension on it or else your reel will be crushed. The same problem is posed in reverse by the cable-laying process, where thousands of meters of cable, weighing many tons, may be stretched tight between the ship and the contact point on the seafloor, but the rest of the cable stored on board the ship must be coiled loosely in the tanks with no tension on them at all. In both cases, the cable must be perfectly slack on the ship end and very tight on the watery end of the winching machinery. Not surprisingly, then, the same machinery is used for both outgoing and incoming winch work.

At one end of the ship is a huge iron drum some 3 meters in diameter with a few turns of cable around it. As you can verify by wrapping a few turns of rope around a pipe and tugging, this is a very simple way to relieve tension on a line. It is not, however, very precise, and here, precise control is very important. That is provided by something called a linear engine, which consists of several pairs of tires mounted with a narrow gap between them (for you baseball fans, it is much like a pitching machine). The cable is threaded through this gap so that it is gripped on both sides by the tires. Monarch's linear engine contains 16 pairs of tires which, taken together, can provide up to 10 tons of holdback force. Augmented by the drums, which can be driven by power from the ship's main engines, the ultimate capacity of Monarch's cable engines is 30 tons.

The art of laying a submarine cable is the art of using all the special features of such a ship: the linear engines, the maneuvering thrusters, and the differential GPS equipment, to put the cable exactly where it is supposed to go. Though the survey team has examined a corridor many thousands of meters wide, the target corridor for the cable lay is 200 meters wide, and the masters of these ships take pride in not straying more than 10 meters from the charted route. This must be accomplished through the judicious manipulation of only a few variables: the ship's position and speed (which are controlled by the engines, thrusters, and rudder) andthe cable's tension and rate of payout (which are controlled by the cable engine).

One cannot merely pay the cable out at the same speed as the ship moves forward. If the bottom is sloping down and away from the ship as the ship proceeds, it is necessary to pay the cable out faster. If the bottom is sloping up toward the ship, the cable must come out more slowly . Such calculations are greatly complicated by the fact that the cable is stretched out far behind the ship - the distance between the ship and the cable's contact point on the bottom of the ocean can be more than 30 kilometers, and the maximum depth at which (for example) KDD cable can be laid is 8,000 meters. Insofar as the shape of the bottom affects what the ship ought to be doing, it's not the shape of the bottom directly below the ship that is relevant, but the shape of the bottom wherever the contact point happens to be located, which is by no means a straightforward calculation. Of course, the ship is heaving up and down on the ocean and probably being shoved around by wind and currents while all this is happening, and there is also the possibility of ocean currents that may move the cable to and fro during its descent.

It is not, in other words, a seat-of-the-pants kind of deal; the skipper can't just sit up on the bridge, eyeballing a chart, and twiddling a few controls according to his intuition. In practice, the only way to ensure that the cable ends up where it is supposed to is to calculate the whole thing ahead of time. Just as aeronautical engineers create numerical simulations of hypothetical airplanes to test their coefficient of drag, so do the slack control wizards of Cable & Wireless Marine use numerical simulation techniques to model the catenary curve adopted by the cable as it stretches between ship and contact point. In combination with their detailed data on the shape of the ocean floor, this enables them to figure out, in advance, exactly what the ship should do when. All of it is boiled down into a set of instructions that is turned over to the master of the cable ship: at such and such a point, increase speed to x knots and reduce cable tension to y tons and change payout speed to z meters per second, and so on and so forth, all the way from Porthcurno to Miura."

It sounds like it would make a good videogame," I said to Captain Stuart Evans after he had laid all of this out for me. I was envisioning something called SimCable. "It would make a good videogame," he agreed, "but it also makes a great job, because it's a combination of art and science and technique - and it's not an art you learn overnight. It's definitely a black art."

Cable & Wireless's Marine Survey department has nailed the slack control problem. That, in combination with the company's fleet of cable-laying ships and its human capital, makes it dominant in the submarine cable-laying world.

By "human capital" I mean their ability to dispatch weather-beaten operatives such as the Lan Tao Island crowd to difficult places like Suez and have them know their asses from their elbows. As we discovered on our little jaunt to Egypt, where we tried to rendezvous with a cable ship in the Gulf of Suez and were turned back by the Egyptian military, one doesn't just waltz into places like that on short notice and get stuff to happen.

In each country between England and Japan, there are hoops that must be jumped through, cultural differences that must be understood, palms that must be greased, unwritten rules that must be respected. The only way to learn that stuff is to devote a career to it. Cable & Wireless has an institutional memory stretching all the way back to 1870, when it laid the first cable from Porthcurno to Australia, and the British maritime industry as a whole possesses a vast fund of practical experience that is the legacy of the Empire.

One can argue that, in the end, the British Empire did Britain surprisingly little good. Other European countries that had pathetic or nonexistent empires, such as Italy, have recently surpassed England in standard of living and other measures of economic well-being. Scholars of economic history have worked up numbers suggesting that Britain spent more on maintaining its empire than it gained from exploiting it. Whether or not this is the case, it is quite obvious from looking at the cable-laying industry that the Victorian practice of sending British people all over the planet is now paying them back handsomely.

The current position of AT&T versus Cable & Wireless reflects the shape of America versus the shape of the British Empire. America is a big, contiguous mass, easy to defend, immensely wealthy, and basically insular. No one comes close to it in developing new technologies, and AT&T has always been one of America's technological leaders. By contrast, the British Empire was spread out all over the place, and though it controlled a few big areas (such as India and Australia), it was basically an archipelago of outposts, let us say a network, completely dependent on shipping and communications to stay alive. Its dominance was always more economic than military - even at the height of the Victorian era, its army was smaller than the Prussian police force. It could coerce the natives, but only so far - in the end, it had to co-opt them, give them some incentive to play along. Even though the Empire has been dissolving itself for half a century, British people and British institutions still know how to get things done everywhere.