The Submarine Telegraph Cable

Coat identical wires with several candidate materials — wax, varnish, natural rubber, a thermoplastic — and leave them submerged in strong brine, checking leakage to the water with a meter every day for a week.

Expect most to fail progressively rather than suddenly: the reading creeps rather than jumps.

That creep is the honest signal. A cable does not stop working when its insulation fails; it gets quietly worse until it is useless, at a depth where nobody can look at it.

Test insulation by watching a number drift over time, never by a single pass/fail measurement.

Make two specimens: an insulated wire alone, and the same wire with several steel wires laid helically around it. Hang increasing weight from each until failure.

Expect the armoured one to carry far more, and — more importantly — expect the plain one's conductor to stretch and thin before it breaks.

Then measure resistance under load. A stretched conductor's resistance rises before anything visibly fails.

This is the design rule: separate the member that carries force from the member that carries signal. The same reasoning puts a steel core in an overhead power line and aramid yarn in a fibre-optic cable.

Submerge a long insulated wire in brine, with the water as the second electrode, and measure the capacitance between wire and water. Then repeat with twice the length.

Expect capacitance to be proportional to length, and to be surprisingly large.

Now charge it and disconnect it, and find that it holds charge.

Say what the cable now is: not a wire, but a resistance and a capacitance smeared along every metre. Nobody designed that in. It is an unavoidable consequence of putting a conductor near a conductive ocean, and it is the thing that nearly killed the enterprise.

Send a short square pulse into your submerged line and observe the far end on an oscilloscope.

Expect it to emerge rounded, delayed and spread out — and expect the spreading to grow sharply with length.

Now send pulses faster and faster until consecutive ones merge into an unreadable ripple. Record that rate.

That figure is the cable's speed limit, and note what sets it: not attenuation, but dispersion in time. A signal you can still detect but can no longer tell apart from its neighbour is just as lost as one that never arrived.

Take the smeared output and try two approaches. First, raise the sending voltage. Second, leave the voltage low and detect with the most sensitive instrument you have.

Expect raising the voltage to help hardly at all — a bigger hump is still a hump, and it still overlaps its neighbour — while sensitive detection lets you read a much weaker but still distinct signal.

This is the counter-intuitive lesson the first Atlantic cable had to learn the hard way. When the problem is smearing, power is the wrong lever. Detect gently, slow down, and shape what you send.

The first Atlantic cable worked for three weeks and then died — of impatience. The 1858 cable carried messages between Britain and North America and then failed. The immediate cause was the use of very high voltage from an induction coil in an attempt to force a readable signal through, which broke down insulation already damaged by rough handling. The engineering error and the conceptual error were the same one: treating a smearing problem as a weakness problem.

William Thomson — later Lord Kelvin — supplied both the theory and the instrument. His analysis of the cable as distributed resistance and capacitance predicted that signalling rate falls with the SQUARE of length, which told everyone that a longer cable must be run slower, not louder. His mirror galvanometer, which threw a spot of light from a tiny mirror on a suspended magnet, could read currents far too small for any mechanical instrument. The permanently successful cable was laid in 1866 by the Great Eastern, and the 1865 cable, lost mid-ocean the year before, was grappled up and completed too.

A tree was a strategic material. Gutta-percha came from a small number of species in southeast Asia, was harvested destructively, and by the end of the century the stands were severely depleted. An entire global communications system rested on a botanical supply chain — the same shape of dependency as rubber for tyres, or particular minerals for batteries now. Ask what the irreplaceable material is; that is usually where the fragility lives.

It changed what 'news' meant. Before 1866 a message to America travelled at the speed of a ship, roughly ten days. After, it was minutes. Markets on two continents began to move together, diplomacy stopped being conducted by ambassadors with weeks of discretion, and the phrase 'the world got smaller' was coined about this and not about aviation. The Loading Coil blueprint is the sequel: the smearing found here is exactly what Heaviside's mathematics and Pupin's coils later learned to cancel.

Honest limits. Very low data rate — the early Atlantic cables managed a handful of words per minute. Enormously expensive to make, lay and repair, and a fault is somewhere under kilometres of water. Gutta-percha is a natural product of variable quality and finite supply. The line is a shared single channel, so its capacity is a hard political and commercial constraint, not just a technical one. And nothing about it is repairable by the user in any sense — this is the first communications technology that only a nation or a consortium could own.