01Executive summary
Four milestones in the first industry built deliberately out of laboratory physics — and the first in which a submarine cable forced engineers to design against a transmission medium they could not inspect.
Chappe's optical semaphore of 1794 proved the demand existed and showed what it cost to meet it mechanically. Cooke and Wheatstone put an electric telegraph into service beside a railway in 1837, and Morse developed a different and ultimately more economical instrument in America. The Atlantic cable entered service in 1866 after a decade of expensive failures.
02The optical telegraph: the right problem, the wrong physics
Humans have needed to send messages faster than a person can travel for as long as there has been government over more than one settlement. Signal fires, smoke, drums, runners, pigeons and relayed shouting all served. Chappe's system of 1794 was the first practical improvement on them: a line of towers six to sixteen kilometres apart, each carrying a pivoted beam with movable arms, read through a telescope from the tower on either side.
It worked well. Over the 760-kilometre line of 120 towers between Paris and Toulon, a message is said to have taken ten to twelve minutes. The French government ran the first line between Paris and Lille and the system spread quickly; Britain built lines to the Channel ports against the possibility of a French landing, and the first American line connected Martha's Vineyard to Boston in 1800 to report ship arrivals.
Every tower needed a person on duty, so operating cost scaled linearly with distance and continued whether or not any message was sent. And the whole line stopped in fog, rain or darkness. Both are inherent to signalling by line of sight, not defects of execution — which makes this the same kind of limit as mechanical television scanning in a later series. When a limitation follows from the mechanism rather than from the implementation, the answer is a different mechanism, and recognising which case you are in is one of the more valuable judgements available.
03The electric telegraph: assembled from four laboratory results
The electric telegraph is the clearest single illustration of this series' argument, because its prerequisites can be listed and every one of them is a scientific result obtained by someone with no commercial intent.
| Requirement | Supplied by | Year |
|---|---|---|
| A continuous current | Volta's pile | 1800 |
| A current producing mechanical movement at a distance | Oersted's electromagnetism | 1820 |
| A practical electromagnet | Sturgeon, later improved by Henry with many turns of insulated wire | c.1825 |
| A way to predict behaviour over a long line | Ohm's law | 1826 |
By 1830 the demand was insistent enough that dozens of people were attempting an electric telegraph. Ampère had suggested in 1820 using a circuit per letter with a magnetic needle at the far end; Ritchie demonstrated it on a small scale in 1830 and Schilling worked out a system in 1832 without producing a practical instrument.
Cooke saw electromagnetic experiments in Heidelberg in 1836 and had built a needle instrument within three weeks. Faraday introduced him to Wheatstone, who had already done substantial telegraph work, and the two formed a partnership in 1837. Before the agreement was signed they had a line running about 1.6 kilometres from Euston along the London and Birmingham Railway, and had sent messages over it. Their early instruments needed five or six wires and as many needles, each pivoted between the halves of a double coil, swinging left or right to indicate letters.
Right of way and a customer in one
A railway offers a continuous corridor already under single ownership, and the railway itself needs to know where its trains are. Telegraph and railway grew together for good structural reasons, and the same logic later attaches fibre to rail and pipeline corridors.
Fewer wires
A needle system needing five or six conductors is expensive per kilometre. Encoding letters as sequences in time rather than as parallel signals trades bandwidth for conductors, and over long distances that trade is decisive.
Morse, working independently in America from 1832, completed his first instrument in 1836 and learned from Gale of Henry's improved electromagnets, which greatly increased the distance over which he could signal. His receiver was a pen held against a paper strip driven by clockwork, marking when the electromagnet was energised. The public credited the system after it helped catch a murderer in 1845 — a man who poisoned a woman at Slough and took the train to London, described ahead by telegraph. Publicity is not engineering, but a technology that has demonstrated public value gets capital, and one that has not does not.
04The Atlantic cable: engineering something you cannot inspect
Submarine cables were working in Europe by the early 1850s, with a Channel service from November 1851 and later links to Scandinavia and across the Mediterranean. Crossing the Atlantic was a different order of problem.
- 1856The Atlantic Telegraph Company is organised.
- 1857The first attempt between Ireland and Newfoundland fails after 540 kilometres when the paying-out machinery causes the cable to part.
- June 1858A second attempt fails similarly.
- August 1858A cable is completed. It works for about two months, degrading steadily, and ceases to function on 20 October.
- 1858–65Seven years of work on insulation, mechanical strength, paying-out apparatus, receiving instruments and signal type.
- July 1865Using the Great Eastern, the cable parts after 1,900 kilometres in about 3,800 metres of water and the attempt is abandoned.
- July–September 1866A new cable is landed at Newfoundland; the Great Eastern then finds and recovers the 1865 cable, splices it, and lands a second.
Sir William Thomson, later Lord Kelvin, was largely responsible for the eventual success. The difficulty is that a long submarine cable is not a wire; it is a distributed capacitance and resistance, so a sharp pulse sent at one end arrives at the other smeared out over time. Signalling faster makes successive pulses overlap until nothing is readable. That is not something trial and error resolves efficiently — it needs a theory of the line and instruments sensitive enough to read a very weak, very slow signal. The first cable was damaged in part by attempts to force signals through with high voltage, which is exactly what someone without the theory would try.
Finding a cable lost the previous year, in nearly four kilometres of water, grappling it from the seabed, raising it without breaking it, splicing to it and completing the run is a marine operations achievement at least equal to laying a new one. It also means that after four failures the company finished the season with two working cables. Building the capability to recover from failure, rather than only the capability to succeed, is what turned a decade of losses into a permanent industry — and it is the same argument as designing landing attempts into missions that have already delivered their payload.
The economics were understood at the time in a way that is easy to lose. While the short-lived 1858 cable was working, a message from London to Canada countermanded the departure of two regiments, saving an estimated £50,000 — a substantial fraction of what a cable cost. Communications infrastructure is valued on the decisions it changes, not on the traffic it carries.
05Takeaways for current practice
- Distinguish limits of mechanism from limits of execution. An optical relay needs staff at every tower and clear weather, and no refinement changes that.
- Trade one scarce resource for another deliberately. Morse traded time for conductors, and over distance that is the right direction.
- Where the medium is distributed, you need the theory. Capacitance along a long cable is not something trial and error resolves cheaply.
- Build the recovery capability, not just the delivery capability. The 1866 season succeeded because the previous year's loss could be retrieved.
- Value communications on decisions changed, not messages sent. One countermanded order paid a visible fraction of the asset.