Your Data Doesn't Travel Through the Sky — Woody Magazine

Woody Magazine

We write about the things that aren't news

Jul. 27, 2026 (Mon.)

Science

Your Data Doesn't Travel Through the Sky. It Follows a Route Opened 160 Years Ago Today.

On 27 July 1866 a telegraph cable finally held across the Atlantic. What carried it was not a stronger current. It was a better ear.

This sentence reached you along a strand of glass roughly the thickness of a human hair, lying on the seabed a few kilometres down. It did not come from space.

We picture the internet overhead. The word cloud does that work, and so does the sight of Starlink satellites filing across a clear night. But data moving between continents travels almost entirely underwater. The International Telecommunication Union puts submarine cables at more than 99 per cent of international data exchanges. When the US Federal Communications Commission last measured it, using 2013 data, satellites accounted for 0.37 per cent of American international capacity.

That 99 per cent deserves a footnote. TeleGeography, which tracks the industry, went looking for where the statistic came from and traced it to a single FCC report drawing on 2013 data. Their own careful phrasing is 95 to 99 per cent of intercontinental traffic. Either number points the same way: down, not up.

The route opened 160 years ago today.

Heart's Content, 27 July 1866

That morning a ship far too large for the harbour came into Heart's Content, a fishing settlement on the east coast of Newfoundland. The Great Eastern was the biggest vessel afloat, built to carry 4,000 passengers and never profitable at it. Her owners had torn out the saloons and lowered three enormous tanks into the hull to hold coiled cable.

She had left Valentia Island, off the south-west coast of Ireland, on 13 July with 2,730 nautical miles of cable below decks. Fourteen days later 1,852 miles of it lay on the ocean floor and the ship rode at anchor in Trinity Bay.

They loaded far more cable than they actually used. A line has to be paid out longer than the straight-line distance to follow the floor beneath it, and a line pulled taut is a line about to part. That surplus matters later.

The Atlantic crossing of 1866. Valentia Island and Heart's Content are the eastern and western termini jointly nominated to UNESCO's tentative list. The grey point marks where the previous year's cable parted. Sources place it about 600 miles short of Newfoundland, so it is drawn as an approximate position only. Coastlines from Natural Earth; the route is an indicative line between the two landing points.

The cable that had already died

This was not the first success. In August 1858 a cable worked.

Queen Victoria sent President James Buchanan a message of 98 words. It took sixteen and a half hours to come through, about ten minutes a word. By the standards of 1858 this was a miracle; the same message carried by ship took ten days. Cities on both shores held parades.

Then it died. The signal came and went across roughly a month, and by the reckoning of Thomson's own university the days it actually worked added up to 23. In that time it carried 732 messages, among them news that the rebellion in India had been put down.

Two men, two diagnoses

The Atlantic Telegraph Company employed two men to think about electricity, and they agreed on almost nothing.

Wildman Whitehouse was an English surgeon and a member of the Royal College of Surgeons. He had taken up telegraphy as a hobby and turned it into a career. William Thomson held the chair of natural philosophy at Glasgow and would later become Lord Kelvin, the man whose name now sits on the absolute temperature scale.

Their problem was that signals crossing the ocean arrived blurred and late. Nothing on land behaved this way.

A wire lying in seawater is not really a wire. Copper core, a layer of insulation, and then the whole conducting bulk of the Atlantic pressed around it: the arrangement behaves less like a cable than like an extremely long capacitor. Charge banks up along its length. Send a pulse and it does not arrive at the far end so much as fill the line first and then seep out, smeared and slow, until dots and dashes run together. Pour water into an empty pipe and it comes out the other side. Pour it into a pipe packed with sponge and nothing emerges until the sponge is soaked.

Whitehouse read this as a shortage of force, and his prescription followed: push harder. He wired induction coils to the line and drove pulses of several thousand volts into it. Accounts of the figure differ. Some put it at two to three thousand volts, others reckon it went past ten thousand.

Thomson read the same behaviour as a problem at the receiving end. The signal was arriving; nothing in the room could see it. The current emerging after two thousand miles of ocean came to perhaps one hundred-thousandth of what an incandescent bulb draws.

The diagnoses diverged, so the treatments diverged. One of the treatments killed the cable. Several thousand volts went through the insulation.

But did one man really kill it?

An objection is reasonable here. If a cable held eight years later, perhaps the 1858 line was simply badly made.

Historians have argued exactly that. The cable was manufactured and stored carelessly. It sat coiled on a wharf through months of sun, and gutta-percha, the insulating gum of the period, does not tolerate heat well. No specification governed the purity of the copper. On this reading the line was finished with or without Whitehouse.

Even so, what he did is not in dispute. Thomson's own university records it plainly: the company's electrician insisted on using his instrument rather than Thomson's, and burned the cable out by applying several thousand volts. The stations recovered legible messages only after switching to Thomson's receiver. The company dismissed him before the cable had finished dying and sent Thomson to take charge of the Irish end.

The cable was probably poor. The man who forced several thousand volts through it shortened whatever life it had.

What won was not a stronger current

Thomson's answer was the mirror galvanometer.

A tiny mirror is fixed to a magnet and hung on a filament. A faint current turns the magnet by a fraction of a degree and the mirror turns with it. A lamp throws a beam at the mirror and the reflection lands on a wall several metres away, where that fractional rotation becomes a visible slide of light. Anyone who has tilted a laser pointer by a millimetre and watched the dot leap across a room understands the geometry.

He had not increased the current. He had lengthened the scale.

The 1866 cable used his galvanometer and nothing else. It ran at eight words a minute, against ten minutes a word. The receiver did not manage that alone. The 1866 cable was a different object, remade to fresh specifications from the purity of its copper to its insulation.

The same failure, repeated on deck

The electricians were not the only ones making this mistake.

On a cable ship everything turns on the brake. Pay out too freely and you squander cable you cannot replace mid-ocean. Hold too tight and the line, dragged by its own weight and the ship's motion, parts.

That is precisely what happened in 1857. The ship lifted on a swell, the load on the cable spiked, and the moment came to release the brake. Nobody released it. The cable parted 330 miles out and sank beyond recovery.

Over the following winter William Everett, who had been chief engineer of the Niagara, took the same post for the expedition and rebuilt the paying-out machinery. The point of the redesign was not a firmer grip. It was a mechanism that released quickly of its own accord as tension climbed toward the breaking point.

In 1865 the Great Eastern lost the cable 600 miles short of Newfoundland, on 2 August. The account Charles Bright compiled preserves a contemporary report on the loss. It attributes the accident to the retarding strain placed on the line, and adds that proper handling of the machine would have prevented it.

Set the two failures side by side. The electricians pushed harder and burned the insulation. The engineers gripped harder and snapped the line. The remedies rhyme as well: Thomson stopped shouting and started listening, Everett stopped holding and started letting go. A line about to part is saved not by a firmer grip but by releasing it in time.

160 years on

The Heart's Content station stayed in service until 1965. Five more cables joined the same two points between 1866 and 1894. Ireland and Canada have jointly placed both sites on UNESCO's tentative list.

The network has grown. The International Cable Protection Committee now counts roughly 500 systems and about 1.8 million kilometres of cable. Glass has replaced copper, and eight words a minute has become tens of terabits a second.

Some things have not changed. A ship still crosses the ocean paying a line out behind it, and the line still lies on the bottom, unburied along most of its length.

The three places in this story. Gold marks the 1866 Atlantic telegraph cable; the dashed line is the Curie cable used for seismic sensing. The ring in the western Pacific is Taiwan's Matsu Islands, where two cables were cut in 2023. Some 500 systems and 1.8 million kilometres of cable now cross the seabed; only the routes discussed here are drawn, and each is indicative, since real cables detour around seafloor terrain.

Who cuts them

Ask what cuts a submarine cable and most people picture submarines and divers. The statistics are duller than that.

70–80%

Share of submarine cable faults caused by fishing gear and ships' anchors
International Cable Protection Committee

The ICPC records 150 to 200 faults a year. Between 70 and 80 per cent come from ordinary human accident, chiefly commercial fishing gear and ships' anchors, rather than sabotage. Somewhere in the world a repair ship puts to sea about three times a week.

The dullness is what makes the network exploitable.

On 2 February 2023 a cable serving Taiwan's Matsu Islands went down under a Chinese fishing vessel. Six days later a freighter took out the second. The two lines were each other's backup, so the islands lost both at once.

Some 13,000 residents fell back on a microwave link. A text message took fifteen to twenty minutes to send. Full service returned after 50 days.

Deliberate? Speaking days after the second break, Wong Po-tsung, vice chair of Taiwan's National Communications Commission, said there was no indication the incidents were intentional. Cables to Matsu had been damaged some thirty times between 2017 and 2023, nearly all of it read as accident.

Which is the point. Accidents are so routine that intent can hide inside them. From the outside nothing distinguishes an anchor dragged on purpose from an anchor dragged by mistake. Nor is there much slack in the repair fleet. The ICPC tracks about 60 cable ships worldwide, 15 of them in the Indo-Pacific, most privately owned and built for the profitable business of laying rather than mending.

Listening to the same line differently

Recently a group of scientists made Thomson's choice again.

In 2021 a Caltech team published a paper in Science describing how they turned a Google-owned cable into a seismometer. The Curie cable runs some 10,000 kilometres from El Segundo, California, to Valparaíso, Chile.

They attached nothing to it. They watched the ordinary telecommunications traffic already inside. Light travelling through optical fibre carries a polarisation, and when the cable flexes, that polarisation shifts by a tiny amount. The team sampled the shift twenty times a second.

Across nine months they registered around twenty earthquakes, among them the magnitude 7.7 event off Jamaica in January 2020. They picked up ocean swell as well.

Nobody touched the cable. They changed how they listened.

In 1866 Thomson read a trembling spot of light on a wall to hear a whisper from the far side of an ocean. Today researchers read light trembling inside glass on the floor of the same ocean to hear the planet move. The apparatus is unrecognisable. The instinct is identical.

The Last Word

The Atlantic did nothing to those cables. Voltage burned the insulation; the brake severed the line. Both were force added in the hope of preventing failure. Whoever pushes harder is the last to see what breaks.

Sources

Woody Magazine is edited and published by Woody. Claude AI is used as a tool in the editorial process, and all editorial judgment and final responsibility rest with the editorial desk. Readers are encouraged to verify independently.

이 블로그 검색

Daily Woody