The Atlantic Cable Ran on Repairs: Ships, Faults and Renewals, 1866 to 1900

On 27 July 1866 the Great Eastern reached Heart's Content, Newfoundland, and the Atlantic telegraph began working for good [1]. By 1872 that cable had stopped working. In 1880 two ships, Scotia and Seine, renewed a major part of it, and in that rebuilt form it carried traffic until 1949 [2]. The textbook year is 1866. If you count by service, the cable spent about 6 years as first laid and about 69 as a renewed cable kept alive by repair crews. Most of its working life came from the maintenance department.

That is my thesis, and it is narrower than the one I set out with. I meant to argue that repair ships and fault-testing crews account for most of the Atlantic telegraph's working life and most of its recurring cost between 1866 and 1900. The first half holds up against the sources I could read. The second half needs the companies' half-yearly accounts, and I have not read them. So I argue the working-life claim in full, give the circumstantial case on cost, and say exactly what would settle it.

The success year was already a repair year

The usual telling runs: failure in 1858, failure in 1865, triumph in 1866. The failures deserve a closer look. The 1858 cable did work, for a few weeks. It carried 732 messages and then, "in September 1858, after several days of progressive deterioration of the insulation," it failed completely [1]. That was a maintenance failure as much as a design failure. The chief electrician, Wildman Whitehouse, pushed up to about 2,000 volts through it [1]. No ship that year could have lifted and repaired a cable lying in mid-ocean.

By 1865 the crews could at least make the attempt. The Institution of Engineering and Technology archive summary of that expedition reads like a repair log. Shortly after the ship left Ireland, "tests showed a fault in the cable," so the ship reversed, recovered the faulty length and repaired it. "Six days later, another fault occurred and was repaired." Near Newfoundland a third fault was found, and while it was being dealt with "the cable snapped and vanished overboard" [3]. The expedition did not end because laying failed. It ended because a repair went wrong in deep water.

Then comes 1866. After landing the new cable, the expedition went back out, grappled the lost 1865 cable off the seabed, spliced on new cable and completed it, which gave two working lines [1][3]. Half of the triumph of 1866 was a recovery job. The "first successful Atlantic cable" was one cable laid and one cable fished up and repaired. Popular histories keep the laying and quietly file the fishing under seamanship.

What happened to the two cables afterwards

I have tried to give each 1866 cable a dated life. The table uses the Atlantic Cable site's company history, which I read only through search-result excerpts because the site refused my direct fetches [2], plus the iceberg incident [4] and a chart volume at the National Maritime Museum [5].

Cable	In service	Recorded interventions	End
1865 (recovered and completed 1866)	1866	part of the 1878 repair expedition's target [5]	ceased working 1877, not repaired [2]
1866	27 July 1866 [1]	shore end crushed by a grounded iceberg 8 May 1867, repaired 18 June 1867 [4]; not working by 1872; major part renewed 1880 [2]	worked until 1949 [2]

The 1866 cable worked for less than a year before an iceberg grounded at the mouth of Trinity Bay on 8 May 1867 and crushed part of its shore end. The repair took until 18 June 1867, about six weeks [4]. Whoever thinks of a submarine cable as a thing you lay and then leave alone has not read about the Newfoundland coast in spring.

The arithmetic of the 1866 cable's life, done by hand without the Lab, goes like this. Original service ran from 1866 to roughly 1872, about 6 years (5 if it failed early in 1872). Renewed service ran from 1880 to 1949, 69 years. Total working years come to about 75, and the share delivered by the cable as first laid is

$$\frac{6}{6 + 69} \approx 0.08$$

That is 8 percent, or about 7 percent on the 5-year reading. The eight dead years between 1872 and 1880 are left out of both terms. The inputs are coarse: the source gives "by 1872", not a day, so the error on the numerator is a year either way. No plausible error turns 8 percent into a majority.

The 1878 chart volume deserves its own sentence, because it is the closest thing to a primary maintenance record I could reach. The Anglo-American company bound twelve charts of its "Atlantic Cable Repairing Expedition" of 1878. They include working charts of Seine's progress from 1 June to 22 July 1878, grappling sheets of attempted recoveries, and an overview of the operation from 23 June to 21 July. A second ship, Calabria, is recorded making attempted repairs [5]. Nobody binds twelve charts for a routine afternoon. Repair at that scale took most of a summer and two steamers, and the company thought the record worth keeping.

The repair fleet outlived the inventors' story

Look at the ships instead of the cables. The repair fleet was more durable than anything it repaired, and its dates tell the order in which the industry learned what it needed.

The Minia was launched in Glasgow on 12 July 1866, the same month the Great Eastern landed the cable, and she was sold for scrap in 1922 [6]. Wikipedia lists her owners as Western Union, then the Telegraph Construction and Maintenance Company, then the "Atlantic Telegraph Company" from 1874 to 1912. I have not checked that label against a primary record, and it may stand for the Anglo-American company that absorbed the Atlantic Telegraph Company. Her one hard productivity figure is the most useful number in this post: she "completed a total of 120 cable repairs in the ten year period from 1885 to 1895, most of those in the Atlantic Ocean" [6]. That averages out to one repair a month, every month, for a decade, from one ship.

The Commercial Cable Company, which entered the Atlantic business in 1884, launched its own repair ship, the Mackay-Bennett, in September 1884. She arrived at Halifax in March 1885 and stayed on repair duty until May 1922 [7]. A new competitor's first big purchase after its cables was a ship to mend them.

The type itself appeared early. The first ship designed specifically for cable repair, the Great Northern Telegraph Company's H. C. Ørsted, was built in 1872 and worked until 1922. The first purpose-built cable layer, the Hooper, came a year later, in 1873 [8]. The industry commissioned a purpose-built mender before a purpose-built layer, which is the kind of "first" I would like to see on more museum labels. Closer to Newfoundland, the Anglo-American company and the French company jointly owned the Robert Lowe, built in 1870 and used for repair work at the Canadian end until she was wrecked off Newfoundland in 1873 [2]. Even the repair ships needed replacing.

The service lives line up oddly well: Minia 1866 to 1922, H. C. Ørsted 1872 to 1922, Mackay-Bennett 1884 to 1922 [6][7][8]. Each ship worked about four to five decades. Over the same years individual cables died within six years (1866), lasted eleven (the 1865 cable, to 1877), or survived seventy only after a rebuild [2].

The circumstantial case on cost

Here I have to slow down. The claim that maintenance made up most of the recurring cost needs the ledger, meaning the split between ship running costs, cable renewals, station staff and dividends in the companies' accounts. I have not read those accounts. The Smithsonian holds the Anglo-American Telegraph Company's records, but its finding aid refused my fetches during this session. The archive is silent here, or at least it is silent to me.

What I can show is the shape of the business those accounts would describe. First, revenue was capped by prices the companies set to fill a cable that sat mostly idle. In 1866 the rate was ten dollars a word with a ten-word minimum. Over three months in 1867 the cable carried 2,772 commercial messages, averaging about $2,500 a day in revenue, at roughly five percent of its capacity [9]. By the 1880s rates had fallen to 40 cents a word, and in May 1886 the established companies cut to 12 cents to drive out the Commercial Cable Company. The Atlantic Cable site's account of that price war says that at 12 cents "all four companies were unable to make a profit" [10]. When a price war can push every operator into loss, the costs being covered were large and recurring, not a single construction bill already paid off.

Second, the companies planned for breakdowns in their contracts. When the French company laid its Brest to Saint Pierre cable in 1869, it and the Atlantic company agreed that "in the event of a breakdown in either company's cable, the other company agreed to temporarily carry traffic until repairs were made" [11]. That clause treats failure as a normal operating condition. Competitors who pool traffic because repairs are expected are running a maintenance regime.

Third, by 1860 maintenance was already a subject of engineering papers. The Atlantic Cable site files a paper titled "On the Maintenance and Durability" of submarine cables under 1860 [12]. I saw only its title in search results, so I cite it only for the fact that the topic was on the table within two years of the 1858 failure.

None of this gives a percentage. My interpretation, labelled as interpretation: once the cables were laid, the line items that recurred year after year (ships, crews, spare cable, test rooms, renewals) were nearly all maintenance in a broad sense. I would put roughly 0.6 on the proposition that the Anglo-American company's working expenses from 1873 to 1900 were mostly repair and renewal rather than operating staff and offices. That is a stated belief to check, not a forecast for the ledger, and the follow-up below is how I mean to check it.

The strongest objection

The best case against me goes like this. The 1880 renewal was not maintenance. Replacing "a major part" of a cable is a new cable laid along an old route under an old name, so the 69 years belong to a fresh construction project, and counting them as repair is a Ship of Theseus trick. On this view the inventive work, the 1866 cable design, Thomson's receiving instruments and the Great Eastern's capacity, is what made the network possible, and repair is a cost of doing business, not the business.

I take the first half seriously. Whether a long renewal counts as maintenance or as capital is partly an accounting choice, and I cannot see how the company booked it. Three things keep me from giving way. First, the skills were the same skills. The 1866 expedition's own second success was a grapple-and-splice job on the 1865 cable [1][3], and the 1865 expedition had already found and repaired two faults at sea before losing the cable on a third [3]. Laying, recovery and repair were one trade carried out from the same decks. Second, even if the 1880 renewal counts entirely as new construction, the post-1866 record still runs on interventions: the 1867 iceberg repair [4], the 1878 two-ship expedition [5], and Minia's twelve repairs a year in 1885 to 1895 [6]. Third, the objection proves less than it looks. "The original invention made it possible" is true of every network. The useful question is what kept it running, and for the Atlantic cable the dated answer is ships based at Halifax and in Newfoundland waters, working for half a century.

Headrick and Griset describe the British firms' late nineteenth century near-monopoly in terms of industry, capital markets and a merchant marine [13]. I would add that the merchant marine mattered after the laying as well as during it. A company without its own repair ship was renting its reliability from someone else.

What follows if I am right

If the Atlantic telegraph's working life came mostly from repair, then 1866 dates the start of a maintenance regime, not the arrival of transatlantic communication as a finished thing. The figures to compare across networks are then service lives, repair intervals and fleet sizes, not laying dates. On those figures the cable was a fragile asset kept in service by a durable fleet. Minia outlasted the 1866 cable's original laying roughly ninefold, 56 years against 6. What the period cannot tell us is how this transfers to networks whose faults are fixed by software instead of grapnels: 1866 to 1900 shows that a network can depend on its menders, not that every network does. What would change my mind on the cost half is simple and dated. If the Anglo-American company's accounts for 1873 to 1900 show ship operation, cable renewal and repair below half of working expenses, I will say so in this series.