Twice an hour, for three minutes, the North Atlantic fell silent. At a quarter past and a quarter to, every wireless operator on every ship within reach of 500 kilohertz stopped transmitting, pressed his headphones a little harder against his ears, and listened to nothing. The clock on the radio-room bulkhead had those two windows marked in red so that nobody could forget them. If anything was going to come through, it would come through then.

What the operators were listening for was not a word. It was a rhythm: three short, three long, three short, run together as a single character with no gaps between the letters. It stood for nothing. It abbreviated nothing. It had been chosen precisely because no other sequence on a crowded band sounded remotely like it.

Eighteen years after that rhythm became international law, a radio officer at a muddy aerodrome south of London faced the same problem all over again, and discovered that this time the answer could not be a rhythm at all. Aeroplanes did not tap. Aeroplanes talked. And so aviation went looking for a word.

SOS never stood for anything

Ask most people what SOS means and they will tell you “save our souls” or “save our ship”. Both are inventions, applied to the signal long after it existed. The letters were reverse-engineered from the pattern, not the other way round.

The signal was German first. Imperial radio regulations that took effect on 1 April 1905 listed a Notzeichen — a distress mark — consisting of three dots, three dashes and three dots. The following year the first International Radiotelegraph Convention met in Berlin, and the agreement signed on 3 November 1906 put the sequence into international law in the plainest possible language. Article XVI: ships in distress shall use the signal, repeated at brief intervals. It came into force on 1 July 1908.

Nine elements, alternating short and long, sent without the pauses that would normally separate one letter from the next. On a band full of commercial traffic, weather, spark-gap crash and atmospheric noise, that shape was unmistakable even to an exhausted operator at the end of a watch. Marconi’s own company had already tried its own call — CQD, specified on 7 January 1904 — but CQD was three ordinary letters and sounded like three ordinary letters. The German idea was better, and it won.

The Titanic did not invent the SOS

This is the story everyone knows, and it is wrong. The Titanic is routinely described as the first ship to send an SOS. She was not, and she was not close.

By April 1912 the signal had been in use for years. The American steamer Arapahoe sent it on 11 August 1909 after breaking a shaft off North Carolina. The Kentucky sent it on 4 February 1910, leaking badly off Virginia. The Merida sent it on 13 May 1911 after a collision, the Niobe on 30 July 1911 after running ashore, the Lexington on 28 August 1911 in a hurricane, the Sterling on 3 December 1911 after another collision. The Ontario sent it on 9 April 1912 — five days before the Titanic struck ice.

What the Titanic did was something stranger and more human. Her two Marconi operators, Jack Phillips and Harold Bride, did not settle on one call at all. They mixed them, alternating the old company signal and the new international one through the whole ninety minutes. Marconi men had been slow to abandon CQD; the pattern from Berlin was official but it was not yet instinct. Afterwards it became instinct very quickly indeed, and CQD faded out.

Three minutes of silence, twice an hour

The sinking changed the rules. A second convention met in London and was signed on 5 July 1912, coming into force a year later. It made 500 kilohertz the primary seagoing frequency rather than one option among several, standardised the procedure, and imposed the discipline that gave the band its peculiar rhythm.

Twice an hour, every station working 500 kHz shut up for three minutes: from fifteen to eighteen minutes past, and from forty-five to forty-eight. Nothing but distress traffic was permitted in those windows. A weak signal from a small ship with a modest aerial, which would have been buried under commercial traffic at any other moment, had a clear channel twice an hour, every hour, for the best part of a century.

Marine radio-room clocks were built for it. The red sectors on the dial mark the radiotelegraph silence periods; a second pair in green, at the hour and the half hour, mark the equivalent windows on the radiotelephone frequency 2182 kHz. It is a clock designed around the idea that listening matters more than talking.

Then the radio learned to talk

Croydon, south of London, was Britain’s aerodrome for the Continent in the early 1920s, and the route that defined it ran to Le Bourget in Paris. Aircraft were small, engines were unreliable, and the Channel was cold. The traffic was Anglo-French in a way that mattered: the men on the ground at either end did not necessarily share a first language.

Crucially, aircraft radio was not telegraphy. It was wireless telephony — a voice in a headset, not a key under a finger. And a voice cannot send SOS, because SOS is not a word. You can say the three letters aloud, but at that point you have thrown away everything that made the signal work. The pattern was the point, and speech has no pattern.

Someone had to notice this and do something about it. On Friday 2 February 1923 two English provincial newspapers, the Gloucestershire Echo in Cheltenham and the Dover Express & East Kent News, carried the news within a day of each other. The Echo ran it under a headline that says the whole thing in five words: “MAY-DAY” FOR S.O.S. The explanation underneath is the clearest statement of the problem anyone has ever written.

The Dover paper, the same day, described what the new call actually sounded like in use. During a recent test, it reported, an RAF flying-boat coming down in the Channel gave the international distress signal three times by wireless telephony and reported that her engines had failed.

Three times. In February 1923, in a newspaper, before any convention had blessed it. The repetition rule that pilots still follow a century later was there from the first documented use.

Who actually coined it, and how sure can we be

The standard account names a man. Frederick Stanley Mockford, senior radio officer at Croydon, was asked to find a distress word that everybody would understand, thought about where the traffic was going, and proposed m’aider — a compressed form of venez m’aider, come and help me — anglicised into two syllables that an English speaker could pronounce and a French speaker could recognise. The Historic Croydon Airport Society, which keeps the site’s records, states that Mockford conceived the phrase in 1923, that the Air Ministry approved it, and that the word was chosen because it resembled nothing else in operational use.

It is a good story, it is told by people with a legitimate claim on the history, and it may well be true. But it is worth being honest about what the paperwork actually shows, because the paperwork is thinner than the confidence with which the story is usually repeated. The etymological reference Word Histories, which traced the earliest printed appearances, puts the caveat bluntly.

The detail that gives that caveat its force is simple: neither of the February 1923 newspaper reports mentions Mockford at all. They announce a new signal, explain why it was needed, and describe a test. The name arrives later. That is not proof of anything either way, and it is not a reason to strike a man from the record. It is a reason to write “attributed to” rather than “invented by”, which is a distinction aviation history is generally too casual about.

The rule book still spells out the French

Here is the detail that settles the argument about where the word comes from, and almost nobody quotes it. The regulations do not merely permit the French reading. They print it.

The International Telecommunication Union’s Radio Regulations, Article 32, define the radiotelephone distress signal as the word MAYDAY, pronounced as the French expression m’aider. The United States codifies the same sentence almost word for word in 47 CFR 80.314. A regulator does not normally explain the etymology of a term it is defining. This one does, because the pronunciation Ist the specification: the whole value of the call rests on it sounding like one particular thing and nothing else.

The format follows from the same logic. MAYDAY spoken three times, then THIS IS, then the identification, also three times. Britannica notes that the word was adopted for radio because it is short, easily understood and unlikely to be mistaken for other sounds — which is the twentieth-century way of saying what the Gloucestershire Echo said in 1923 about the letter S.

What it does to a frequency

In aviation the procedure sits in the FAA’s Aeronautical Information Manual at paragraph 6-3-1, and the wording there is careful in a way worth noticing. The initial communication from an aircraft in distress should begin with the signal MAYDAY, preferably repeated three times. Preferably. Not must. A pilot who is busy flying an aeroplane that is trying to kill him is not going to be failed on phraseology.

What the word does is legal rather than technical. Distress communications have absolute priority over everything else, and the AIM puts it in one clean line: the word MAYDAY commands radio silence on the frequency in use. Its junior partner PAN-PAN, from the French panne, breakdown, warns other stations not to interfere but ranks below distress. One word clears a channel; the other asks politely.

The emergency frequencies are 121.5 MHz and 243.0 MHz, both line-of-sight, with 121.5 guarded by direction-finding stations and by a good number of military and civil aircraft. A century after a flying-boat put down in the Channel, the architecture is the same: say the word, and the noise stops.

The Morse watch ended on 1 February 1999

The older signal outlived the technology that justified it by a remarkable margin, and then stopped almost overnight. The International Maritime Organization amended the Safety of Life at Sea convention in 1988 to require the Global Maritime Distress and Safety System, and on 1 February 1999 the compulsory human listening watch ended. Digital selective calling, satellite EPIRBs working through Cospas-Sarsat and Inmarsat terminals took over the job that a man with headphones had done since before the Titanic.

Ninety years of silence periods closed with a date on a calendar. Amateur operators still mark it. The frequency itself has moved on: the 2019 World Radiocommunication Conference reallocated 500 kHz to NAVDAT maritime data.

But the signal is not repealed, and this is the part that surprises people. Annex IV of the international collision regulations lists the recognised distress signals, and it still contains both. Item (d): a signal made by any signalling method consisting of the group three dots, three dashes, three dots in the Morse code. Item (e): a signal sent by radiotelephony consisting of the spoken word Mayday. A torch flashed at a ship’s bridge in that rhythm is a distress signal in law today, just as it was in 1908.

Two answers to the same question

What separates them is not age or nationality or any of the folklore. It is the channel. Morse is a medium with no vocabulary, so the distress call had to be a shape. Voice is a medium with nothing but vocabulary, so the distress call had to be a word — and it had to be a word robust enough to survive a bad headset, an engine at full power, a frightened voice and, in 1923, an aerial trailing behind a biplane over open water.

One was picked by German regulators for the way it sounded to an ear trained on static. The other was picked, by somebody at Croydon whose name we cannot quite prove, for the way it sounded to an ear listening through a hiss. Both answers were right. Neither has been improved on in a hundred years.

Sources: International Telecommunication Union Radio Regulations Article 32; 47 CFR 80.314; FAA Aeronautical Information Manual 6-3-1; COLREGS Annex IV; Gloucestershire Echo and Dover Express & East Kent News, 2 February 1923, as reproduced by Word Histories; Historic Croydon Airport Society; International Maritime Organization; Encyclopaedia Britannica; Snopes.