First Actual Case

There is a moth in a museum in Washington. It is a couple of inches across, it has been dead since 1947, and it is stuck to a page of an engineering logbook with a strip of tape, beneath a line of handwriting.

The line reads: 1545 Relay #70 Panel F (moth) in relay. First actual case of bug being found.

It is one of the best known objects in the history of computing, and it is almost always described as the moment the word bug entered the language. The sentence taped above it says, in four words, that this is not what happened.

The machine

The Mark II was finished at Harvard in September 1947. Howard Aiken directed the work, with Edmund Berkeley and Grace Hopper on the team, and the United States Navy paid for it because it wanted ballistics calculated faster.

It was a relay machine. Not valves, not transistors: thousands of electromechanical switches, each a small arm pulled across a pair of contacts by a coil. It could add in an eighth of a second and multiply in three quarters of one, which was several times quicker than the Mark I and is a rate a modern processor exceeds several billion times over.

The point about relays is that they are mechanical, they are open to the air, and the gap between their contacts is about the size of an insect.

At a quarter to four on the afternoon of 9 September, somebody working through a fault found one wedged in relay 70 on panel F. They got it out, taped it to the log, and wrote the line.

The word was seventy two years old

The earliest known use of bug for a fault in a machine is in The Operator, a journal of scientific telegraphy, on 15 August 1875. "The biggest 'bug' yet has been discovered in the U.S. Hotel Electric Annunciator."

It is worth knowing what that machine was, because it makes the sentence ordinary rather than quaint. An annunciator was the board behind a hotel desk that showed which room had rung the bell for service. Every room had a wire, and the board dropped a little flag or lit a number when a guest pulled the cord. It was among the first pieces of electrical equipment that ordinary people used without knowing they were using it, and when it went wrong the men who fixed it already had a word for the reason.

Thomas Edison was using it routinely three years later. In March 1878 he wrote to William Orton that he did find a bug in his apparatus. In May he wrote to William Preece wondering whether the static charge was the bug. In November he set out what he meant, in a letter describing what invention actually consists of: months of intense working study and labour are required, he wrote, once the bugs, as such little faults and difficulties are called, show themselves.

That is a definition, written down by a man explaining his trade to somebody outside it, in 1878. The word was already established enough to need explaining rather than inventing.

Read the November letter again and notice what Edison is actually describing. He is not complaining about faults. He is saying that the faults are the work: that an idea arrives quickly and then months of intense study and labour are required, because the bugs show themselves only once the thing is built and running. That is a fair description of software written a century before there was any.

Seventy two years separate the annunciator at the U.S. Hotel from the moth in relay 70.

Why the word actual is doing all the work

Read the log entry as an engineer would read it and the joke is obvious.

These men had been saying all week that they were hunting a bug. They meant a fault. Everybody in the room, and everybody in every workshop like it going back to the telegraph, meant a fault. Then one of them opened a panel and there was a literal insect stopping the current, and the funny thing was not that a bug had been found. It was that this time the bug was an actual bug.

That is what first actual case means. It is a punchline, and it only works because the audience already knew the word. An entry recording the birth of a term would not need the qualifier at all.

Whose story it became

Grace Hopper worked on that machine and told the story for decades, and it became attached to her so firmly that many people believe she coined the word. She did not, and neither did anybody else in the room that afternoon. Like Edison seventy years earlier, she was using slang she had inherited.

What she did do was keep the page. The moth stayed taped where it was put, went with the machine's records, and ended up in a national collection rather than a bin, which is a more unusual outcome than finding an insect in a relay.

The date

Not every account agrees on the year. Some put the log entry at 9 September 1945 rather than 1947, and the page itself carries the day and month more clearly than it carries the year.

The machine settles it. The Mark II was not finished until September 1947 and did not go to the Naval Proving Ground at Dahlgren until March 1948. Nobody found a moth in relay 70 in 1945, because in 1945 there was no relay 70.

What debugging kept

The word survived the transition out of hardware entirely. There are no relays left to trap anything, and a modern fault is a wrong assumption written down in a language the machine reads, but the vocabulary held: bug for the fault, debug for the hunt.

Both come from an era when the fault could be a physical object lodged somewhere in a machine you could open with your hands. On one recorded occasion, in the middle of the afternoon at Harvard, it was.

The story survives because of that. There is no photograph of the annunciator at the U.S. Hotel and nobody kept whatever was wrong with it, and Edison's bugs were arguments and burnt contacts that left nothing behind. A moth can be taped to a page. Nearly eighty years later the thing everybody remembers about the word is the one occasion it was literal, which is the wrong way round, and is why the log entry has to keep explaining itself.

The moth is still taped to the page.

Sources

• US Naval History and Heritage Command, "The First Computer Bug, 9 September 1947", photograph NH 96566-KN, courtesy of the Naval Surface Warfare Center, Dahlgren (the moth found trapped between the points at relay 70, panel F, of the Mark II Aiken Relay Calculator while it was being tested at Harvard University on 9 September 1947, and the operators affixing it to the log with the entry first actual case of bug being found).
• Wordorigins.org, "bug (computer)" (the earliest known citation of bug for a machine fault, in The Operator, The Journal of Scientific Telegraphy of 15 August 1875, reading that the biggest bug yet has been discovered in the U.S. Hotel Electric Annunciator; the full text of the Harvard log entry including the time of 1545 and relay 70 panel F; the explanation that this was the first actual insect rather than the first defect and that the workers were using established terminology; and that Grace Hopper enjoyed retelling the story and was widely but wrongly believed to have coined the term).
• Thomas A. Edison Papers, "Edison Invents the Technical 'Bug'" (Edison's letter to William Orton of 13 March 1878 saying he did find a bug in his apparatus; his letter to William Preece of 19 May 1878 asking whether the static charge is the bug; his letter of 13 November 1878 describing months of intense working study and labour being required once bugs, as such little faults and difficulties are called, show themselves; and the note that the Oxford English Dictionary credits an 1875 journal reference as the earliest documented use).
• Wikipedia, "Harvard Mark II" (completion in September 1947 under Howard Aiken with Edmund Berkeley and Grace Hopper, financed by the United States Navy for ballistic calculations; delivery to the Naval Proving Ground at Dahlgren, Virginia in March 1948; an addition time of 0.125 seconds and a multiplication time of 0.750 seconds; and the extraction of the insect and its taping into the log on 9 September 1947).