Mary, at Several Speeds

When Thomas Edison's phonograph reached the offices of Scientific American in December 1877, it asked the editors about their health. It also asked how they liked the phonograph and wished them good night. A dozen or more people gathered round the little machine on the desk. Someone turned its crank; metal spoke.

The editors could explain the process. Their report described the diaphragm, the points touching the recording surface and the cylinder travelling along a threaded shaft. They could also hear its limitations. Many words were indistinct, even though the inflections belonged unmistakably to a human voice. The phonograph was convincing before it was consistently clear.

It had emerged from a laboratory already occupied with sending messages. At Menlo Park, New Jersey, Edison and his assistants worked on telegraphs and telephones: devices that transmitted words, and mechanisms that could mark a moving surface. In July 1877, Edison considered attaching an embossing point to a diaphragm and letting the vibrations of speech indent rapidly moving paraffined paper. Could the marks give the sound back?

Charles Batchelor, his chief experimental assistant, later remembered how much of that question depended on someone's hands. John Kruesi adapted parts of an automatic telegraph and soldered a needle to a diaphragm. Batchelor cut strips of paraffined paper, set the point to press lightly and pulled the paper beneath it while Edison spoke.

Batchelor recalled a barely recognisable fragment of “Mary Had a Little Lamb” coming back, followed by more experiments and better articulation. He wrote this account around 1905–08, nearly thirty years afterwards. It describes the paper trials, before the familiar cylinder machine, and gives the work to several people: one made the parts, another prepared and moved the recording surface, another supplied the voice.

By late November, a cylinder design had been assigned to Kruesi. He turned drawings into a grooved metal drum on a screw-threaded shaft, with a crank to rotate it. Batchelor tested and adjusted the completed instrument. This was the form Edison, Batchelor and Edward H. Johnson took to Scientific American on 7 December.

A groove beneath the voice

A sheet of tinfoil covered the cylinder. Beneath it ran a spiral groove, leaving a narrow path where the foil had no solid metal immediately behind it. The operator turned the crank; the shaft's thread moved the cylinder sideways as it revolved. A point resting against the foil could therefore follow one continuous spiral rather than passing over the same ring again and again.

Speech supplied the movement in the other direction. Sound entering the mouthpiece shook a thin diaphragm. A point attached to its centre moved with it, pressing small indentations into the foil above the groove. The pattern recorded changes in the diaphragm's movement.

The machine described that December had a separate reproducing diaphragm and point on its opposite side. With delicate spring pressure holding this point against the indented foil, another turn of the crank reversed the chain. The marks moved the point; the point shook the diaphragm; the diaphragm disturbed the air. The listener heard the result.

The cylinder also carried time. Turn it too slowly on playback and each vibration arrived farther from the next, lowering the pitch and stretching the words. Turn it faster and the voice rose. Scientific American suggested regulated clockwork to keep recording and playback at the same rate. Until the drive became dependable, the person at the handle helped determine who appeared to be speaking.

A lecturer tries again

At New York's Chickering Hall on 23 March 1878, Professor J. W. S. Arnold had a roomful of listeners and a stage crowded with apparatus. His assistant, Dr Miller, rearranged instruments while the audience waited. Twenty minutes after the appointed hour, people began stamping their feet.

Arnold eventually reached the machine they had come to hear. According to the next day's New York Times, he fitted an already indented strip of foil, fastened its edges and held a paper funnel at the opening. The first playback produced wheezing sounds that nobody understood.

He returned the cylinder to the beginning and tried more slowly. “Mary Had a Little Lamb” became recognisable, in a voice the reporter compared to an old man speaking with water in his mouth. A third, very fast turn made the same rhyme shrill. One recording had supplied several speakers.

A fresh sheet did better. Arnold talked, shouted and sang into it, and the phonograph repeated his performance. After the lecture, much of the audience came onto the stage for a closer introduction. Miller used an already recorded strip and added interruptions to the rhyme, recording orders to be quiet or go to bed among Mary's travels. On the next playback, the machine delivered the altered version.

A closer look · 9:52 · English · automatic captions

Recreating an 1878 phonograph exhibition

Play film

The repertoire suited an instrument still learning to be intelligible. Familiar rhymes made missing words easier to recognise; spelling, numbers and short greetings offered manageable tests. At a Chicago demonstration on 22 May, electrician George H. Bliss admitted that he was inexperienced at operating the machine. He leaned over the mouthpiece and shouted. The alphabet came back clearly, while a longer sentence broke up.

The Chicago Tribune followed the uneven progress. A rendition of “Yankee Doodle” initially returned a jumble; a second attempt yielded a recognisable tune and most of the words. Cornet playing was judged perhaps the best result of the afternoon. The machine's small diaphragm also cracked. A demonstration could contain both a successful performance and a damaged part.

A library on metal sheets

At Menlo Park, Edison was already imagining a much larger repertoire. A World reporter's visit, reprinted in Scientific American Supplement in April, found him singing with a workman called Martin. They used a double mouthpiece for “John Brown's Body”, Edison taking the tune and Martin the bass. Only occasional distant bass notes returned. Edison thought Martin had not sung loudly enough; another recording on the same foil improved matters, though the performances did not align everywhere.

Downstairs, amid working lathes and machines, Edison showed the reporter another model: a circular plate with clockwork in place of the hand crank. He outlined a publishing business in recorded concerts, songs and novels. Listeners would keep an album of sheets, select a composition and let the instrument perform it. Young people could dance without requiring anyone to sit out and play the music.

These were plans, rather than a description of a service already available. The same interview that promised an album for the parlour recorded the difficulty of getting two men into a machine in the laboratory. Its mechanical format was still changing, and a successful impression needed to survive removal, handling and another encounter with the point.

The rhyme that became attached to Edison's beginnings is itself heard through a later performance. The surviving recording of Edison saying “Mary Had a Little Lamb” dates from 12 August 1927, the phonograph's Golden Jubilee. He was demonstrating how he remembered making his first record. The 1877 recording had not been saved.

At Chickering Hall in 1878, preserving the evening's performances had not been the final concern. When the visitors had heard enough, Miller tore the used foil into pieces and handed them out. The audience scrambled for them.