“Une trace impérissable de ces fugitives mélodies”

The earliest known recording of the human voice from which intelligible sound has been recovered was made by the French scientist Édouard-Léon Scott de Martinville on 9 April 1860—seventeen years before Edison’s first phonograph recording. It is a bit of “Au clair de la lune”.

On 20 April another recording, longer and with better fidelity, was made. Earlier recordings exist, but for lack of calibration they have not been converted. Scott calibrated his later recordings with a tuning fork of known pitch.

The recovery was carried out by First Sounds, a group devoted to preserving, recovering, and publishing old recordings.

Below is a sketch from Scott’s patent application (“brevet d’invention”), dated 25 Mar 1857.

Source: Scott de Martinville, Brevet d’invention et Certificate d’addition p17; digitized by David Giovannoni (FirstSounds.ORG). Sound-bell at left, cylinder (coated with soot) at right. Note the stylus at the right end of the sound-bell; when recording, this would be applied to the cylinder so as to scratch away the soot.

Scott first described his techniques in Principes de phonautographie, a seven-page document deposited at the Académie des Sciences on 26 Jan 1857. It seems to have been written in part to establish priority, and to allot credit, because he had recently hired a “constructer” to build the device (and, one suspects, to overcome problems of realization that Scott himself had failed to solve). His preamble merits an extended quotation. If I were of the prophetic school of historians, I would say it was ripe with implications for the future.

Y a-t-il possibilité d’arriver, en ce qui concerne le son, à un résultat analogue à celui atteint dès à présent pour la lumière par les procédés photographiques? Peut-on espérer que le jour est proche où la phrase musicale, échappée des lèvres du chanteur, viendra s’écrire d’elle-même sur un papier docile et laisser une trace impérissable de ces fugitives mélodies que la mémoire ne retrouve plus, alors qu’elle les cherche? Pourra-t-on, entre deux hommes réunis dans un cabinet silencieux, faire intervenir un sténographe automatique qui conserve l’entretien dans ses plus minutieux détails, tout en s’accommodant à la vitesse de la conversation? Pourra-t-on conserver à la génération future quelques traits de la diction d’un de ces acteurs éminents, de ces grands artistes qui meurent sans laisser après eux la plus faible trace de leur génie? L’improvisation de l’écrivain, lorsqu’elle surgit au milieu de la nuit, pourra-t-elle se retrouver le lendemain avec sa liberté, cette indépendance complète de la plume, instrument si lent à traduire une pensée toujours refroidie dans sa lutte avec l’expression écrite?

Je le crois. Le principe est trouvé. Il ne reste plus que des difficultés d’application, grandes sans doute, mais non insurmontables dans l’état actual des arts physiques et mécaniques.

Scott de Martinville, Principes de phonautographie, 1857; from Firstsounds.org).

[My translation:] Is it possible to arrive, with respect to sound, at a result analogous to what has been attained at present for light using photographic methods? May one hope that the day is near when the musical phrase, leaving the lips of the singer, will inscribe itself on docile paper, leaving an imperishable trace of those fugitive melodies that memory can no longer recall when it seeks them? Will we be able to place between two men within a silent chamber an automatic device that will preserve their conversation in its minutest detail, even while accommodating itself to the speed of speech? Will we be able to preserve for future generations various traits of the eloquence of one of those great actors, those great artists who die without leaving behind them the faintest trace of their genius? Will the improvisation of the writer, arising in the middle of the night, admit of being retrieved the next day in its freedom, its entire independence from the pen—that instrument so slow in translating a thought which loses its ardor in its struggle with written expression?

I believe so. The principle has been found. There remain only difficulties of application, no doubt significant, but not insurmountable given the current state of the physical and mechanical arts.

Scott—who to my knowledge never managed to play back any of his recordings—is imagining, for the reader whom he hopes to persuade of the importance and originality of his project, the benefits of a record of sounds, an extension of memory and the stenographer’s pen. Sound and song proverbially exist only in the moment: Scott promises that that transient being will effect its own trace in “docile” (that is, teachable) paper. His device is a phonautograph, a “sound self-writer”, but interestingly enough his model for the active portion of the sound-bell is the typanum, the ear-drum, nature’s own phonautograph, by which sounds impress themselves eventually upon the brain. He has managed, as the ear does, to make the vibrations of the air—and not, like his predecessors, merely the vibrating source of sound—leave their trances in a durable medium.

Philosophers fuss about theory much more than experiment, and even when (as increasingly they have) they turn their attention to experiment, they tend to idealize—to ignore the apparatus, the laboratory, the craft knowledge of those who build the apparatus (Scott has hired an “habile et savant constructeur” to realize his plans) in favor of the results, that is, the recorded observations with which theory can be sensibly confronted. The phonautograph itself neither confirms nor refutes any theoretical claim; but—so Scott promises—to “submit to vision and to measurement by precision instruments” the aerial vibrations of sound will compensate for the limitations of our hearing. It will permit the counting of those vibrations and eventually the discovery of the laws of timbre (p4).

No doubt I’m not the first to notice that the nineteenth century is the century in which nature began to write her own book. A bit of searching suggests that self-registering instruments appear in the 1790s (a self-registering thermometer and barometer are both described in that decade). That revolution—for so I think it deserves to be called—took place independently of the great theoretical developments of the period. It gave us first those instruments that inscribed what we now call a graph of their measurements of some quantity like temperature or pressure—instruments that permit nature to represent itself still only symbolically—and then photography and phonography. So basic are these techniques that we hardly notice them, or the great novelties they have introduced into the production of knowledge. Descartes’ world, and even Kant’s, was one in which nature, so to speak, might well “lose its ardor” (se refroidir, “cool off”) in the difficult passage through a human mind to a written record. In the world of Einstein, the “observer” has become nature itself, busily accumulating mountains of inscriptions and images.

▶ My source for this item was Anecdote du Jour. On the measurement of temperature in general, see Hasok Chang, Inventing temperature (Oxford, 2004) and Jan Golinski, “‘Fit instruments’: thermometers in eighteenth-century chemistry” in Frederick L. Holmes and Trevor H. Levere, eds. Instruments and experimentation in the history of chemistry ((MIT, 2000); on machines that reproduce the sounds of speech or that speak, see Thomas L. Hankins and Robert J. Silverman, Instruments and the imagination (Princeton, 1995). On “the control of the representational process by automatic means”, see Lorraine Daston and Peter Galison, Image and objectivity (Zone, 2007), esp. c3.