The French intern strained to hear through the headphones. After half a minute, she nodded. “Yes,” she confirmed. “I hear it. He says: ‘… le six novembre [November 6th] en Haut de la Tour de Monsieur Eiffel [at the top of Mr. Eiffel’s tower].’” Then, she told us, she heard a violin playing faintly.
In those moments, in an audio lab in the subbasement of the Library of Congress, I was among a small group of listeners who learned the content of the old phonograph cylinder recording.Footnote 1 The unmarked record came in a box with two dozen others associated with William J. Hammer, Thomas Edison’s representative to the Exposition Universelle in Paris in 1889. For decades, the Hammer box sat mute in museum storage at the Smithsonian’s National Museum of American History (NMAH).Footnote 2 There had been clues as to what we might hear, among them a loose scrap of paper in the box with a terse handwritten note, author unknown: “Violin played on the Eiffel Tower Nov. 6, 1889.” But with the cylinder’s content now audible, key identifications fell into place. The recording, dating from the end of the Paris fair, originated at a party in Gustave Eiffel’s apartment on top of his grand monument and on a phonograph Edison had presented to the French civil engineer. For the event, Hammer reportedly operated the machine and recorded a number of invited artists, including the violinist, Henri Berthelier, who played a “Romance” by Henri Vieuxtemps.Footnote 3
The recording did not reveal its sounds while playing on an old machine, though. We listened to a computer sound file, a digital object instead of the old wax cylinder, extracted from the original recording with twenty-first-century state-of-the-art techniques and equipment dubbed IRENE. Remarkably, IRENE performed this feat without touching the cracked and brittle 130-year-old record.
Introduction
The immediate inspiration for this essay is not just that one cylinder, but three in the collection of sound recordings at the Smithsonian’s NMAH. In addition to the record created atop the Eiffel Tower, a second cylinder relates to a balloon ride over the Paris countryside that W. J. Hammer organized after the fair ended. Both cylinders were part of his personal collection. The third originated in Alexander Graham Bell’s well-equipped sound laboratory in Washington, D.C. Previously considered unplayable, all three yielded their sonic content for public release in 2019, thanks to IRENE: Image, Reconstruct, Erase Noise, Etc., each of the critical sonic recovery functions it performs.Footnote 4 Only after we recovered their sounds was it apparent that all were made within about a month of each other at the end of 1889. Created independently but considered together, these three records are material and acoustic evidence for crucial but overlooked events in 1889. They point to the co-evolution of two then new competing audio technologies—the phonograph and the graphophone—and unprecedented numbers of listeners (Figure 1).
Three cylinder recordings from 1889. Left to right, graphophone cylinder with recording of Alexander Graham Bell’s voice, Edison phonograph cylinder with recording of violin performance on the Eiffel Tower, and Edison phonograph cylinder associated with a balloon ride over Paris, France. Image: National Museum of American History.
This essay will juxtapose the three recordings from 1889 with the experience of hearing them again in 2019, thanks to IRENE. This pairing offers an opportunity to examine how each period envisioned the technical opportunities and social purpose of these new sound technologies in their respective times. Inspired by the work of Sheila Jasanoff and others who have developed the concept of “sociotechnical imaginaries,” what follows here focuses on the role of listeners’ imaginations and inquires how their notions shaped the meaning, use, and material aspects of recorded sound and playback.Footnote 5 This work also responds to a recent appeal in a special issue on sound archives in the journal Technology and Culture for more “carefully contextualized microhistories … of particular sound technologies under distinct historical conditions, specifying their social-spatial relations with actors, objects, sites, and settings.”Footnote 6 Using both physical records and sounds to supplement documentary evidence, this detailed look, first at 1889 and then 2019, aims to meet these microhistory conditions. A third influence on this essay are historical studies on interactions between inventors and technology’s users, with listeners constituting a special category of users.Footnote 7 As a contribution to sound studies, this comparative look manifests the field’s attention to both sonic phenomena and the means by which we come to know and experience those phenomena. The experience of interpreting audio revivals reinforces an important observation from sound scholars Joeri Bruyninckx and Alexandra Supper about the iterative nature of this kind of research: “knowledge through sound is always in the making, resonating between listeners, material artefacts and cultural contexts.”Footnote 8
In the case of the recordings from 1889, the availability of their sounds today stimulated a small project at the Smithsonian to document them as museum objects and yielded unexpectedly rich content and context. The research results, sketched in the next pages, go beyond descriptive cataloging to detail the circumstances around the recordings and sharpen our focus on listeners in sound playback’s earliest days—listeners who have been mostly overlooked in the grand, captivating, and often-told invention narratives. Until recently, most of what we know about the earliest audio recordings is not from the sounds themselves. Instead we have relied on written descriptions of sounds or listeners’ responses to them without fully understanding listening as a historical phenomenon.Footnote 9 The sounds recently recovered from 1889 will supplement those written descriptions and leave room for much-needed historical work beyond the brief accounts here. The phonograph introduced new categories of listening in public, but, as Kate Lacey has pointed out, “listening has been a curiously absent category in most treatments of media history and in most theorizations of the public sphere.”Footnote 10
In the case of the IRENE technology, with an invention story mostly unknown to a wide public, it is possible to probe the motives of the inventors, roughly gauge present-day listening appetites for nineteenth-century sounds, and speculate about future uses of IRENE. The ability to hear these cylinders today and the rich opportunities for historical research that come from looking at and listening to them, also make a powerful case for preservation. Together the two episodes from 1889 and today illustrate how two very different groups of listeners heard sounds in new ways, imagined what possibilities lay before them, and enthusiastically helped shape new directions for using sound recording.
Part 1: Listening to Sound Recordings in 1889
Recording and playback in 1889. The three audio artifacts of 1889 survive from a critical period of competition between two emerging technologies. In the late 1880s Thomas Edison’s organization promoted his phonograph, and Alexander Graham Bell’s Volta Associates championed their graphophone.Footnote 11 Kicking off the rivalry was the dramatic introduction of Edison’s phonograph in 1877, a prototype machine that recorded and reproduced sounds with a sheet of tinfoil. While there had been efforts to represent sound visually before this, never before had it been possible to play it back. The phonograph was a public sensation. Thanks to adoring press coverage and agents who demonstrated the talking machine far and wide, Edison earned the nickname “the Wizard of Menlo Park” and international fame as an inventor.Footnote 12 An outgrowth of Edison’s work on telegraph and telephone technologies, his “talking machine” was, in his initial view, fundamentally a device for storing and reproducing human speech, for “preserving the sayings, the voices, and the last words of the dying member of the family—as of great men.”Footnote 13 Edison’s preservation impulse undoubtedly revealed his age’s preoccupations with death and speaking to the future, as Jonathan Sterne has discussed at length.Footnote 14 Recordings also reflect the period’s cultures of collecting, with sound carriers the newest artifacts, the newest manifestations of the consumer’s material world. The nature of recordings as potent objects was evident from the earliest tinfoil phonograph demonstrations, when those in attendance received fragments of the recordings as souvenirs. That potency persisted as new types of sound carriers appeared and proliferated.Footnote 15
After about a year, Edison shifted his inventive focus to electric lighting, but not before suggesting many fanciful future uses of the phonograph in an 1878 article for the North American Review. He imagined that his new invention might be employed for dictating letters; holding content for talking books, student lessons, or advertising announcements; serving as the basis for toys or clocks; and in recording telegraph and telephone communications.Footnote 16 These weren’t all necessarily his own ideas. Although he credited no one else, Edison gave us a hint that others had made suggestions. The reception of the phonograph, he acknowledged, depended on “that peculiarity of the invention which brings its possibilities within range of the speculative imaginations of all thinking people,” [emphasis added] and “to the almost universal applicability of the foundation principle, namely, the gathering up and retaining of sounds hitherto fugitive, and their reproduction at will.”Footnote 17 From the very beginning, then, Edison recognized listeners played a role in shaping the protean technology.
Despite the phonograph’s astonishing ability to capture and reproduce sound, it had considerable shortcomings. Unsatisfactory sound quality and a fragile tinfoil record topped the list of its faults. One of Edison’s investors, Gardiner Greene Hubbard, prompted his son-in-law Alexander Graham Bell and the Volta Associates—Chichester Bell and Charles Sumner Tainter—to improve on Edison’s work. The associates devised a more durable recording medium—a wax-coated cardboard cylinder—and reengineered the machine to engrave the surface of the wax in response to sound waves, rather than emboss the surface in the style of Edison’s tinfoil machine. The associates obtained the basic patents in 1886 and, benefitting from their location in Washington, D.C., attracted the attention of three of the capital’s experts in stenography and reporting. The latter formed the American Graphophone Company to sell the Volta machines as mechanical stenographers.Footnote 18
The work of the Volta Associates provoked Edison to return his focus to sound recording. In short order came his “Improved Phonograph” and then his “Perfected Phonograph.” He too adopted the wax cylinder as the medium for recording and playback and explored several spinoffs. Most notable were a talking doll with a phonograph in its torso and a portable phonograph for military use. Examples appeared in his display at the Paris Exposition.Footnote 19
Commercialization began midway through 1888, when, in one of the oddest business episodes in U.S. history, both the Volta and Edison camps independently sold their invention rights to Pittsburgh entrepreneur Jesse Lippincott, who formed the North American Phonograph Company to promote both technologies as dictating machines for business offices.Footnote 20 Lippincott organized his business structure around regional member companies, each with its own sales territory and machines to lease at $40 per year. As these franchises started up, listeners consisted mainly of small groups of potential customers and the merely curious, who attended private demonstrations scattered across the United States.Footnote 21 But dictation machine rentals weren’t enough to sustain the business. “Sound recording,” according to David Morton, “was starting to look like a failed technology.”Footnote 22 That is until several of North American’s “sub-companies” altered their rental machines, set them up in public places like train stations and hotel lobbies, and loaded them with pre-recorded cylinders to offer entertainment for a nickel a play.Footnote 23 This form of public listening began as a trickle in 1889, when graphophones were first available for lease, and when cylinders, prerecorded by franchisees or by Edison’s operation to demonstrate the virtues of the machines as dictation equipment, prompted users to repurpose the machines for public mechanical entertainment. Radio would go through a similar transformation thirty years later, when, diverging from RCA’s vision of the medium as means for transmitting business-to-business Morse code messages, ham operators began airing phonograph concerts.Footnote 24
The fate of Bell’s graphophone was irrevocably intertwined with Edison’s phonograph. As sound archivist Sam Brylawski has pointed out, the differences between the two machines seem minor today, but just over a century ago, Edison’s interests squared off fiercely against the Volta Associates and their business representatives. For adopting the cylinder as a recording medium, Edison would ultimately lose in the courts, which handed patent priority of invention to the Volta Associates and their successor companies in the early twentieth century.Footnote 25 But in 1889, Edison seemed to have the upper hand, especially after his vigorous and highly publicized return to sound recording technologies and his triumphal public displays at the Paris Exposition (Figure 2).
Listening to the phonograph at the Paris Universal Exposition of 1889.
From Le Monde Illustré, September 14, 1889, 173.
Listening at the Paris Exposition. Just as small, scattered gatherings of listeners in the United States generated interest in and demand for recorded sounds, so did throngs in Paris at the Exposition Universelle in 1889. Running for six months from May 6, the exposition commemorated the French Revolution, celebrated the French Republic, and showcased the host country’s achievements.Footnote 26 The fairground space allocated to the United States was enormous, and it signaled French sympathies for another republic. America’s large footprint was also possible because many European monarchies chose to stay away from an event marking the overthrow of a sovereign and did not mount national displays.Footnote 27 Such nineteenth-century world’s fairs were a form of public spectacle, where instead of sovereign power the “power of capital” was on full display.Footnote 28 This one was no exception. Dominated by the imposing Eiffel Tower, the Paris Exposition in 1889 exhibited the technical optimism of the day and, in material form and extensive programming, advanced the ideals and rhetoric of progress, nationalism, and imperialism, all against the background of recurring French social unrest and political crises.Footnote 29 What was different from previous world’s fairs was the prominence of recorded audio to exercise that power over the huge, fair-going crowds.
It is impossible to say with precision how many of the thirty-two million visitors heard recorded sound during the fair’s six-month run, but all of them had an opportunity from the event’s opening day to hear phonographs and graphophones. The graphophone display, operated by Percival Lee Waters, was interactive in the sense that visitors could listen and record themselves. A modest affair, the display featured eight machines, outfitted with ear tubes for one listener at a time. It nevertheless had a prominent place near the Bell Telephone and Western Electric exhibitions in a section devoted to U.S. industry in the Palais des Expositions Diverses (Figure 3).Footnote 30
The graphophone exhibition at the Paris Universal Exposition of 1889. From Exposition Universelle de 1889, Smithsonian Libraries and Archives.
Edison had given William J. Hammer oversight of the phonograph portion of his extensive displays.Footnote 31 Hammer started the fair with one machine, but ultimately installed twenty-five in multiple locations.Footnote 32 A phonograph pavilion, purposefully built with foot-thick, sawdust filled walls, insulated those inside from the din of the open gallery. The space, meant for VIPs, was intended “for the reception and entertainment of distinguished guests.”Footnote 33 Most visitors listened on the public floor through ear tubes, sometimes as many as six sets per machine. Hammer estimated that between five thousand and fifteen thousand per day came by to listen. Another Edison phonograph specialist, A. Theodore Wangemann, counted fifteen hundred a day during his visit to the fair. Newspaper hyperbole from Edison himself put the figure at thirty thousand daily visitors.Footnote 34 Although the exact figures are unknown, these visitor estimates, even at the low end, stand in marked contrast to the small groups of listeners at demonstrations of the machine in the United States at the same time.
What did visitors hear on the recordings at the fair? Nothing is known of the graphophone’s sonic offerings, but the Edison content is well-documented. The Edison operation sent cylinders, each two minutes in duration, manufactured in the Edison works, and recorded in the Edison studio in New Jersey. Blank cylinders permitted on-the-spot recording at the fair as well, where a piano had been installed in the Edison pavilion. Like counting visitors, exact numbers of recordings are difficult to determine. In June, Wangemann arrived in Paris to help with the phonograph portion of the Edison display and do a promotional tour of Europe afterward. A log indicates the Edison operation shipped 654 cylinders to Paris for his use. But Edison wrote Hammer in May, perhaps optimistically, that he was ready to send two hundred pre-recorded with music then, and an additional hundred a week thereafter (Figure 4).Footnote 36Footnote 35
The Edison phonograph exhibition at the Paris Universal Exposition of 1889. William J. Hammer Collection, Courtesy of Archives Center, National Museum of American History.
Hammer asked for recordings of Edison’s voice to share with visitors.Footnote 37 But the cylinders, logged in New Jersey with a brief description of their content, were, in the assessment of music historian Annegret Fauser, “mostly music, with the repertoire consisting of the French and American national anthems, songs, short solos and ensemble pieces, and band music … a broad selection of light music with popular waltzes, polkas, and marches by, among others, Johann Strauss, Emil Waldteufel, and Franz Lehar.” Also included were recordings of “opera extracts and other classical repertoire that would have been played in any bourgeois salon of the time, whether in New York or Paris.” In contrast to historical accounts that Edison was against using the phonograph for music, Fauser argues his operation supplied plentiful recorded music for the fair.Footnote 38
Customarily, these early sound recordings began with an announcement of what listeners were about to hear, usually omitting the identification of the performers but assisting the listener in imagining who they were.Footnote 39 The Edison studio obliged with introductory statements in a variety of languages, catering to the mostly European fair visitors. Native speakers in the vicinity of the New Jersey works supplied announcements in French, German, and Italian for a set requested in August that year.Footnote 40
The fair got very little attention in the American press, beyond a notice of its opening in May, a few comments about lessons learned for the upcoming 1892 fair in the United States, and Edison’s vaunted visit in August. In contrast, the French press covered the fair and Edison extensively. The Volta Associates graphophone appears in some French accounts, but coverage of Edison’s August appearance at the fair overshadowed not just the graphophone’s modest display, but almost everything else at the fair except the Eiffel Tower—if, that is, Wangemann is to be believed.Footnote 41 Along with the phonograph, Edison’s association with electricity elevated his reception. The phonograph might have been Edison’s most recent innovation, but electricity held a higher cultural value, as the delivery system of power, progress, and modernity. During Edison’s visit to Paris, the fair’s official weekly publication proclaimed: “Edison est en ce moment le roi de Paris.” [At this moment Edison is the king of Paris.] (Figure 5).Footnote 42
Thomas Edison as lead story on the weekly Official Bulletin of the Universal Exposition of 1889, August 31, 1889.
What did visitors say about what they heard? Firsthand accounts of listening do survive, especially in the press. In Musical Encounters at the 1889 Paris World’s Fair, Annegret Fauser writes about the “realities of recorded sound” and their reception at the exposition.Footnote 43 In discussing public reactions to the fair’s sonic experiences, she relies on French journalists and finds that most of them praised the sounds heard from the machines. But in her view, they were reacting with enthusiasm not to the precise content, but “as listeners … to the sheer fact of sound reproduction itself.” The phonograph at the fair, according to Fauser, was a positive means to imagine the future of music and to consider recorded sound’s role in that future. Nevertheless, she also notes that for one prominent listener, the sounds from the phonograph were “an obstacle to the listening imagination.” Nineteenth-century science writer and historian of technology Louis Figuier complained of the machine’s disappointing limitations:
Whether one hears them through the acoustic tube or on open machines, the sounds of the orchestra, the choruses, and the words retain that timbre of a buffoon’s voice and that extraneous noise that remove all illusion and all charm. The listeners look at each other without daring to express their disappointment, but finding that the result did not live up to their expectation.Footnote 44
Two more cylinders. Recorded sound, regardless of its quality at that time, inspired ideas for future inventions and experiences. Considering two more cylinders supports this observation.
Edison’s representative at the 1889 fair, W. J. Hammer experimented with one such possibility when his duties in Paris were over. He arranged to do something he’d always wanted to—take a ride in a gas-filled free balloon. On November 14, 1889, in the company of two other Americans, he made that voyage, but not before nearly getting arrested. As Hammer and his associates prepared for their balloon ride, filling the balloon at La Villette gas works north of Paris, he remembered “some French military officers” arrived to accuse the assembled group of plotting espionage (Figure 6). Allowed to proceed after explaining their purpose and agreeing not to take pictures of the city’s fortifications, Hammer went aloft with his companions—Dr. Rufus G. Wells, an experienced balloonist from St. Louis, and Dr. Abbott L. Rotch, founder and director of a private scientific institution outside of Boston, the Blue Hill Meteorological Observatory. The trio took lots of instruments with them and did many experiments. They traveled for about four hours over about seventy miles, at an average speed of twenty miles per hour, at an average height a little more than half a mile. With a giant horn for amplification, borrowed from the recently closed Edison exhibit at the fair, the aeronauts shouted down to people below as they floated by.Footnote 45
William J. Hammer and his associates prepare for their balloon ride over Paris, France, in November 1889. Note the giant phonograph horn, borrowed from the Edison exhibit at the Universal Exposition, to communicate with people below as they floated by William J. Hammer Collection, National Air and Space Museum Archives.
During the balloon ride, Hammer dropped a dozen sealed boxes containing sound recordings attached to parachutes. He envisioned this technique could be used for military surveillance. “[A]n officer making observations from a military balloon,” he wrote, “could dictate such observations to a small phonograph attached to his side.” One of those portable phonographs—likely the only one ever made—survives at the Edison National Historic Park, and the Smithsonian has one of those recordings. The surviving cylinder, smaller than the standard record size, comes with a custom box, roughly a three-inch cube (Figure 7). On the box, an inscription reads: “Sent from my free balloon November 14, 1889 Paris from W.J.H,” and another inscription offers this paraphrased instruction: if found, please return to William J. Hammer at the address of his Paris hotel. This recording dates from eight days after the Eiffel Tower record. Despite a very large crack, the cylinder offers us a barely audible male voice. We provided a rough transcript of it in a National Museum of American History blog post. We speculated it might be a recording of the balloon’s launch, and solicited reader suggestions by posing the question: “Do you hear what we hear?”
[Cough] ? an Edison phonograph […] American industrial nation, Paris exposition 1889. Are you ready? Do not go yet. Bonjour messieurs et mesdames. […] Into the […] phonograph […] everybody on the ropes. Lift off! […] Lift off, lift off […] lift off […]Footnote 46
We received one response that offered a convincing transcript for a portion of the sound file:
A cylinder William J. Hammer recorded for his balloon ride on November 14, 1889. National Museum of American History.
In 1888, Edison had written about the virtues of wax records and their ability to “speak for themselves” without waiting “dumbly for centuries to be deciphered” like the cuneiform cylinders of the ancients. He couldn’t have predicted how time would exact a heavy toll on the waxy surface grooves.Footnote 48 “Deciphering” is exactly what is now required to listen and understand, because of surface irregularities and other forms of degradation in the years since these cylinders were recorded.
Back in the United States, Alexander Graham Bell had resumed experimenting with recordings after a few years’ break. The third cylinder discussed here, recorded on a graphophone cylinder dated December 6, 1889—roughly a month after the balloon recording— survives from Bell’s laboratory in a former carriage house behind his father’s home in Washington, D.C. Bell recorded himself, his father Alexander Melville Bell, a man who identified himself as David Hess or Haas, and perhaps a fourth speaker. The first words of the recording identify location and Bell’s voice:
Eighteen hundred eighty-nine, December sixth, the phono-graphophoneFootnote 49 has made its appearance in the Volta Laboratory in Georgetown, D.C., for the first time on this day. How does it come out? Alexander Graham Bell.Footnote 50
Later on the recording, Bell speaks again and whistles a tune. The graphophone recording is by far the clearest of the three cylinders considered here, but it is impossible to say whether this is because it had better sound quality initially or because it withstood the passage of time better than the two Edison cylinders.
Although Bell has the name recognition in this story, most of the technical credit for the graphophone belongs to Charles Sumner Tainter, who started as Bell’s key experimenter in the D.C. lab in 1880 and stayed with the project long after Bell and his cousin departed. Tainter, a talented instrument maker in his own right, saw the cylinders and machines into production in Bridgeport, Connecticut, and continued to make improvements in both until his health failed. He eventually went to San Diego.Footnote 51
Documentary evidence supplements the content of the Bell cylinder. In his notebook, Bell wrote that he had resumed audio experiments and had been exploring ways to make multiple copies of a single recording. At that time, all cylinders—whether Edison or Volta—were recorded one at a time. Entries beginning in February 1890 indicate that Bell’s work shifted from cylinders to discs, because it was easier to make multiple copies from the latter.Footnote 52 The shift likely had two origins—not just in competition with Edison, but with Emile Berliner, too, who would eventually have success in patenting and making disc records commercially practical. Whatever his immediate inspiration, Bell longed for a new project and dreamed of concrete technical solutions to the simultaneous recording of multiple copies.
Part 2: Listening to IRENE’s recovered sounds
Material things. Three cylinders described above survived from 1889 until today mostly by chance. The two Edison cylinders, part of Hammer’s personal collection of technical and scientific objects, passed at his death in 1934 to his daughter Mabel. IBM acquired them from her and donated them to the Smithsonian in 1962.Footnote 53 The Volta cylinders had a more direct route from Alexander Graham Bell to the Smithsonian’s U.S. National Museum, largely due to his long service to the institution as one of its governing regents. The museum holds roughly four hundred objects pertaining to sound recording and playback—discs, cylinders, and equipment—from Bell’s Volta Laboratory.Footnote 54
These sound carriers ended up in the object collections at the Smithsonian, instead of in archives or library-based “special collections.” It is likely the staff at the time considered these materials artifacts because of their associations with prominent inventors and their status as audio incunabula, that is, artifacts from the earliest days of sound playback. They survived as historically important objects because of that designation. Removing an object from any museum collection once it has been “accessioned,” that is, legally acquired, has been an infrequent and often controversial practice. On the other hand, badly broken records from the period, considered unplayable and useless to researchers, mostly ended up in the trash rather than in archives.
The records’ classification as artifacts, not archives, also meant they received attention from the object conservators at the NMAH, who assisted beginning in 2011 in preparing the objects for sound recovery and important research. This was an essential step before the cylinders yielded their content. In the case of the three records discussed here, the Eiffel Tower cylinder in the Hammer box posed the biggest challenge. Most of the twenty-five cylinders were stuck in place on the box’s wooden pegs, originally designed to protect the fragile wax records from touching each other while in transport and storage. But over the years, the pegs swelled and effectively trapped them. Senior objects conservator Beth Richwine removed most of the recordings from their pegs simply by letting the wood shrink in a room with lower humidity. Mary Wilcop devised methods to clean and repair the cylinders, performed materials analysis, investigated the white discoloration on some of the recordings, and saw them safely packed and transported to and from the Library of Congress, where they underwent IRENE’s recovery process.Footnote 55
Of the other twenty-four cylinders in the Hammer box, we scanned fourteen in April and May 2018 at low resolution to determine whether they held recoverable audio. Twelve of those proved viable and now await rescanning at higher resolution to obtain sound files suitable for public release. Eleven other cylinders from the box still need repairs to ensure safe handling before scanning.
IRENE. The NMAH cylinders’ survival depended mostly on luck and the essentially conservative nature of museum collections practices. Hearing them, well past the time anyone alive today thought possible, depended on the development of a specialized digitization process, a set of state-of-the-art techniques and instrumentation meticulously applied, all without touching the surface of the recording.
That process emerged by chance, too. Physicist Carl Haber first learned about the global problem of endangered audio while listening to a radio interview with Mickey Hart, a drummer for the Grateful Dead and a passionate preservationist for the music of Indigenous peoples around the world. Haber, an experimental physicist at Lawrence Berkeley National Laboratory (LBNL), wondered if the techniques he used to image microelectronic devices—most recently in developing silicon particle detectors for the ATLAS Project at CERN’s Large Hadron Collider outside Geneva, Switzerland—would work on the grooves of fragile recordings. Beginning in 2003 Haber, LBNL postdoctoral fellow Vitaliy Fadeyev, and other collaborators, especially Earl Cornell at LBNL and Peter Alyea at the Library of Congress, adapted physics instrumentation to make images of sound by scanning the record’s surface at extremely high resolution, creating a three-dimensional map, and, with custom algorithms and computer software, running a virtual stylus through the map to create a digital sound file.Footnote 56 The devices and techniques they assembled to accomplish this are collectively called IRENE (Image, Reconstruct, Erase Noise, Etc.), in tribute to the first recording Fadeyev and Haber scanned—the Weavers’ 1950 version of “Goodnight, Irene.” Just as work on the telegraph and telephone conditioned developments in sound recording, innovation in one field unexpectedly benefitted another (Figure 8).
IRENE at Library of Congress scanning an experimental wax disc recording with the voice of Alexander Graham Bell, made April 15, 1885. Author’s photograph.
Considerable U.S. federal support has moved IRENE’s development forward over the past two decades, but IRENE systems are still rare.Footnote 57 There are now six on institutional campuses—at LBNL; the University of California, Berkeley; the Roja Muthiah Research Library in Chennai, India; the Library of Congress, where there are two; and most recently, the Moscow State Conservatory in Russia. As of 2014 there is a seventh IRENE: the Northeast Document Conservation Center in Andover, Massachusetts, provides IRENE’s services to revive recordings for paying clients. And soon there will be an eighth. Thanks to a Save America’s Treasures challenge grant from the National Park Service and private donors, the Smithsonian’s National Museum of American History is building its own IRENE system.Footnote 58 The full story of IRENE’s origins and development over the past two decades is beyond the scope of this essay, but it has all the elements of most good invention yarns—a flash of insight, relentless hard work, numerous collaborators, jealous rivals, powerful advocates, and grateful beneficiaries.Footnote 59
There are other audio recovery technologies available, both optically based and modern versions of stylus transfers.Footnote 60 With alternatives that depend on running a stylus over the original recording’s grooves, access to the content of historical sound carriers has often been at the expense of preservation. When the content takes precedence, the sound carrier is then disposable. IRENE’s chief benefits are the system’s noncontact characteristics that avoid further damaging the historic recordings and the software’s ability to reconstruct damaged records, in the process reducing noise and estimating with algorithms where sounds should be where the surface is too damaged to reproduce the original.
So then what, technically, are we listening to when we hear sounds recovered with the IRENE system? A grooved disc or cylinder is an analog sound carrier, where walls of the groove represent the original audio’s wave form as engraved or embossed by a moving stylus in response to the original sound.Footnote 61 The IRENE process creates a high-resolution digital map of the surface of a disc or cylinder. Then, processing the image may involve the optional removal of certain sources of noise or distortion, that is, the imperfections in those grooves—dust and dirt particles, mold, scratches and cracks—that would otherwise create pops and clicks in the digital recording if not cleaned away.Footnote 62 Finally, software calculates the motion of a virtual stylus moving through the virtual record’s grooves as represented by the map, reproduces the audio content, and makes a standard digital sound file (.wav).
The process that identifies and removes discrete noise sources, as Carl Haber has explained to me, may then apply an algorithm to insert signals in place of the missing or removed data. The algorithm analyzes the audio signal generated from the surface just before and after the missing or noisy data. Then, based on a mathematical rule, it compensates for the flaw with sound that makes sense in context when constructing the output sound file. In this way IRENE acts as a technical interpolator, much like analogous processes that remove scratches or other blemishes from images.Footnote 63 Studies of perception show that human listeners can similarly construct continuity, that is, an understandable stream of sound from cacophony or interruptions, in a process that synthesizes multiple sounds into coherence.Footnote 64
Modern ears, accustomed for over a century to hear increasingly clear sound recordings, may have difficulty comprehending recorded sounds from 1889, even when the IRENE process does its best work, because the original recording process or years of surface degradation may have occurred. A recording’s surface, including cumulative ills, is what we hear from an IRENE-recovered digital sound file.
Decisions about technical matters influence what we hear in digital files in the most fundamental ways. One key specification for reformatting analog sound carriers to digital files that is crucial for determining what we hear is the sampling rate, that is, the number of periodic samples of the source material taken to reproduce it. Audiophiles, librarians, and archivists initially distrusted digital files for sound preservation purposes because the first compact discs in the 1980s came with significant data compression and, for knowledgeable listeners, consequent loss of the original sounds and degradation in sound quality.Footnote 65 There is now general consensus that the standard compact disc (44.1 kHz) is “inadequate,” but debate continues about how high the sampling rate should be for digitally reformatting analog recordings and audio tapes.Footnote 66 A growing agreement is that 96 kHz is a workable preservation standard. Such disagreements over “the ground rules of reproducibility,” as Jonathan Sterne pointed out, began with the technology in the 1870s and continue to this day.Footnote 67
What categories of recordings have IRENE’s listeners heard? Since 2005, Haber and his collaborators have retrieved sounds with IRENE from an array of acoustic carriers, including phonautograms, the paper tracings from the experimental “sound writer” of Leon Scott de Martinville; Edison tinfoil recordings; commercial discs; recordings of Kwakiutl culture made by Franz Boas electrotyped from cylinders for the Berlin Phonogramm Archive; aluminum transcription discs of Slavic folklore archived at Harvard University; lacquered discs from radio broadcasts and others from personal recordings; and plastic dictation belts. IRENE has reanimated the voices of Jack London, Howard Hughes, Janis Joplin, and Alexander Graham Bell.Footnote 68 The process was especially suited to the custom work needed to revive a sampler of experimental records made between 1881 and 1885 in different formats and materials at Bell’s Volta Laboratory.Footnote 69
Listeners’ observations about these diverse sounds can be found in global press coverage, in general comments on the internet when the sounds are posted for public access, and from academic researchers. The discovery of a recording of Alexander Graham Bell’s voice generated intense public interest.Footnote 70 Likewise, reactions to the largest collection of sounds recovered to date, the Phoebe A. Hearst Museum’s nearly three thousand cylinders carrying native Californians’ song and speech, indicate a present-day appetite for these sounds from scholars and contemporary Native communities.Footnote 71
Today’s listeners have added a layer of meaning that the collecting anthropologists didn’t intend, according to Rebecca Lemov, a researcher remarking on the ability to listen to the voice of Ishi, a California Yana who lived for a time at the University of California Museum in San Francisco until his death in 1915. Rather than using the recording to analyze linguistics in the face of the Native speakers’ disappearance, Lemov writes, “the value of these sound archives was to connect emotionally with Ishi, to resurrect his voice, despite the mediations of time and format.” For a class she was teaching on the history of anthropology, she discovered:
There is something radically different about being able to summon Ishi’s voice at will and with little effort even in fragmentary form…. The significance of this cannot be overstated. Even as Ishi has become less known—his photographic representations, physical artifacts, and bodily remains increasingly disputed—the sound recordings have taken a different trajectory: one that promises immediacy, a sense of direct, resurrected presence.Footnote 72
Brian Hochman, revisiting Jonathan Sterne’s discussion of the earliest use of sound recording for ethnography to capture Native languages on the edge of extinction, describes a similar vitality in an online Passamaquoddy Native culture site with sound. Hochman concludes that “the online archive functions as a creative platform for the persistence of Passamaquoddy culture rather than as a technological memorial to its expected extinction.”Footnote 73
Because to date so little audio has been revived from this earliest period of playback, it is difficult to predict what value the sounds, once thought lost forever, might have and to whom. Since the Smithsonian’s sound files are posted for open access, for example, we have only an incomplete answer to a key question: What, specifically, do these sounds add to our historical knowledge? There are some indications that the fresh sounds from Alexander Graham Bell’s Volta Laboratory, retrieved by IRENE in 2011 and 2013, have captured the attention of literary scholars in search of insights on Shakespeare and nineteenth-century poets. Historians and media scholars have used the files for research and coursework. Not limited to academic users, Volta sound file links appear on websites of phonograph enthusiasts, too.Footnote 74 From this small sampling of present users, it is easy to conclude that an expanded supply of early sounds might have continuing interest to wide audiences for multiple uses.
The present-day physicists and technicians working with IRENE imagine a future in which our sonic cultural heritage, in all its various specialties, is saved before it disappears, preserved digitally, and shared widely for listeners to make what they will of the sounds. Efforts to develop noncontact methods like IRENE, according to Haber, are intended “to get access and to share research.”Footnote 75 Another of his goals is to establish general mass-digitization techniques that “replace the need to maintain a variety of legacy playback systems in the future.”Footnote 76
This optimism prevails despite a bleak outlook for recorded sound preservation. Some have even called the state of our audio heritage a crisis.Footnote 77 Recent studies have sketched the degenerating physical condition of many analog recordings in the United States, the dire state of recorded sound preservation, and the barriers to improvement. One analysis estimates that there are over 250 million preservation-worthy audio holdings in U.S. collections, with 80 million of those requiring specialized attention. Pessimists predict that preservation for all these worthy items might run to millions of dollars, not counting expenses for digital storage and access to the files.Footnote 78 Real numbers for the cost of digitization are developed on a case-by-case basis because the work is customized to the condition, format, and quantities of recordings to be digitized. Developing an estimate can be a costly endeavor, all on its own, especially if institutions have not surveyed their holdings. Mike Casey, who heads audio-visual preservation services at Indiana University, summarized the problem in sobering terms: “archives hold very large numbers of analog and physical digital recordings on obsolete audio and video formats that are actively degrading, some of which contain content with high research value. We have a relatively short time window to save these recordings.” Footnote 79 And the difficulties are international, with wealthier nations more likely to organize and fund preservation efforts while smaller scattered collections around the world disappear. “Without them,” wrote Dietrich Schüller, “our view of the cultural and linguistic diversity of mankind would be incomplete. Their loss would mean a substantial deprivation of cultural, linguistic, and ethnic minorities in terms of their heritage, their history and their identity.”Footnote 80
As old media deteriorates and new media technologies change relentlessly, preservationists and institutional archivists face challenges unimagined just a few years ago. They are on a quest, trying to reach an impossible preservation goal as they face circumstances indescribably difficult to overcome. They are making hard decisions about what to save, what methods to use, and how to pay for the significant costs when funds are short for the work of cultural salvage. Some collaborative endeavors have begun, but at this time there is no single organized effort beyond widespread acceptance that digitization is—for the time being—the path toward the future.Footnote 81 Whatever preservation innovations may come, what matters is how we choose to use them. If the phonograph of 1889 epitomized the optimistic view of technological innovation and progress, today we are living with the unanticipated consequences of our widespread ability to record and play back sounds. We now have so many examples, and we have inadequate means to preserve them.
Researchers have a stake in the fate of sources. Although historians prefer to study the past, we can start by envisioning what it would be like to lose the sound sources our work depends on and then act in concerted ways across disciplines—with scientists, archivists, conservators, preservationists of all sorts—to craft a future in which sources endure and do so in less endangered ways. This, by the way, is not a new idea. From the earliest days of sound recording, it was clear the medium was ephemeral, and the search for a better record began immediately after Edison’s tinfoil ushered in the playback era. The American Graphophone Company, for another example, had an “indestructible” record in the works in 1904, a record meant to last “thousands of years” and to preserve a national collection of voices.Footnote 82
Nor is preservation an ideal without complications. Jonathan Sterne has written that now there is simply “too much to collect and not enough sense of, or agreement about, what should be collected.” Once collected, those sounds may or may not be preserved. Without giving up on “the responsibility to preserve or to remember,” he reminds us that losses of recordings, like losses of other forms of documentation, are inevitable, and such losses are characteristic of writing history. Writing history, he argues, has always depended on fragments of information, “available for present analysis in extremely skewed and partial form.”Footnote 83
The consequences of degraded or lost sources extend beyond professional needs for sources to write history and touch on larger cultural questions. Historian Abbey Smith Rumsey, writing about human cultural memory in the digital age, reminds us that wiping out evidence of culture “became a central strategy in subduing civilian populations” during the wars of the twentieth century. Hopeful that digital information gives us opportunities to use old sources in new ways, she takes heart that IRENE can help extract information from fragile fragments of our cultural history. She adds her voice to those who advocate taking steps to ensure that our shared knowledge endures and remains accessible. Her call to action is “to imagine how we will reconstruct our memory systems to accommodate abundance,” the bounty of information in the digital age. Her book When We Are No More reminds us that because we remember the past, we can imagine the future.Footnote 84
Acknowledgments
Thanks to the IRENE group—Carl Haber, Peter Alyea, Earl Cornell, Mary Wilcop, Shari Stout, Morgan Burgess, Beth Richwine, and Janice Ellis—for their work and the Grammy Foundation and Smithsonian Women’s Committee for funding that work. Thanks also to Edison Papers staff—Paul Israel, Daniel Weeks, and Louis Carlat—for thoughtful advice, to Kristen Frederick-Frost and Karen Hanson for comments on a draft of this article, and to this issue’s helpful editors and peer reviewers for suggestions.
