Noise Underground: Transduction and Infrastructural Interference in German Deep-Shaft Coal Mines

Abstract

Infrastructures built to move goods, people, and information across distance support the functioning of everyday tasks, logistical networks, and industrial lifeworlds across both the Global South and the Global North. What work does noise perform in their functioning? How does noise matter for their design? This paper suggests that noise within infrastructures elicits knowledge and alteration that it recursively eludes, but that such epistemes of representation and intervention are historically specific to the transductive chains through which noise manifests itself as interference. Working across two different historical examples, this paper compares the incommensurable ways that engineers and scientists understood and addressed the problem of noisy interferences within signaling systems engineered for communication within the deep-shaft coal mines of Germany's Ruhr in the 20th century. This comparison requires decoupling the concept of noise as an audible phenomenon from its disruptive capacity manifested as interference within the transmission of signs, two aspects of noise often conflated. More broadly, the paper proposes a heuristic of transduction as an alternative framework for STS studies of infrastructures—one that shifts attention away from their elemental components (copper, steel, concrete) toward how noise within sociotechnical systems involving energy conversions and signal transmission comes to matter.

Introduction

Noise poses challenges for knowledge insofar as it is often construed either as the opposite of desirable sound (Mody 2005), signals (Siegert 2015), or speech (Mills 2021), or as disruptive to them. Noise calls attention to itself when it disrupts infrastructures, spaces, and protocols for producing knowledge and for the problems it poses to control, order, and semiosis (sign processes). ¹ In much of the scholarship on noise across anthropology and science & technology studies (STS), noise is understood as an audible phenomenon that can be conceptually framed as polluting or interruptive. The trouble that I have with this literature is that it often conflates the aesthetic and physical aspects of noise as something audibly registered with its disruptive qualities and capabilities. This conflation does not enable historians or ethnographers of science and technology to distinguish how noise acting as interference within communicative or sociotechnical systems and knowledge infrastructures may not always be sensed as something heard.

In what follows, I propose the notion of transduction as a heuristic for thinking through noisy interferences as an abiding feature of infrastructures, both environmental and epistemic. Such a heuristic helps to disaggregate the work of noise as an audible or physical phenomenon from its constitutive role within infrastructures, which are sociotechnological ensembles that require both material and energy to move information, goods, and people across space (Larkin 2013, 176). The concept of transduction, broadly understood as the process of converting energy or information from one form to another, also offers a compelling approach with which to better analyze the historical specificity of resistances and interferences within such infrastructures. To date, the operation of various kinds of energy transfer and signal transduction, integral to making infrastructures work, has been largely overlooked in scholarly work on the epistemic and political matter of infrastructures (Anand, Gupta, and Appel 2018).

In this article, I deploy transduction heuristically to draw attention to the noisy aspects of signaling infrastructures used in the deep-shaft coal mines of the former West Germany. To develop this argument, I draw upon two empirical examples of interferences within signaling systems to show how such Störungen (disruptions) became problematized as epistemic things (Rheinberger 1997) incommensurable with one another. This discontinuity foregrounds attention to transduction as a key analytic for parsing the historical specificity of noise in knowing more-than-human technical assemblages, one that shifts analytical attunement toward more multimodal understandings of how infrastructures become perceptible in action. I suggest that transduction offers a mode of tuning into the indeterminacy and leakage of infrastructures, not as aberrations but rather as constitutive components of their functioning. As such, it offers important insights for STS scholars interested in thinking about the noise of interference within these systems and the continuous labor and knowledge it elicits but recursively eludes.

Epistemes of Interference and the Turn Toward Transduction

The redefinition of noise as interference, its expansion to all categories of the signal, and the comparison of its effects to hearing loss rendered deafening as endemic to industrial modernity. (Mills 2011, 136)

An understanding of noise as a disruptive interference has a history, one that is thoroughly intertwined with the history of telephony and telecommunications engineering (Mills 2011; Sterne 2003). New apparatuses for testing hearing (Tkaczyk, Mills, and Hui 2020), inscription devices (Latour 1987, 67-70; Latour and Woolgar 1986, 51), and cultural techniques (Krämer 2017; Siegert 2015) facilitated its embedding as a scientific object (Daston 2000, 12), problematized for its disruption in the transmission of signals across telecommunication lines that entailed various transductive operations.

In recent years, anthropologists and STS scholars alike have turned toward an analytic of transduction to attend to the knowledge infrastructures that condition the possibility for the transfer of signals and the conversion of physical phenomena into sensory objects (Spackman 2020) and speaking subjects (Roosth 2009). For sound scholars, the historical specificity of transduction as a technical term emergent within the development of telecommunications systems affords an opportunity to bridge cultural analysis with technical description by “offering a conceptual language partially shared between scholars in the humanities and in engineering and science circles” (Helmreich 2015, 223). One of its earliest technical references appears in the 1923 Bell System Technical Journal, where the “transducer” is characterized by its ability to modify the “spectrum of interference presented to terminals of the selective network.” ² From its very technical inception, then, engineers noted how the physicality of the transducer, whose operation alters a signal in one medium such that it can continue to propagate across other media, modulated the kind of disturbances that ensued. Noise as a phenomenon of disturbance and interference is, in other words, built into and made perceptible within the material functioning of transduction.

Scholarship by media and sound researchers has foregrounded how specific assemblages of technologies, practices, and discursive formations materialize noise. Douglas Kahn (2013, 70) describes how incorporating the earth into telegraph circuits opened up a new world of noise in a doubled sense: “Spluttering and bubbling, jerking and rasping, whistling and screaming”—these were the audible qualities of what comprised the sounds of the earth, a “meaningless jangle” brought into audition, which in turn doubled as another kind of noise, as “noisy” interferences known by the disruptions they produced within the traffic of messages. This doubled ontology of noise, both acoustic and performative, elicited infrastructural innovation in the form of an additional metallic line to close the telegraphic circuit and restore the smooth (i.e., self-identical) transmission of the message between sender and receiver.

I build upon these scholarly insights into the historical ontology of noise as something both contingent upon the infrastructural systems it iteratively disrupts, and actively molding the solutions produced to exclude it. To do so, I draw upon and extend methods from media archaeology. These methods involve mapping technical media diagrammatically to understand objects as forms of “materiality-in-action” (Parikka 2011a, 65). Diagrams serve here as a starting point for understanding the information patterns, circuits, and relations through which objects work and the “micro-temporalities” they entail (Ernst 2011). The historical genealogies written using these methods, however, have maintained a telecommunications emphasis on noise as that which affects and interferes with a particular “media system” (Parikka 2011b): the telegraph system, radio, and so forth. As infrastructures, however, such systems arguably comprise multiple kinds of technical devices, including the pivotal operation of transducers necessary for altering signals composed of varying kinds of energy across different media.

My turn toward mapping the historical specificity of transducers within communication infrastructures stems from an unexpected historical anomaly I encountered during archival research into signaling systems in mines, and how disruptions to these circuits became problematized by the engineers designing them. The circuits acted as infrastructures in multiple ways. Infrastructures are, as Larkin (2013, 329) writes, “matter that enable the movement of other matter.” Due to their dual ontology as being both things and a relation between things, these objects “create the ground on which other objects operate and when they do so, they operate as systems” (Larkin 2013, 329). Mines cannot operate without numerous kinds of infrastructure, both the processes involved in carving out and maintaining the negative space of the mine, as well as the systems of tunnels, vertical shafts, railways, and ventilation systems necessary for enabling the removal and movement of coal from seam to surface organized around technical and managerial standards (Carse 2017, 28). As these infrastructures became mechanized around the turn of the 20th century, from the introduction of pneumatic drill presses to electrical power, signaling systems that direct the movement of goods, workers, and machines also became re-engineered by applying knowledge of electromagnetism, already used in telegraphic systems above ground, to the subterranean space of the mine. These infrastructural changes generated noise. Mapping noise within these circuits, however, by conceptualizing it as the spectrum of disturbances that surfaced within operations of transduction, indicates a surprisingly historical deafness to audible noise as a source of interference for engineers in the early part of the 20th century in Ruhr. A transductive heuristic helps us to understand this disjuncture in the historical ontology of disturbances by asking us to attend to how conversions of signals across different channels crystallize matter and meaning.

In what follows, I undertake an archaeology of transduction to hone in on a metaphoric fissure that separates one heterochronous episteme of interference from another; both epistemes share a concern with noisy disturbances in the semiotic traffic mediating communication across the subterranean spaces and the surface but understand them in incommensurable ways. In my reference to fissure, I draw upon work by anthropologist Sophia Roosth (2018), who sees in Lévi-Strauss's reminiscences of geology an allegory for ethnography. In her essay on coeval temporalities, Roosth cites Lévi-Strauss's encounter with two green plants growing on two different ammonites, a spatial vision where different temporal epochs become juxtaposed through ethnographic writing. In both geology and ethnography, eons can “stratify, condense, and collocate in a single moment” the imagination of heterochronous things (Roosth 2018, 75). Inspired by this insight, I juxtapose one episteme of interference with another to invite attention into how hybrids of machinic and bodily techniques involving transduction become purified to produce incommensurable understandings of noise that, in turn, motivate the technological aesthetics of infrastructure and its imaginaries.

The first episteme of interference I unearth details observations and interventions into disturbances across sign systems designed to communicate acoustic signals amid the electrification and mechanization of underground mines in the first half of the 20th century. Analyzing archival materials from former mine administrators and technical journals indicates that noise as an audible phenomenon within the mine environment was not listened to by engineers as a source of disturbance. Rather, engineers located sources of accidents and errors in the transmission of acoustic signals solely with the technological apparatus of the system; their visual observations implicitly acknowledged the role of workers’ transductive capacities as necessary for signal transduction, yet they purified these (human) energies and resistances out in their judgments regarding the cause of interference in signaling. Reduced to an automated interface for encoding, decoding, and relay within engineering formulations of signal disruption, human transductive capacitances and capacities could, according to these engineers, be overcome by modifying the technological design of the nonhuman aspects of the circuit.

Within the second episteme of noise I trace in this article, noise has become an environmental thing emitted across the medium of air in the mine and transduced by the human ear, producing effects of masking that hide or render silent the hearing of acoustic signals (i.e., desired sounds). The effect of noise on the transductive capacities of the ear becomes a scientific object within this episteme, marshaling the energy and attention of engineers, mine administrators, and scientists. These actors used cultural techniques such as audiograms to inscribe the results from testing workers’ hearing that were then translated back into the aesthetics of acoustic signaling. Such testing apparatuses made noise known by transducing it into something else. Such systems often used proxies to “stand in” (Mulvin 2021) for the environmental noise of the mine, understood as a totality of sound pressure waves.

Through these apparatuses, audible noise morphs into the visual flatness of a data space acquiring density, durability, and salience (Daston 2000), de-sonified and quantified. It also acquired epistemological authority, shaping how engineers sought to reformat acoustic signaling and informing bureaucratic regulations to protect workers. Yet environmental noise ³ —ephemeral and bound contextually to the spacetime of its actualization—remained underdetermined and thus a potential parasite of future disruption. Within this episteme, environmental noise tunes the transductive capacities of the human ear otherwise; it can lead to ringing in the ears (tinnitus) or a temporary displacement of hearing thresholds, a phenomenon that became a focus of scientific study in the 1960s. Cumulative exposure to high intensities of noise can result in the diminished capacity of inner cells to vibrate sympathetically with changes in sound pressure levels, a resistance that opens up other transductive pathways, such as hearing through cochlear implants or sensing environmental vibrations through bone conduction.

In both epistemes, transductive operations combine human and technological hybrids in the design of communication systems that move the matter of information by choreographing the transmission of signs across different classes of energy. Instead of focusing on a specific channel or medium, I suggest that transduction as a scientific and engineering concept—which recognizes electrical resistance and disturbance as part of the physical process involved in converting signals across media—offers an underutilized heuristic for STS scholars to attend to events of disturbance within the work of infrastructures, particularly those involving sensing and signaling practices (Blok, Nakazora, and Winthereik 2016; Gabrys 2019; Helmreich 2007). Given the expansion of such infrastructures within the Anthropocene, “noise” as resistance and disturbance that accompanies all transductive operations moving information across media suggests it is likely pervasive as a material phenomenon, though underanalyzed as an agent. Here, I show how noise becomes materialized as interference within communicative infrastructures, an attunement that might offer a heuristic for fathoming how the unruly turbulences of transduction perturb and recursively shape the knowledge and design of such infrastructures. Rather than an opposition between failure and success (Chu 2014, 352), a transductive approach acknowledges the disruptions and resistances involved in choreographing the movement of signs across space and time, thereby inviting attention to how noise operates parasitically as a potent force at work in the history and anthropology of human–technological assemblages. ⁴

Tuning into Noise in the Archive

It is July 2023. Sitting in the conference room on the first floor of the Montanhistorisches Dokumentationszentrum at their offices on Bessemerstrasse in Bochum, I listen briefly to the crunching and crushing sounds of the neighboring recycling center that accompany the humid air floating in through the open windows. I have begun my descent into the archive as part of my research on sound, noise, and signaling as it informed work and knowledge in the deep-shaft coal mines that dot the landscape of Germany's Ruhr region. Extending far below ground, these mines often make themselves felt as a presence aboveground through unexpected subsidence that can divert train tracks and crack housing foundations or roads, or through tales about the Ewigkeitskosten (eternity costs) of mining, a term that circulates in German media and policy reports to reference both projected and ongoing costs required to maintain decommissioned mines in legally mandated ways. ⁵ But what sort of remains and reminders do the sounds and noise of underground mines leave?

To begin to answer this question, I search the archival holdings using keywords, such as “Signalanlagen” (signaling systems), “Geräusche” (sounds), “Zeichen” (signs), and “Lärm” (noise). The archive's holdings came into existence as a mandate to preserve the “schriftliche Hinterlassenschaft” (written remains) of Germany's former mines, which include hundreds of hard coal mines, ore works, and potash mines as a “Gedächtnis des deutsche Bergbaus” (memory of German mining). ⁶ The founding of the Bergbau-Archiv Bochum (BBA) on July 1, 1969 coincided with the consolidation of the remaining mines in the region under the Ruhr Aktionsgesellschaft, in response to a series of “crises” in the industry, which at that time was already understood to be entering an enduring structural change (Friedrichs 1996). Consequently, the archive acquired the stewardship of files, maps, photos, and films from the long era of German mining. Numbering around 7,000 linear meters of shelving and including over 330 holdings, the archive is divided into four groups: business and corporate holdings; holdings of mining associations and organizations; pre- and posthumous personal collections; and special archival collections.

Identifying and requesting the corresponding holdings that appeared in their database based on this keyword search allowed me to request certain files for further review. Intermittently over the rest of the year, I gingerly worked my way through sheets of cost suggestions, correspondence between mine owners and other businesses and state authorities, product catalogs, photographs, and engineering diagrams. At the same time, what I sought evaded me. The information about past sonic experiences in this archive could only be derived from nonsonic traces, like mention of equipment purchased, in the photographs documenting noise measurements and the damping of Grubenlüfter (pit ventilators) in the 1970s, in the diagrammatic sketches of signaling circuitry, and in various product advertisements. Such an externalist endeavor marks those who approach sound as a thoroughly historical, rather than transcultural, experience. Sound studies scholar and historian Jonathan Sterne (2003, 13) describes such an approach as one that requires moving to the edges of sound and just outside it, “into the vast world of things that we think of as not being about sound at all…because the very density of sonic experience emerges and becomes perceptible only through its exteriors.”

Since the late 19th century, the predominant way of exteriorizing sound to understand it has been through operations of transduction, which broadly refers to the transformation of energy from one material modality into another. Within the history of sound-reproducing technologies, Sterne (2003, 33) analyzes how the vibrating diaphragm that allowed telephones, phonographs, and other sound-reproducing technologies to transduce signals across different media were themselves artifacts of changing understandings of human hearing, whereby the ear itself came to be understood and modeled as a kind of transducer. Yet in the archive I encountered an unexpected silence in the first half of the 20th century about the role of the human ear in transducing acoustic signals underground. The correspondence, scientific studies, and policy reports authored by mine administrators, engineers, and doctors remained silent, mentioning nothing about the problem of (audible) noise.

Beginning in the 1960s, however, measurements of noise exposure and its effects on the ear began to surface across and within many documents. Thus, although the ear modeled as a transducer is a prominent artifact in Sterne and Mills's histories of hearing technology in the early 20th century, in my research it arrived with a delay. To make sense of this temporal delay, I conceptually utilize the affordances of transduction as an STS approach to attend to how the transductive operations of mines themselves become altered across two epistemes of interference. Documents from the Montanhistorisches Archiv, coupled with articles from Glückauf, a technical journal published between 1865 and 2012 by the VGE Verlag GmbH, serve as my primary historical sources in this archaeological account of interferences in subterranean signaling. Before delving into the underground, however, I provide a brief description of how electrification and mechanization altered this subterranean space and how it became built through processes of extracting the earth's interior, to hollow the underground out from within, making room for human–technological infrastructural assemblages.

Engineering the Underground

The environment of underground mining in Ruhr around 1900, where miners labored to wrest sunshine fossilized in the form of Steinkohle from its geological internment, symbolizes the way in which environments have always been partially “artificial” or technical in a modernist sense. That is, rather than formed spontaneously by nonhuman processes occurring in “nature,” a category conceived of as the antithesis of man (Latour 1993), such environments are shaped by anthropogenic actions and variously referred to as engineered, built, or manufactured. For Lewis Mumford ([1934] 2010) the mine allegorized a surrounding replete with the inorganic, one in which all environmental conditions, such as light and air, that might sustain human life must themselves be “deliberately manufactured” by humans. For Rosalind Williams (2008, 7), the persuasive power of the underground as an exemplary image of a totalizing artificial environment stems from its qualities of enclosure and verticality. Within the Ruhr region, the former heartland of Germany's coal and steel industry during the first half of the 20th century, these dimensions of enclosure and verticality acquired new scales and meanings through the expansion and extension of capital-intensive infrastructures.

Verticality in the Ruhr mines encompassed more than a geophysical feat of digging deeper into the earth; it also speaks to economic restructuring that provided greater capital and control over the various processes involved in rendering underground coal a “resource materiality” (Richardson and Weszkalnys 2014). By the time that the Rheinische-Westfälisches Kohlen-Syndikat (Rhenish-Westphalian Coal Syndicate) came into operation in 1893 ⁷ as part of the vertical integration of the industry, the coal mined in Ruhr had become increasingly geared toward the production of coke (Fettkohle). ⁸ The deposits (Lagerstätte) of Fettkohle largely lay in the northward seams of the Ruhr coalfield, which dip under the cover of secondary water-bearing rock, in an area known as the Emscher-Lippe subdistrict. Thus, as the industry moved northward in the late 19th century, the extraction of coal required new kinds of technology to vertically access these seams, which, in general, lie deeper in the underground than the deposits of Magerkohle and Esskohle that had been mined in the Ruhr valley through small-scale operations in the 18th century (Pounds 1968, 49).

This vertical descent entailed more than just mapping and modeling the underground ⁹ ; to mine seams that lay thousands of kilometers underground required transforming the underground into a spatialized, enclosed environment inhabitable for humans through infrastructuring (Blok, Nakazora, and Winthereik 2016). Similar to the submarine cyborg (Helmreich 2007) and the NASA cabin (Aronowsky 2017), burrowing deep into the earth and shuttling men and equipment down to work the coalface required setting up systems of monitoring, control, and communication to make it possible for biological life forms (human, horses, etc.) to labor in that space. Unlike the submarine and the space cabin, however, the boundaries distinguishing the insides of the cyborgian cabins from its inhospitable outsides were not architecturally fabricated using man-made materials; rather, the three-dimensional spatial enclosure of the deep shaft mine emerged through accumulated processes of subtraction that removed the interior of the earth. Shaft sinking techniques developed over centuries have allowed the otherwise hidden, “interior” subsurface of earth to be exposed as an exteriority for excavating and extracting material resources (Evans and Graham 2020). Its sides formed through millennia of geological processes, the negative space of the mine requires continual care and maintenance to sustain it against abiding forces that would render it inhospitable to human life. Alongside the mechanization and electrification of the underground mine, such maintenance included monitoring the composition of the air, techniques of ventilation using mechanically generated suction, machines to pump out water, and buttresses to hold space against the pressure of the mountain that exerts force on all sides of it (Figure 1).

Figure 1.

Illustration depicting how to channel air flow through multiple shafts and seams. Reprinted from Burghardt (1995, 49, Fig. 17).

Many of these technical objects used to excavate this dark, negative space emitted vibratory waves registered by the human ear as sound, as did the expansion of interlocking infrastructures of railway tracks and winding towers built to buttress the movement of men, machines, and coal or “black gold” between seam and surface. Within the mines of Ruhr, such sonic-emitting infrastructural innovations included the use of pneumatic drills to extract coal from the face of the seam, a threefold increase in the use of drilling machines powered by compressed air for opening up shafts and seams, the introduction of mechanical conveyors, the use of locomotives to move coal from the face to the bottom of the vertical shaft, as well as steam-powered fans to better steer air through the mine using suction (Burghardt 1995). Concomitant with these changes were also the transposition of telecommunication technologies used aboveground since the 19th century to innovate and electrify signal-apparatuses, enabling the transduction of signals across further distances with faster speeds through alterations and extensions in the kind of media used.

To date, STS literature on the underground has remained relatively quiet about the role of sound, noise, and signaling in representing and intervening in the subsurface (Kinchy, Phadke, and Smith 2018; Nystrom 2014), as has the emphasis on infrastructure's “invisibility” in STS (Star 1999). Here, I respond to this silence, building upon Helmreich's (2007) use of transduction to attend to both how the underground became a site of signal routing, monitoring, and controlled feedback spanning the technological and the human, and fusing flesh and information. In doing so, I extend his insight that attending to transduction means listening for disturbances and resistances; in what follows, I try to sound out the historical specificity of noise as connected to operations of transduction that were essential material components of signaling systems in the mines of Ruhr as well as globally (Gould 1914).

A Peculiar Deafness in Engineering Knowledge in Ruhr, ca. 1870-1940

A key node of signaling systems in mediating the movement of coal, rock, machinery, and men was the pit bank, an infrastructural articulation where a vertical shaft from the surface meets the entrance to a subterranean tunnel running roughly perpendicular to it. A series of transductive operations involving encoded signals connected the Anschläger (striker), the name for the worker who initiated the signal at the pit bank, and the Fördermaschinist, a worker above ground who controlled the movement of the pit cage through a rope and pulley system. Such systems relied on new products and circuitry, like those featured in a 1903 Siemens and Halske catalog, whose incorporation into the mines received special attention by engineer Rüdorff (1905) in a Glückauf article. Engineer Rüdorff approvingly reports from an inspection of the mine that the Siemens and Halske electrical signaling system had been operational for a year in a “satisfactory” way. The apparatus, he declares, offers “the greatest conceivable safety against faulty signal transmission with a simple operation [einfache Bedienung] of the equipment” (Rüdorff 1905, 508, emphasis added).

What Rüdorff refers to as a “simple operation” is, as I outline, a programming of noise, understood as interference in coding that disrupts signal transmission, into the communication system itself as a safety feature. The featured apparatus Rüdorff discusses consisted of a Zeigenapparat (pointer apparatus) installed alongside the usual Signalglocken, which consisted of a circular signaling disk on which a Zeigen (pointer) moves. When the Anschläger at the filling station sounded the signal, the pointer apparatus simultaneously transmitted the (optical) equivalent of the signal to the pit bank and the machine room. Here, the optical signals served as signs of what was to come, yet only once the Anschläger sounded the sign from the pit bank, kinetic energy transduced to generate the same audible sign at all other “signal stations,” did it count as that which was to be executed. However, should the acoustic signal sounded by the pit bank not correspond to the optical sign initially issued, the machine operator was expected to then refuse to execute the signaled action. In this sense, the apparatus built into it the possibility of semiotic noise as a safety feature (Figure 2).

Figure 2.

Engineering diagram illustrating trackside signal and telephone system for Zeche Alma (Gelsenkirchener Bergwerks-AG) in the German city of Gelsenkirchen (1911). Image reproduced from attachment to letter Siemens and Halske (1911) from the Bergbau-Archiv Bochum BBA folder 41/1900 with permission from the Montanhistorisches Dokumentationszentrum des Deutschen Bergbau Museums Bochum.

By semiotic noise, I mean the disruption in a sign comprised of a signifier and a signified. I view the binary coding system used in these electrical signal-apparatuses as analogous to Ferdinand Saussure's conceptualization of the sign ([1916] 2011). For Saussure, signs are dyadic, consisting of a “vehicle” or signifier and its signified (i.e., semantic meaning). The link between the signifier and the signified is entirely arbitrary; that is, there is no physical or sensory aspect of a signifier that motivates its meaning, which is given synchronically through the system of langue ¹⁰ in which it is embedded. The semiotic system, codified across numerous mines, acts similarly. Counting rhythms transduced into signals within an electromagnetic medium are decoded by humans into specific commands for action. This circuit of transduction disciplines bodies underground, which are increasingly synchronized into different blocks of action to transduce a supposedly seamless circuit of work (Figure 3).

Figure 3.

Notice board displaying signal codes from Zeche Zollern in Essen, Germany. Reproduced from Telsemeyer and Tempel (1988, 14), reprinted in (Schmidt 2005, 132, Fig. 5).

Such codifications appeared on Anschlagtäfel (notice boards) posted at each workstation ¹¹ across numerous mines, and in the printed mining regulations for cableway systems ¹² (Schmidt 2005, 132). Within Ruhr, one “strike” meant “stop,” two “hits” meant “go.” One followed by two meant “to go slowly,” and so on. But with the introduction of the optical pointer apparatus, two different “sign” languages came into play. Disagreement between the two signs meant interference and nonexecution, whereas correspondence represented another sign at another metalevel, designating that the signal be executed. If we understand noise as interference, then within this communications system, engineers had designed it as a feedback mechanism to introduce redundancy into the system as a safety feature to help ensure that the identity of the signal sent matches the signal executed.

As Parikka (2011b) notes, within the formalized understanding of communications systems that apprehends noise as an inherent feature of such systems, redundancy has become a cybernetic strategy to increase the probability that the signal received matches the signal sent. Parikka notes that such solutions often later evolve into other forms of noise, but in this example, a historical form of noisy interference itself has become programmed into a rudimentary cybernetic system of feedback and control. This system utilizes two different sense-specific channels transmitting information at different speeds to program into the human–technological circuit the possibility of semiotic noise, that is, the mismatch across two different communication channels, as a safety feature. The Fördermaschinist in this diagrammatic circuit acts like a digital interface, interpreting across two binaries: match or no-match, execution or nonexecution.

Within this rudimentary human–machine cyborg, the intermediary of the human as a transducer only figures in its capacity to automatically process informatic binaries. As Helmreich (2007, 628) notes, drawing upon Donna Haraway, cyborgs have been “primarily imagined in a visual, even textual register—as made of inscribed surfaces, of information and codes.” Interiority does not figure here; cybernetics is a behaviorist science and as such, one befitting of the technological optimism that animated the engineering spirit suffusing the pages of the Glückauf magazine. The incorporation of the human in this apparatus as merely the decoder and relay across two different sign systems indicates how the human became figured as one transducer among several within a larger, more-than-human semiotic circuit that fuses flesh, electricity, magnetism, iron, copper, counting, and information. Although in a different episteme, the miner as a transducer becomes reduced to his ear, within this one, engineers instead purify the spectra of noise across transductive operations as solely technological. They locate it in technical media rather than subjects. And even where noise as interference in coding was introduced as a safety feature against accidents, it continued to erupt and interrupt operations, calling attention to itself as a cause of accidents.

Combating Accidents from Noise

Although signaling systems were introduced into the railway system and mining infrastructures as a safety mechanism (Abel 1889; Winston 1998, 23), mismatches between the signal sent and that received transduced further accident. In a 1914 Glückauf article, for instance, a mining officer from Bochum named Kliver noted that it had come to the attention of the Oberbergamt (regional mining authority) in Dortmund that certain shortcomings had been observed within shafts using electrically mediated signal bells. ¹³ To determine the source of the disruptions and whether such systems could be considered “reliable,” the Oberbergamt sought reports in 1904 and 1913 from the district mining officials within its administrative district. Their tabulated observations addressed two kinds of disturbances: (1) “signal strikes” that did not sound; and, (2) ones that were repeated of their own accord, rather than transducing those given by the operator (Figure 4).

Figure 4.

A tabulated summary of signaling defects in Ruhr mines (1914). Reproduced from Kliver (1914, 129).

In the 1904 report, engineers attributed errors largely to technical malfunctioning derived from exposure (Aussetzung) to environmental conditions that degraded the technical apparatus or disrupted the electrical circuitry, or wear and tear of the “elements” of the apparatus that led to weak or false tones. In 1913, reports attributed errors almost completely to the actions of the human operator as the Signalgeber, rather than to the action of the technical object. In his assessment, however, Kliver, argued to the contrary: that this semiotic noise—a mismatch between the sign sent and that received—ultimately stemmed from technical shortcomings in the design of the apparatus.

Responding to the cause of error being attributed to the speed of striking in the 1913 tabulation, Kliver attributed this shift from object to operator as likely stemming from new ordinances ¹⁴ that required more stringent Überwachung (monitoring) of the electrical signaling equipment. However, he ultimately disagreed that “too rapid signaling” was the reason behind these defects. This ostensible human error was “nicht der eigentliche Grund, sondern nur äußere Veranlassung,” (not the actual reason, only the external cause). Consequently, Kliver devoted the remainder of the article to arguing that, based on his engineering knowledge, the problem lay instead with the transductive operations of the signaling apparatus, a shortcoming that could be solved through technological improvements to the device's “sensitivity.”

A focus on the “sensitivity” of the apparatus, rather than workers, as the source of error stemmed from several considerations. One question arose as to whether the sensitivity of the Glockenklöppel (the bell clapper) was, in fact, able to follow the repeated magnetische Einwirkung (magnetic influence) quickly enough, a shortcoming that could easily be remedied by newer products capable of executing 250 to 300 strikes per minute. Moreover, argued the author, the absence of signal strokes was not even in and of itself a cause for concern, if it failed across all signaling places; the danger lay in the absence of a stroke at one, while being present at others. This discrepancy among the strokes transduced across the system stemmed from an “unequalness” in the “sensitivity” of the Wecker (bell) of the same signal system, not the speed with which the human striker struck. This difference in sensitivity might have arisen from not only mechanical differences, such as the length of the clapper, its amplitude, its weight, or even its angle, but also electrical differences; the number of coil windings within the electromagnet of each Wecker, the thickness in the wires of such coils, and the way it was anchored, all affect the Widerstand (resistance) of the Wecker to the electrical current and hence how easily it responded. In other words, differences in the transductive resistances and capacitances of the technical media were the source of semiotic noise and the place to intervene to negate it. ¹⁵

This account indicates how interpretations of noise as disturbance remains key for understanding how technical objects evolve, as well as how matter and meaning articulate with one another. Within such technological utopianism, semiotic noise becomes itself asymmetrically translated into a sign of the shortcomings in the transductive elements internal to the signaling device. Understanding of noise as disruption within these underground signaling systems does not figure within this episteme as something audible. In what follows, I look across the fissure to another sediment of time, when the audible noise of mechanization does become a disruptive actor within the signaling system, a problem addressed by de-sonifying it into a scientific object (Daston 2000). Within this episteme of interference, noise made itself known as an effect of masking. Rather than referential to any source of sound, noise as a masking effect is relative, known through the displacement it induces on hearing thresholds by affecting the capacities and sensitivities of the human ear. The ear as a transducer becomes a key site for materializing noise in order to overcome the problems it poses to the seamless conversion and propagation of signs.

How Noise Became Environmental

If semiotic disruption in underground transduction pathways had previously been the marker of an apparatus's technical imperfection in the early 20th century, instead, in the 1960s and into the 1970s, noise became what we might think of as “atmospheric” (Peterson 2021). It was part of the air down there, in underground mines, an elusive entity to be damped but never eliminated. It contaminated and disrupted the signaling systems, but this time not as an error within a device or a technically closed system of transduction. Instead, it was now the name for a masking effect produced by both sounds within the surrounding environment, and the processing of stimuli into sensation through the transductive mediations of the ear.

A 1976 engineering dissertation, “A Contribution to the Question of Acoustical Signaling in Underground Hard-Coal Mining” (Levin 1976)exemplifies this epistemic shift from concerns about semiotic disruption to the problem of noise as an effect known by how it masked or rendered silent other environmental sounds. It was written by Günter Levin, a certified mine assessor who at that time led the occupational disease area within the mining employees’ liability insurance association of West Germany. Noting the alarming tenfold increase in cases of recognized, occupationally induced hearing difficulty (Lärmschwerhörigkit) between 1966 and 1972, Levin observed that the high noise strain of the workplace not only increased the risk of illness but also raised concerns about accidents due to masking effects. Levin's study appears in the archive alongside numerous other publications and reports concerned with how to combat noise in the workplace, which began to appear following the codification of noise-induced hardness of hearing as an occupational hazard applicable across all trades beginning in 1961. ¹⁶ Compounding the problem of masking from environmental noise was also the “deafening” effect this noise bore on the miner's ear. Since the 1920s, audiograms as part of audiometric regimes of testing had become key inscription devices for understanding how intensities of sound pressure across various frequencies could affect the sensitivity of the ear, modeled as a transducer. To address the problem and identify possible prophylactic measures, Levin drew upon these tools from audiometry, transducing the environmental noise of the mine into a laboratory setting using proxies and applying the technique of the audiogram to reveal the effects of this simulated noise on miners’ hearing thresholds.

Briefly, the flattening technique of the audiogram bridged the sensory world of sound with formulas and diagrams (Krämer 2017). As an inscription device (Latour and Woolgar 1986, 51), it rendered noise perceptible as a visual figure and as a scientific object amenable to quantification, comparison, and mathematical understanding (Krämer 2010). The epistemology of the audiogram's line relied on new units—the best-known, though still controversial, being the decibel (see Mills 2018). The form of the audiogram utilizes a two-dimensional grid, juxtaposing sound intensities along its y-axis, represented and measured in decibels, across discrete frequencies along its x-axis, measured in hertz (Hz). It first came into practice in the 1920s alongside the public and commercial debut of the audiometer. As Mills (2011, 122) has shown, testing scenarios involving transducing sound into audiometers and audiograms enabled an objectification and understanding of noise as “deafening,” which became abstracted from physiological injury to encompass “temporary hearing impairments caused at any point along a telephone circuit: room, transmitter, line, repeater.”

Through a close reading of Levin's experimental setup and conclusions, I extend this analytic of deafening noise and its relation to masking, to foreground its relative and contingent nature. Building upon psychoacoustic theorizations of masking and its integration in the perceptual coding of digital formats (Sterne 2012), I examine the role of the audiogram in Levin's work as a critical component of knowledge infrastructures that render noise knowable through its effects on listening subjects exposed to transduced sound waves. As such, I argue this kind of noise is best conceptualized not as a quantity of sound pressure but as an epistemological object understood relationally through the disruptions it produces upon (human) hearing. The planar surface of the audiogram allows these displacements to be visually figured as shifts, measured in “quantities” of sensation (Mills 2011; Sterne 2012) and in hearing capacity at various frequency bands.

Levin's study situated itself within an episteme that recognized the ear as a transducer that mediated between sound pressure levels in the environment, and electrical signaling pathways to the brain. Medically, work environed by loud and constant sound correlates with observations of deafness as far back as Ramazzini's important 18th-century book De Morbis Artificum Diatriba (Treatise of Artisans and Workers), first published in 1700 in Latin (Ramazzini 1700). However, the modern science of audiometry, emergent in the late 19th and early 20th centuries, allowed such noise to be studied scientifically through observation of its measured effects on hearing thresholds, rather than as statistical correlation. Audiometry helped to quantify shifts in hearing thresholds both from age-related hearing loss, known medically as presbycusis, as well as contingent, though possibly permanent, shifts due to noise environing the “indefensible” ears (Schwartz 2003) of humans. In his study, Levin attempted to account for both sources of noisy disturbances generated by the effects of masking.

Statistical inference offered him the means to judge what percentage of the workplace suffered from “deafening noise” (Mills 2011) due to changes in sensitivity that modified the ear's transductive capabilities. Statistical studies, for instance, reported that 52.7 percent of the workforce employed in underground mining was above the age of forty. Because audiograms showed an age-related hearing loss for this group in the order of 10 to 25 decibels (dB) within frequencies of 3,000 to 4,000 Hz, Levin reasoned that such threshold shifts must be considered in the aesthetics of acoustic signals. Compounding this age-related shift in hearing sensitivity was an estimation that nearly 80 percent of all production-related equipment emitted a sound level over 90 dB(A). Given that at least one-third of all miners worked in proximity to such equipment for twenty to twenty-five years, Levin extrapolated a 28-30 percent risk of hearing deficiencies of 25 dB or more at 500, 1,000, and 2,000 Hz based on hearing tests done according to standardized protocols across Germany. Experimental data from 1,400 reported cases of noise-induced hearing loss from those employed in hard coal mines in 1973 buttressed these statistical inferences, which showed a sharp drop in the curve of the audiogram beginning at 2,000 Hz (see Figure 5).

Figure 5.

Audiogram visualizing hearing loss in reported and assessed BK 26 cases. Reproduced from Levin (1976, 13, Fig. 5).

Such effects posed a conundrum: they provided quantitative evidence of the harmful effects on environmental noise on the miner's aging ear yet, as Rosalind Morris (2008) has shown, the miner has been disciplined to listen for sounds of danger. Engineers and mine administrators thus had to address how to protect ears from the deleterious effects of exposure to environmental noise in a way that still made it possible for miners to hear and listen to acoustic signs. To address this problem, Levin coupled certified studies ¹⁷ of the “damping” effect of hearing protection undertaken by the federal office for physical technics (Physikalisch-Technischen Bundesanstalt) through his own study. The study experimentally tested the hearing of workers to determine the masking effects of wearing earplugs and whether and how it shifted the “monitoring threshold” (Mithörschwelle) of hearing for certain machines commonly found in underground mining workplaces.

As Sterne (2012) notes, the masking of sounds from other sounds had become a central focus of 20th-century research on hearing within Euro–US science, one connected to changing relationships among vibrations, ears, and minds. At its core was the recognition that the physicality of sound as external stimuli produced by variations in pressure levels in a medium differed from the subjective experience of sound: perceptions of loudness, pitch, and duration did not seamlessly correlate with external measurements. Rather, stimuli passing through the ear-as-transducer produced distortions, resistances, and disturbances, which the science of psychoacoustics sought to explain and predict. In Germany, the 1956 publication The Ear as a Communication Receiver by psycho-acousticians Eberhard Zwicker and Richard Feldtkeller helped to scientifically establish the concept of masking. Key to their studies of masking were the distinctions between stimulus quantities and sensation qualities, with the latter clouded by subjective ambiguity. The solution was a set of methods that allowed the “relations between stimulus and sensation” to be “described by curves and mathematical expressions” (Feldtkeller and Zwicker 1956, 54-55, quoted in Sterne 2012, 103). These methods entailed experimental setups in which a listening subject concentrated on a single component of a sensation to abstract it from all others, allowing sensation to “become a thing that one can study,” gaining “density, repeatability, and a measure of materiality” (Sterne 2012, 104).

A similar assemblage involving specific combinations of technologies and transducers, techniques of listening and of documentation, and forms of knowledge was constructed by Levin to test the masking effects of hearing protectors on the “mind's eye” (Sterne 2012) of a miner. To do so, Levin (1976, 17) situated a testing subject between loudspeakers that, on one side, reproduced the sounds of various machines at the same approximate decibel level as they would appear at the distance of one meter to a worker underground (Figure 6).

Figure 6.

Image of experiment participant surrounded by loudspeakers raising his hand to signal hearing the sound of a tone. Reproduced from Levin (1976, 18, Fig 7.2).

At the same time, the testing subject listened to a series of sinus tones whose intensity levels were slowly increased between the octave frequencies of 63 and 8,000 Hz. Upon detecting the sinus tone, the subject would raise their hand, thereby indicating experimentally the “monitoring threshold” of the sound, which also denoted the intensity at which an acoustic signal needed to sound in order to be detected. While the study aimed at quelling concerns that hearing protection endangered workers by rendering important signals inaudible, the series of audiograms the test produced would also play a role in giving form to acoustic signaling (Figure 7).

Figure 7.

A series of audiograms that depict monitoring thresholds in response to the sound of: (a) a pneumatic motor UZ 32 DEMAG; (b) a hammer drill BM 22 DEMAG; (c) a dismantling hammer HF A 11 Hauhinco; (d) a ventilator DL 7–66 GF Turmag; and (e) a pneumatic plunger pump TP 100 Hauhinco. Reproduced from Levin (1976, 20-22, Fig. 8.1-8.5, edited by author).

In this way, the intermediary of the audiogram thus provided the grounds for a grid on which Levin could compare and manipulate results regarding shifts in hearing thresholds, which were determined by transducing sonic emissions from individual machines, to draw conclusions about the appropriate aesthetic parameters for acoustic signals in the external environment of the mine. By adding up the differences in “monitoring thresholds” of the earlier tests conducted on hearing subjects, Levin produced a new audiogram, one that depicted—again harnessing the visual simplicity of a linear curve—another “monitoring threshold” (Figure 8). This threshold, however, related not to a graphed sonic spectrum particular to a single machine, but instead offered a quantitative, relational unit of intensity by which an acoustic signal must be enhanced to counter the effects of “masking noise” that was part of the atmosphere of the mine. What comprised this “masking noise” was not an object or specific sound, but rather an environmental amalgamation—an indeterminate, environing thing—known only through its disturbances on hearing capacity, that is, as that which masks the signal through a shift in thresholds for specific frequency bands.

Figure 8.

An audiogram of two lines show the degree of sound level excess of the monitoring threshold over masking noise with and without ear protection. Reproduced from Levin (1976, 53, Fig. 13).

The difference of such shifts across various frequencies offered insight into how acoustic signals might be “coded” in order to be perceptually registered and transduced by the miner's ear (Morris 2008). For Levin, this combination of audiograms provided quantitative evidence that frequencies between 1,000 and 5,000 Hz could be heard with very little difference in sound level from the masking noise, a finding that indicates the particular sensitivity of the human ear's transductive capabilities within this range (Levin 1976, 53).

In shifting from the geometric figure depicted in the audiogram to its encoded arithmetic values, Levin provides prescriptions for how acoustic signaling devices should be perceptually “coded” (Sterne 2012). By coupling the “average differences between the monitoring threshold and the background noise” given in the audiogram (see Figure 5) with the “standard deviation” in monitoring thresholds at different frequencies for various machines, Levin (1976, 56) reasoned that the following parameters for the sound spectrum of danger signals should ensure the detectability of acoustic signals: 12 dB at 250 Hz, 12 dB at 500 Hz, 9 dB at 1,000 Hz, and 11 dB at 2,000 Hz.

In this sense, the audiogram as a cultural technique allows displacements of thresholds to “stand-in” as proxies for environmental noise, a cultural delegate (Mulvin 2021) that in turn informs the formatting of signaling apparatuses based on the capacities and interferences inhering in the human ear-as-transducer. Noise became known as an effect of masking within this new sonic regime and with it, engineers and administrators focused on what arrives at the miner's ear, whereas previous (engineering) solutions to transductive disturbances elided the role of the human as a signal transducer and, by extension, its role in producing noise. These proxies, however, allowed noise as an environmental thing to only be known and combatted through its de-sonified delegates; its environmental thingness as atmospheric and ephemerally produced through the actualization of the labor process remained indeterminate and potentially disruptive.

Conclusion

Formatting is, as Sterne notes, a way to more closely attend to the specificity of media by analyzing how they are codified in some way; codification could occur through policy, technological design, sedimented habit, or some other means. Yet, converting signaling across media requires coding with and against the noise coproduced within transductive operations. Acoustic signaling apparatuses as a human–technological hybrid rely upon operations of transduction necessary to propagate signs across different classes of energy. To code these signals involves counting rhythms, electromagnetism, and acoustic sounding to transduce them across various classes of energy. Before the sounds of the mine became known as noise (Lärm), engineering this coding involved refining the transductive pathways isolated within the technical apparatus, which are designed to be increasingly immune to environmental intrusion. As noise (Lärm) became named and known as an environmental interference of masking, intrusion became inevitable, and coding the design of acoustic signaling systems oriented toward combating noise as an atmospheric phenomenon deleterious for the ear's transductive potential. Such coding aimed to make signaling perceptible as an audible sensation detected against the noise reverberating throughout the underground mine, as well as the noisy resistances of fatigued inner ear cells that no longer responded to stimuli in the ways that young, “normal” ears did.

To draw out this difference across these two epistemes is to foreground how the historical specificity of noisy interferences within infrastructures involving signaling and sensing practices is coproduced alongside the transducers that make such material-semiotic processes possible. Transducers actualize potential energy whose pathways can be traced but not modeled in advance. It involves mapping out “the actual course inventions follow…corresponding to the discovery of the dimensions according to which a problematic can be defined” (Simondon [1964] 1992, 313). For engineers writing in Glückauf in the first half of the 20th century, noise as an environmental phenomenon that could disrupt signaling infrastructures was not among the spectrum of resistances and interferences that informed technological understandings of transduction at that time. Within another episteme, the audibility of noise de-sonified into visual graphs of intensity became embedded as a scientific object within experimental systems designed to test its effect on the ear's transductive capabilities.

Mapping out the historically specific course of transductive operations, which materially delimit how disturbances come to matter, offers an alternative approach in STS for understanding how infrastructures-in-action acquire significance. Such an approach complements attending to the materiality of infrastructure not by positing its material elements as ahistorical agents, but by tracing how they acquire meaning through the historical operations in which they participate, and which they make possible. At the same time, such a heuristic requires a shift in thinking about the materiality of infrastructure away from elemental terms like “concrete, steel or copper” (Anand, Gupta, and Appel 2018, 25), toward a focus on the operations that render matter with certain properties within individualized technical devices to support energetic conversions and signaling across various media (Lloyd Thomas 2015). Such a shift might reveal that noisy interferences parasitically disrupt infrastructuring not only in the Global South, but also in the Global North, as with the case studies of deep-shaft mining in Ruhr presented here. By extension, differences in perception of the seamless and invisible functioning of certain modernist infrastructures may reside more in the way that the materiality of infrastructures-in-action are conceptualized in STS, and the choice of methods to know them, rather than their situatedness at any particular global location.