Mechanical Instruments and the Aesthetics of Human Performance

by Peter Tracy


A drawing for the London News of September 20th 1862 depicting an Orchestrion by M. Welte


Music Box

Music of Japan. Parsimoniously
from the water clock the drops unfold
in lazy honey or ethereal gold
that over time reiterates a weave
eternal, fragile, enigmatic, bright.
I fear that every one will be the last.
They are a yesterday come from the past.
But from what shrine, from what mountain’s slight
garden, what vigils by an unknown sea,
and from what modest melancholy, from
what lost and rediscovered afternoon
do they arrive at their far future: me?
Who knows? No matter. When I hear it play
I am. I want to be. I bleed away

– Jorge Luis Borges, trans. Tony Barnstone[1]


The established musical instruments of today’s concert stage — violins and pianos, flutes and oboes, timpani and trombones – are relatively recent inventions. These staples of today’s musical practice found their contemporary forms just a couple hundred years ago — in the grand scheme, only a tiny blip in music history. Indeed, harps, horns, and harmoniums alike are still evolving, albeit too slowly to excite much notice: a little tweak here, a new type of valve there, and within a few hundred years your pianoforte could turn into a Steinway grand.

Instrument makers apply the technology of their times to the musical problems they, their customers, or their patrons want to solve. Often times, the goal is to make an existing instrument more versatile or more precise: German horn maker Ed Kruspe solved the F versus Bb horn debate of the late nineteenth century by sticking both horns together, creating the first “double horn” in 1897 and significantly improving the French horn’s accuracy and practical range.

Thus improvements in mechanical skill and knowledge can, in a very concrete way, effect the way music is made. Instruments are not transparent mediums, merely waiting for anyone with skill to pick up and play: their defects and eccentricities define not just what can be played, but how it is played. To aim more fully at uniformity and precision, one would have to either improve the skill of the instrumentalist, improve the instrument, or, in the most interesting cases, remove the player from the equation entirely.

Indeed, self-playing and mechanical musical instruments have a history just as storied as that of their human-wielded counterparts. Musical clocks, barrel organs, humanoid automatons, carillons, serinettes, player pianos – these are just a few of the noise-makers produced by a process of tinkering that has arrived today at fully fledged robot musicians and digital vocaloid singers. It’s clear that, even before recorded music, there was something in capturing sound via mechanical ingenuity, in creating an instrument that could repeat the same performance over and over with the same affect and pacing, that appealed to more than just the mechanically inclined.

The creators of many of the earliest mechanical instruments were inventors to the core, people who, alongside their contribution to music, have laid the foundations for modern mechanical engineering. Hero (also known as Heron), an inventor working in Alexandria in the first century CE, was one such inventor, a mathematician and engineer who’s musical achievements were more a product of mechanical curiosity than a sense of aesthetics. The long list of devices recorded in Hero’s highly influential work Pneumatics includes the first documented steam engine and a famous demonstration of hydraulics known as Hero’s Fountain. Yet Hero also describes a wind-powered organ (the first documented machine powered by wind) and two “theater automatons” which, if they were ever truly realized, would have been astonishingly complex for his day. One of these mobile theaters, showing a play called The Apotheosis of Dionysos, featured a mechanism whereby lead pellets dropped automatically onto a drum head and a cymbal, creating music to accompany mechanical dancers and a statue of the god Dionysos who’s cup overflows with wine[2].

The mechanical innovations of classical antiquity had a profound impact on the work of medieval Islamic engineers, who read and translated works by Hero and his precursors such as Philo (ca. 280 BC –ca. 220 BC). Among these Islamic polymaths was Ismail Al-Jazari (1136–1206), an inventor, engineer, and mathematician (among other things) who is credited with an equally prolific number of inventions. Often dubbed the father of modern robotics, Al-Jazari is perhaps most famous for The Book of Knowledge of Ingenious Mechanical Devices, in which he describes fifty mechanical devices and provides instructions for how to construct them. His “elephant clock”, a weight powered water clock in the form of an elephant, included a humanoid automaton who bangs on a cymbal and a mechanical chirping bird. Almost unbelievably, he also describes a full robot band that floated on a boat to entertain guests at royal drinking parties.

On the one hand, mechanical music can consist merely of a whistle powered by falling water, a ball bearing falling automatically onto a drum, or a bell chiming when enough liquid has flowed out of a vessel. On the other, it can consist of fully fledged model humans, who might even make facial expressions and have programmable movements. “Between these extremes”, write scholar Charles B. Fowler, “lies a fascinating variety of mechanical contrivances that play music re-corded on paper, wood, and metal rolls, on perforated metal discs, and through other devices.” He goes on to explain that, as he wrote in 1967, “five means [had] been employed to activate these machines — water, sand, weights, springs, and electricity.”[3]

A reconstruction of hers wind powered organ

One of these in-betweens is the formerly ubiquitous barrel organ which, although it requires a human to turn its crank, simply regurgitates music that has been encoded on a system of raised pins on a barrel. By spinning the barrel, the organ grinder both operates the instrument’s bellows and allows these pins to trigger levers which open the organ’s pipes for each note. Thus, if correctly encoded, a barrel organ can give a rather convincing if slightly clunky performance of whatever popular song, aria, or minuet the occasion calls for.

“As early as 1670”, writes Fowler:

“large organs with the tunes set up in pins on a wooden barrel, were used in English churches. The choir director played the barrel-organ by cranking a handle. Several hymn barrels were supplemented by an Easter and a Christmas barrel. Around 1700, a smaller barrel organ appeared, designed for home use. Three sets of metal pipes and one of wood played eight to ten jigs, marches, and love songs, with a soft, mellow sound…. at a time when few people were  able to play a keyboard instrument, the barrel organ provided musical entertainment and enrichment in the home.”[4]

Thus, mechanical music was an integral part of European music from a much earlier date than one might suspect. Indeed, although figures like Handel, Haydn, Mozart, and Beethoven do not seem to have written music specifically for barrel organs, they did not seem to feel that writing music for the mechanical instruments of their day was beneath them. Handel wrote music for carillon, an instrument which consists of tuned bells operated by and elaborate system of pulleys and levers, as well as for the automatic clocks of Charles Clay, a clock maker for King George I who is perhaps most famous for ordering one of his most elaborate musical clocks, built over the course of twenty years, to be destroyed three days before his death. Haydn was good friends with Father Primitivus Niemecz, a cellist, librarian, and instrument maker who build a series of mechanical organs which, to this day, preserve over thirty pieces by Haydn written specifically for them. Beethoven wrote a piece, Wellington’s Victory, originally intended for a contraption known as the panharmonicon, invented in 1805 by Johann Nepomuk Mälzel. This machine was one of the earliest of the orchestrions, so-called because they consisted of elaborate cabinets (not unlike those of Hero) which attempted to mimic the sound of whole orchestras through complete automation.

Whistling birds complete with real feathers and automated flutters, pinned-cylinder music boxes teaching canaries how to sing popular songs, costumed humanoid robots that play the hammered dulcimer or the flute or the drum: what do we get from these things? That is, why not just have a human play the organ or the piano or the dulcimer instead of programming a machine to do it for us?

For my part, I always find myself coming back to the music of Conlon Nancarrow (1912–1997), an American-born, Mexico-based composer who was preoccupied with complex mathematical relationships in music. Due in part to this fixation (as well as his relative musical isolation in Mexico), Nancarrow turned to the player piano to realize compositions which were increasingly impossible for human players to perform. Nancarrow, then, answer’s this question by creating music for which a human player cannot act as a substitute and for which the mechanical, lifeless precision of the machine is an essential facet of the work itself.

Yet, perhaps paradoxically, Nancarrows music has a charm and whimsy that almost works to obscure its rigorously mathematical formal qualities. It is oddly satisfying to watch one of Nancarrow’s meticulously punched piano rolls wind its way down and to hear the pianola strike itself exactly in time, every time. Indeed, there’s almost something silly about it: what kind of madman would write a tempo canon in a ratio of 3/4/5/7 only for it to be part of a composition known, if only unofficially, as the Second Boogie-Woogie Suite?[5]

Mechanical music from antiquity to the middle ages to the present day has at its heart a certain curiosity, a certain unquestioning appetite for novelty: if we can have a fully robotic band, why not build one? What would it be like to capture music in such a way, to listen to music played without a human hand? Still, automatic and mechanical musical instruments do raise some thornier questions: can music and human performance ever be separated, and should they be? Can a performance be emotionally compelling if performed by an automated robot with no feelings of its own? Does the creation of self-playing instruments further sideline musicians, who’s economic position in the last few if not the last hundred years cannot be described as anything but precarious?

These are questions of long-standing, with no clear answers. Thomas Pattenson provides just one example of the ambivalence with which musicians of the past must have met the spread of mechanical music:

“In the middle of the eighteenth century,”, writes Thomas Pattenson, “the composer Johann Joachim Quantz noted that ‘with skill a musical machine could be constructed that would play certain pieces with a quickness and exactitude so remarkable that no human being could equal it  either with his fingers or with his tongue.’ Quantz thought that such a machine could only excite astonishment, a sensation that would soon wear off once listeners understood how the         mechanism worked. In his view, mechanical instruments ultimately served to highlight the aesthetic primacy of live performance, which could be attained only by human musicians.”[6]

Despite the reservations of musicians like Quantz, by the early twentieth century an aesthetics of mechanical precision had pervaded the work of young composers like Stravinsky to George Antheil (1900–1959). The German composer Paul Hindemith (1895–1963) “instructed the performer of the “Ragtime” movement from his Piano Suite op. 26 (1922) to ‘play [the] piece very wildly, but always firmly in rhythm, like a machine’”, and went so far as to write music for mechanical organ to accompany the performance of an experimental ballet in 1926. Indeed, mechanical music was at the heart of debates about the role of technology in the arts and in society which raged in the early twentieth century: In 1927, Hindemith wrote that “no other aspect of musical life has been so hotly disputed in recent times as that of music made by mechanical instruments.”[7]

For mechanical music enthusiasts, musical machines meant no imperfections, no impurities, and no room for error. As in the work of Nancarrow, there was (and is) a mathematical, rationalist strain to the arguments in favor of instruments like orchestrions and player pianos. But there was also something of an emotional shift in the early twentieth century, an increasing gaggle of composers and musicians forwarding an aesthetic of dry precision that would go on to have profound repercussions for twentieth and twenty-first century music making. “By separating performance from the presence of musicians”, writes Pattenson, “the advocates of mechanical music challenged conventional aesthetic assumptions and raised unsettling questions about the technological mediation of musical expression.”[8]

A humanoid flute playing automaton built by Innocenzo Manzetti in 1840

Today, many people listen to most of their music via “technological mediation” — recorded, compressed, audio streamed over the internet. Yet one would be hard pressed to find many people who think of their speakers or headphones as akin to Al-Jazari’s elephant clock or Hero’s wind organ. These devices, although quite different in their mechanisms, all aim at creating sound outside of the need for a live performer, and the mechanism which makes this possible is, in each case, far from neutral. Recordings can be edited and recording sessions can use millions of dollars of equipment; speakers can be surround-sound and seemingly true to life or can compress and distort recordings beyond easy recognition. The ubiquity of high quality recordings and the unprecedented access to them provided by  the internet raises further questions: how does recording change our perception of live performance? How does the existence of “ideal” recordings change the musical practice of performers? What is recorded music “worth” in both a monetary and aesthetic sense? How does this effect those who try to make music for a living?

Mechanical music, then, is all around us, and it is not an unalloyed good. Nor, however, are mechanical instruments or recording technology simply monstrous automatons out to replace flesh and blood musicians. At least in the case of mechanical instruments, what they offer is something categorically different from a human performance, even when they perform a piece playable by human hands. Indeed, many of the mechanical instruments of old have a certain charm to them, an almost animal-like lack of self consciousness that lends their performances an odd sort of sincerity. As opposed to a human performer, these machine’s cannot make mistakes, and while the lack of this risk robs musical performance of much of its emotional weight and excitement, it does lend it a certain whimsy and clunky assurance that would be impossible to recreate with a human touch.

The real competitor for the attention of audience members is the high-quality recording, which can be edited and re-recorded to offer the listener impossibly polished interpretations of their favorite pieces from the comfort of their own home. Still, that same risk and that same humanity is lacking from much recorded music: one knows how it will play, and even if there are mistakes on the recording, they will be the same mistakes each time. To be surrounded by others in a concert setting is to be part of a social world that recordings cannot recreate. Live performance, then, can never be fully captured nor replaced. Although the gap between mechanical music and live, human performance has closed significantly since Hero’s time, that gap seems to be insurmountable at some fundamental level.

Al Jazaris Elephant Clock as depicted in his Book of Knowledge of Ingenious Mechanical Devices

The clock described in Borges’ “Music Box”, is a simple contraption: Japanese water clocks of the type he likely had in mind consist only of two pieces of bamboo, with water flowing from one into the other until the larger vessel tips over with a satisfying clunk. That, too, is mechanical music, albeit of a more elegant kind: simple processes yielding deceptively simple results. Thus, if one is tempted to despair (along with so many) that automation will take away jobs and opportunity from ordinary people, it would be well to remember that musical automatons have been in existence for hundreds of years, and that, even in the presence of thousands of recordings stretching back decades, people continue to seek out live performances. Recordings, player pianos, robot bands, singing software programs — these things will never and can never replace the give and take of live human performance, the social contexts and the interactions which it fosters. Realizing this, we might just be able to take a page from Borges and enjoy the eccentricities of the machines without fear.


A playlist of music played on or written for self-playing, automatic, and mechanical instruments, from the carillon to the barrel organ and beyond.

Early Music Seattle: Mechanical Music


[1]     Jorge Luis Borges and Tony Barnstone, “Music Box.” Poetry 199, no. 6 (2012): 536–37.

[2]     Philip Steadman, “The Automata of Hero of Alexandria.” In Renaissance Fun: The Machines behind the Scenes, 111–32 (UCL Press, 2021), 115.

[3]     Charles B. Fowler, “The Museum of Music: A History of Mechanical Instruments.” Music Educators Journal 54, no. 2 (1967): 45.

[4]     Fowler, 46.

[5]     See Kyle Gann’s The Music of Conlon Nancarrow, Music in the Twentieth Century (Cambridge ; New York, NY, USA: Cambridge University Press, 1995).

[6]     Thomas Pattenson, “‘The Joy of Precision’: Mechanical Instruments and the Aesthetics of Automation.” In Instruments for New Music: Sound, Technology, and Modernism, 18–51 (University of California Press, 2016): 24.

[7]     Paul Hindemith, “Zur mechanischen Musik,” in Aufsätze, Vorträge, Reden, ed. Giselher Schubert (Zurich: Atlantis, 1994), 19.

[8]     Thomas Pattenson, “‘The Joy of Precision’: Mechanical Instruments and the Aesthetics of Automation.” In Instruments for New Music: Sound, Technology, and Modernism, 18–51 (University of California Press, 2016): 20.