…For what is the heart, but a spring; and the nerves, but so many strings; and the joints, but so many wheels, giving motion to the whole body, such as was intended by the artificer? –Thomas Hobbes, Leviathan (1651)
[Figure 1. Bernardino Baldi’s axe-wielding arm. By permission of Carnegie Mellon University Libraries, Special Collections (Call number TJ210.2 .H37 1589).]
The device shown in figure 1 is a hybrid of machine and human parts. Made up of an arm mounted to a notched wheel and weight-driven pulley, it resembles a piece of fanciful clockwork: the disembodied limb of a jacquemart, perhaps — the automaton figure who strikes the hour on late-medieval clocks. But this arm holds an axe, not a mallet.
While the device’s operation is easy to work out — the gear turns, the arm moves, the axe falls, the action repeats — the diagrammatic quality of the image invites the viewer to deconstruct its mechanism: to follow the pull of weight, the clockwise advance of the gear, the arm’s ratcheted swing and drop. And yet there’s also something cryptic here. The image ensnares as much as it instructs, suggesting a kind of aliveness that somehow isn’t confined to the machine’s muscled arm. Instead, the joining of the two — where arm ends and armature begins — makes the image uncanny. It blurs categories that shouldn’t overlap, mingling flesh and instrument, anatomy and machine.
This illustration appears in Bernardino Baldi’s Italian translation of On Automata, a first-century treatise on self-propelled machines by the Hellenistic engineer and technologist Hero of Alexandria (Grillo 2019). Published in Venice in 1589, Baldi’s translation is regarded as the first printed book on robotics, marking the beginning of a rich literary tradition of writing on automated machines. The work reflects Baldi’s humanistic training in ancient languages and classical mechanics, in that it offers a meticulous excavation of centuries-old technology reimagined for sixteenth-century use. The book accordingly embodies a fundamental paradox in Renaissance culture: it is simultaneously ancient and modern, backward-looking and cutting edge.
The word “automaton” offers up a similar contradiction. From the Greek αὐτόματον, the word refers to a thing that moves of its own will. In this sense, automata are related to Aristotle’s definition of an animal: a beast or being that moves by itself. But so-called autonomous devices offer only a semblance of autonomy. Beneath their lively exteriors are cog-and-wheel mechanisms designed to execute a set of predetermined functions. It is this narrower meaning that we now understand by the word: automatons are precisely will-less, their rote actions constrained by mechanical subroutines and human ingenuity. They are more like intricate puppets, in this respect, than proto-robots. But in Baldi’s sixteenth-century context, this was far from a settled view; automatons had not yet been fully drained of life.
In the seventeenth century, just as machines were understood to perform a kind of aliveness, human bodies were described in mechanical terms. French philosopher René Descartes (1596–1650) in his Discourse on the Method (1637) famously likened the parts of human anatomy to machinery. He was particularly interested in the function of the eye, comparing it to a lens embedded into the wall of a shadowy camera obscura. Figure 2 illustrates an experiment Descartes designed to demonstrate the eye’s optics. It shows a bearded figure in a darkened room peering through a bisected eye. The retinal wall has been removed to reveal the cornea and lens, and lines describe paths of light, refracted through the eye’s vitreous humour onto points T, S, and R. Meanwhile, Descartes compared the circulation of blood and bodily humors to a kind of self-contained hydraulic system: vapors and fluids, like those in a heat-driven siphon, rise and fall with the temperature of the body and its parts, while sensations — heat, cold, touch, pain — are channeled through pipe-like nerves to the seat of perception in the brain.
[Figure 2. Descartes’ illustration showing an experiment designed to demonstrate the eye’s optics. By permission of Posner Memorial Collection, Carnegie Mellon University Libraries, Special Collections (Call number Q155 .D43).]
Descartes’ mechanistic anatomy left plenty of room for debate: Dutch mathematician and physicist Christiaan Huygens (1629–1695) derided the ideas of those “new philosophers” who imagined the human body as “nothing but Clocks and Engines of Flesh” (quoted in Riskin 2016, 70). Huygens’ skeptical view was not commonly held, however; Descartes’ philosophy of machine-like humans and humanlike machines remained alive throughout the seventeenth century. Fleshy engines and mechanical bodies animate the work of leading scientists in the period, including Thomas Hobbes (1588–1679; see the epigraph to this essay) and Giovanni Alfonso Borelli (1608–1679), whose groundbreaking treatise on muscles employed Cartesian metaphors to equate muscular contraction with the mechanical action of levers and pulleys. And for Borelli’s contemporaries, the analogy cut both ways: automatons, by imitating the motion of living things, seemed to flicker with life.
This paradox of animated inanimate machines suffuses Baldi’s book. His woodcut image of the axe-wielding arm almost gives the conundrum a physical form: is the device performing its predetermined and purely mechanical function, or is it straining to exert a darkly vital force? Of Automata is full of such “machines that move themselves” (machine se moventi) — to use Baldi’s evocative translation of the Greek — though many operate in a more lighthearted mode: one of Baldi’s (or Hero’s) devices is a carousing automaton Dionysus who pours a libation of milk or wine while automated cymbals and drums clatter and crash in the background.
Reprinted in a second edition in 1601, Baldi’s book tapped into the period’s fascination with automata and other self-propelled machines. But these objects were not exclusive to the Renaissance, nor indeed to Western Europe. Automata and other mechanical tricks have captured the imagination of technologists working across cultures and centuries. In 1206, for example, the Islamicate engineer and polymath Ismail al-Jazari described several self-propelled devices in The Book of Knowledge of Ingenious Mechanical Devices. These included clocks driven by water and heat, mechanical servants, and automata moved by the subtle working of siphons and pumps. This last category included a fountain shaped like a peacock, “in which flowing water triggers little figures to offer the washer first a dish of perfumed soap powder, then a hand towel” (Riskin, “Frolicsome Engines”).
Another book on automata in Special Collections is a work by Japanese engineer Hosokawa Hanzō Yorinao (1749–1796). Printed at the turn of the nineteenth century in Kyoto, the book depicts several Japanese karakuri — automatons, clockwork, and other mechanized tricks and technologies — showing the cross-cultural reach of self-moving machines and their long history in Japan. Titled Karakuri Zui [Manual on Automatic Mechanical Devices], the work pairs instructive notes on how to build and operate Hosokawa’s devices with meticulously drawn woodcut illustrations.
[Figure 3. The copy of Hosokawa’s Karakuri Zui in Special Collections is bound in three slim volumes. By permission of Carnegie Mellon University Libraries, Special Collections (Call number TJ181 .H87 1808).]
The origin of the word karakuri is unknown, though it may derive from the verbs karamu (to twine) and kuru (to reel), suggesting the internal pulleys, springs, and weight-driven mechanisms that propelled the devices. Many of Hosokawa’s inventions, like Baldi’s before them, join puppetry and conjuration, including a “whistling little drummer,” (Kazuo 2012, 179); a tea-serving automaton that glides forward with eerie poise; a sake bowl with an automaton turtle at its base; and a mechanical cockfighting scene complete with a spring-loaded dog who “frighten[s] the cocks who fly away” (199). One of Hosokawa’s most elaborate machines is an automaton fish that swims up an artificial stream. The mechanical tableau stages a Chinese legend about a carp that swims to the top of a waterfall and transforms into a dragon (Kazuo 2012, 169). Mounted on a magnet hidden behind the waterfall’s cascade, the fish “climbs automatically […] transforms into a dragon, creates clouds, and then rises into the sky.” Even Hosokawa’s clouds are given gaseous form. The book instructs the machine’s operator to trap tobacco smoke in a vacuum-sealed box. At the moment of the fish’s transformation, the box’s lid springs open, releasing a puff of acrid smoke. Thus, as Hosokawa concludes in his accompanying commentary, “when the dragon appears the dark cloud appears with it” (177).
[Figure 4. Hosokawa’s metamorphic fish. By permission of Carnegie Mellon University Libraries, Special Collections (Call number TJ181 .H87 1808).]
Hosokawa’s automata emerged from a moment of intellectual and cultural ferment. Exposure to Western science — rangaku, or “Dutch learning”— in Edo Japan (1603–1867) led to innovations across the scientific disciplines. Rangaku scholars studied imported European texts, replicating and adapting foreign mechanical, medical, and scientific principles to fit local traditions. While this Western influence is apparent in Hosokawa’s book, it appears in transmuted form: Baldi’s automata reflect Renaissance humanism’s impulse to recover and refine ancient knowledge, but Hosokawa’s work blends indigenous culture and craft with new and foreign technology.
[Figure 5. Karakuri mechanics. By permission of Carnegie Mellon University Libraries, Special Collections (Call number TJ181 .H87 1808).]
What Baldi’s and Hosokawa’s books show us is that the history of robotics is a great deal weirder and more wonderful than we might expect. These books suggest that, from the sixteenth to the nineteenth century, there was a period of sustained innovation that unfolded in parallel across cultures and continents. Yet early designs for automata did not always result in functioning prototypes, and the early modern world was not uniform in its embrace of self-moving machines. By the end of the seventeenth century, skepticism began to gain ground.
Offering a satirical case in point, Jonathan Swift’s Gulliver’s Travels (1726) describes a machine whose rotating parts generate new knowledge. Described in Lemuel Gulliver’s fictional travelogue of his visit to the island of Laputa, the device — a full “twenty foot square” — comprises tiny “bits of wood” that rotate on “slender wires.” Carrying paper labels marked with individual words, these small wooden blocks are turned repeatedly by scholars to produce a nearly infinite series of word permutations and sentence fragments. When the scholars “found three or four words together that might make part of a sentence,” they promptly recorded them into a growing number of large volumes. The “Professor,” Gulliver’s Laputan host tasked with oversight of the machine, suggests that this massive collection of “broken sentences” might one day be pieced together to form a “compleat body of all the arts and sciences” (Swift 1726, vol. 2, 73–75).
[Figure 6. The Laputan text generator from Swift’s Gulliver’s Travels. By permission of Carnegie Mellon University Libraries, Special Collections (Call number R3724 .G7 1726).]
Besides delivering a joke at the expense of Britain’s machine-obsessed Royal Society, Swift’s Laputan ChatGPT seems to predict our own preoccupation with machines that claim to generate new knowledge (I’d argue that Swift’s account offers the first intimation of a Large Language Model, or “LLM”). But Swift’s tone suggests that he wouldn’t have shared in our enthusiasm for these tools. His depiction of science-by-vending-machine instead ridicules mankind’s misplaced reliance on technology. But it also marks something of a turning point. Following the seventeenth-century’s feverish embrace of automatons and other mechanical conceits, Swift exposes the field’s diminishing returns: what at first glance might appear to be a thinking collaborator turns out to be a box muttering random strings of words that only sometimes cohere into meaningful phrases.
By the eighteenth and nineteenth centuries, European automata had moved beyond practical or ritualistic functions into the realm of entertainment. Jacques de Vaucanson’s Digesting Duck (1739) and Wolfgang von Kempelen’s chess-playing Turk (1770) enthralled audiences with their performances. The duck flapped its wings, swallowed bits of proffered grain, and even “digested” food (the duck’s counterfeit droppings evidently elicited gasps of surprise). Fortunately, the waste-producing mechanism was an illusion: the nibbled grain remained undigested in the automaton’s lifeless stomach. The Turk, similarly, was not an autonomous thinker but a cleverly concealed hoax: a diminutive operator (the device was quite small) guided its moves from a compartment hidden beneath its chessboard.
The impulse to invent self-moving machines did not end with these Enlightenment-era deceptions, of course. While Special Collections mainly curates the prehistory of robotics, it also holds more recent works, including a first edition of Ivan Sutherland’s A Walking Robot (1983). The book features several illustrations of Sutherland’s “Trojan Cockroach,” a machine with an insectile frame and articulated legs designed to traverse uneven terrain. Despite its steampunk and lumbering appearance (figure 9 appears to show the machine leaking large quantities of fluid), the robot’s hulking form maintained a startling equilibrium. Threaded through with joints, cables, and servomechanisms, the Cockroach’s legs distributed the device’s weight to maintain balance with remarkable efficiency. And though it was developed as part of cutting-edge locomotion research — it was an important precursor to the development of robotic bipedal motion — the machine’s segmented anatomy and creeping gait evoke something more ancient. It resembles a mechanical beast that might have appeared in Baldi’s On Automata, among Hosokawa’s karakuri, or in one of Leonardo da Vinci’s notebooks. Recalling these precedents while still signaling something new, Sutherland’s robot encapsulates humankind’s enduring fascination with “machines that move themselves.”
[Figure 7. Ivan Sutherland’s “Trojan Cockroach.” From A Walking Robot (Pittsburgh: 1983). By permission of Carnegie Mellon University Libraries, Special Collections (Call number TJ211 .S9873 1983).]
[Figure 8. The “Trojan Cockroach” during a field test. By permission of the Carnegie Mellon University Archives.]
[Figure 9. The “Trojan Cockroach” during a field test. By permission of the Carnegie Mellon University Archives.]
This essay is an edited excerpt from Rare Books & Ancestral Machines: A Handbook to the Posner Center for Special Collections at Carnegie Mellon University (Carnegie Mellon University Press, 2025).
