Technologies of stage magic: Simulation and dissimulation

Introduction

Magic consists in creating, by misdirection of the senses, the mental impression of supernatural agency at work. (Nevil Maskelyne (Maskelyne and Devant, 1992 [1911]: 110))

The art of conjuring bases its deceptions upon manual dexterity, mental subtleties, and the surprising results which are produced by the sciences … The physical sciences – generally chemistry, mathematics, and particularly mechanics, electricity, and magnetism – supply potent weapons for the use of the magician. (Jean Eugene Robert-Houdin, 2006 [1868]: 29)

The craft of stage magic has come under various kinds of scrutiny for what it might reveal about the workings of human minds and cultures. Psychologists and neuroscientists have studied its implications for mental processes, particularly perception and attention (e.g. Hyman, 1989; Macknik et al., 2008). ¹ At least one sociologist has studied the production of the magician’s stage persona and forms of interaction with audiences (Nardi, 1984). More contextual accounts have made links to theatre (e.g. Mangan, 2007) and early cinema (e.g. Solomon, 2010). Others have sought still broader interpretations of conjuring’s place in society and culture (e.g. Cook, 2001; Coppa et al., 2008; During, 2002). Simon During’s (2002) epic and insightful book charts how, through its entrepreneurial show-business methods, ‘secular magic’ emerged as distinct from the occult, but carried forward preoccupations with things such as cruelty, the exotic, and the sex and race of its performers (p. 1). Continuing an interest in stage magic, but taking a different direction, this article considers it as a site to learn about technology – particularly, the ways in which technologies are performed and received, and the deeply entangled agencies of humans and machines.

The historical focus for the present account is Europe in the 19th and early 20th centuries, a time and place of great inventiveness and success for magical entertainment, culminating in a ‘Golden Age’ from about 1890 to 1914 (Dawes, 2007: 109). Conjuring shows of this time were deeply technological accomplishments. Typically visible on stage was the figure of a magician, often with an assistant or two, manipulating various props now familiar to magic audiences, like wands, bird cages, steel rings and person-holding cabinets. Also at play were hidden gadgets, sometimes known as ‘gimmicks’, with unfamiliar names like reels, pulls, the servante and the gooseneck. And beyond the immediate stage, reaching back in time, was an extended network of magicians, magical inventors, apparatus, apparatus builders, instructional texts and so on. In the climax of a magic trick, it was a particular alignment of all these actors and arrangements that gave rise to what the conjuring theorist Nevil Maskelyne described as ‘the mental impression of supernatural agency at work’.

Take the following trick performed by the great French magician Jean Eugene Robert-Houdin in Paris in the 1840s. He brought a wooden chest on stage and showed that he could lift it freely. But when spectators were invited to do likewise, they found it too heavy to raise off the ground. The Light and Heavy Chest illusion, like most conjuring tricks, rested on the impossibility of breaking commonsense physical laws – in this case, the conservation of mass. But equally essential, to this or any magic effect, was the way simple material props expressed a strange agency through the re-ordering of natural events in which they participated. The wooden chest was somehow complicit in allowing the magician to lift it while preventing others from doing so.

During the 18th and early 19th centuries, the style of conjuring performances and public displays of science had acquired a common format of ‘experiments’, ‘apparatus’ and ‘effects’, presented with wand or pointer. ² Over this same period, conjuring had taken on many of the techniques and materials of laboratory science. But this accelerated greatly in the second half of the 19th century, as a variety of new technologies found their way into magic shows, such as electricity, electromagnets, large glass mirrors used with sophisticated optics and thin steel wires arranged under principles of load distribution. ³ Along with these came scientific ways of thinking. By the early 20th century, a key instructional text for magicians declared a taxonomy of conjuring methods with three ‘orders’: ‘manipulative’, involving sleight of hand; ‘mental’, involving calculation; and ‘physical’. The physical order was seen as ‘by far the most extensive and important’ and contained six ‘classes’: ‘mechanical, optical, acoustic, electrical, chemical and molecular’ (Maskelyne and Devant, 1992 [1911]: 113–114).

As these back-stage technologies grew in scale and sophistication, something else was happening front-stage. A modern style of conjuring emerged around the middle of the 19th century, characterized by minimal apparatus and a natural manner of performance. What this entailed was a new way to present each arrangement of apparatus so that its strange effects were apparent to spectators but its inner workings were not. Understanding the modern style is the central theme of this article, with the promise that it holds more general insights for the way technologies are presented and received.

To investigate the deployment of technology in the modern style of stage magic, I will draw on Lucy Suchman’s (2007) notion of ‘human–machine reconfigurations’, which she developed to account for certain productions of computerized technologies that aspire to be intelligent, including ‘worldly’ robots (Suchman, 2011: 119) and the smart features of interactive gadgets and other human–technology assemblages, such as those involved in endoscopic surgery. Building on the insights of actor–network theory, especially Latour (1993), Suchman argues against the view of agency as something inherent in a singular person, machine or object. Rather, it is better understood as relational and enacted in subtle reconfigurations of a network of humans and artefacts. Applying this to stage magic allows us to see it as an early form of simulation. Special arrangements of apparatus were used to produce the appearance of supernatural agency through translations of various other agencies within the network of performers, assistants, apparatus builders, other magicians, instructional texts and spectators. From this viewpoint, I will attempt to show that in simulating supernatural agency, magicians deployed a set of deceptive techniques that will be described collectively as ‘dissimulation’. If simulation involves the production of an effect, then dissimulation refers to the complementary means by which spectators are prevented from knowing about the secret methods and mechanisms behind that effect. Importantly, dissimulation implies more than concealment; it implies that the secret methods and mechanisms are rendered absent.

The use of the terms ‘simulation’ and ‘dissimulation’ here is motivated partly by their appearance in the writings of a 20th-century conjuring theorist, Dariel Fitzkee (1945: 132). Fitzkee appears to have borrowed the sense of Francis Bacon’s very early and prominent usage of the dualism in his 17th-century essay on the various ways people conceal their true selves from each other. According to Bacon, simulation is when a ‘man industriously and expressly feigns and pretends to be that he is not’, and dissimulation is when ‘a man let’s [sic] fall signs and arguments that he is not as he is’ (quoted in Pitcher, 1985: 77). In another prominent use, consistent with the meaning developed here, Jean Baudrillard (2006 [1994]) distinguished the terms in his writings on simulacra: ‘To dissimulate is to pretend not to have what one has. To simulate is to feign to have what one doesn’t have’ (p. 3).

The next three sections of the article develop this account of the modern style of stage magic, explaining its sources in magicians’ writings and elaborating the key notion of dissimulation through the analysis of specific tricks and performance techniques. The naturalness and minimalism of the modern style are examined as tools of dissimulation, in relation to both apparatus design and the performer’s manner. The following section then considers the broader significance of this view of stage magic by drawing a parallel with public performances of science and technology. While the term ‘dissimulation’ has not been widely used in Science and Technology Studies (STS), it is seen to overlap with the familiar ‘erasure of human labors’ around technological performances (Suchman, 2007: 238). A specific comparison is made with recent productions of computerized life forms, especially those of social robotics (e.g. Alač et al., 2011; Brooks, 2002), these being the focal technologies of Suchman’s account. This raises additional comparison with the mechanical automata of the 18th and 19th centuries that often shared the stage with magicians (Riskin, 2007). Finally, the article concludes by considering what has been revealed about stage magic’s peculiar reconfigurations of human and machine agency.

The modern style of conjuring: simulation and dissimulation

A conjuror is not a juggler; he is an actor playing the part of a magician, an artist whose fingers have more need to move with deftness than with speed. (Jean Eugene Robert-Houdin, 2006 [1868]: 39)

[T]he magician’s audience is not called upon to sympathize with human emotions, but to take an interest in things which are entirely out of the common, and in events which are only interesting from the fact that they occur. (Nevil Maskelyne [Maskelyne and Devant, 1992 (1911): 55])

A conjurer’s business is not to deceive, but to evoke wonder by artistically perfected magical effects. (Samuel Sharpe, 2000 [1936]: 248)

A background for the present account is provided by magicians’ own histories of their craft, notably those of Sidney Clarke (1926–1928), Henry Ridgely Evans (1928), Geoffrey Lamb (1976), Milbourne Christopher (1973), Edwin Dawes (1979) and Jim Steinmeyer (2001, 2003). But the key sources are a small body of reflective instructional texts written by and for magicians on the principles of good conjuring. ⁴ The authors, who articulated the tenets of the modern style, are described here as conjuring theorists to mark their special status. Only a few figures in the history of conjuring have succeeded in publishing purely theoretical treatments of the craft, as opposed to the numerous practical manuals on how to perform specific tricks. Robert-Houdin, the performer of the Light and Heavy Chest illusion, took early ownership of the modern style through his influential performances and writings, in particular, his widely read autobiography Confidences d’un prestidigitateur (Robert-Houdin, 2006 [1858]) and his two most famous instructional books, Les secrets de la prestidigitation et de la magie (Robert-Houdin, 2006 [1868]) and Magie et Physique Amusante (Robert-Houdin, 2006 [1877]). ⁵ Two later works confirm the modern style’s significance throughout and beyond the golden age and form two more anchors of the current account. Our Magic (1992 [1911]), by Nevil Maskelyne and David Devant, expresses the thinking of Victorian England’s most successful conjuring company and is perhaps the clearest statement of modern magic’s ideas. Neo-Magic: The Art of the Conjurer (Sharpe, 2000 [1932]), with related writings by the esteemed commentator Samuel Sharpe, marks the end of the golden age and provides both a critique of the modern style while endorsing many of its tenets. ⁶ These three sources, of Robert-Houdin, Maskelyne and Devant, and Sharpe, have been intermittently republished up to the present day and are still widely cherished in societies of magicians.

The modern style of conjuring arose at a time when public displays of science were in sharp decline (Shapin, 1988). It was as if conjuring was taking centre-stage in the ‘theatre of marvels’ being vacated by science, until it was later succeeded by film in the 1920s (Solomon, 2010: 4). ⁷ Magicians faced the challenge of how to continue to draw on science and technology in the face of advancing public knowledge and diminishing interest in scientific spectacles. Thinly disguised and theatricalized science would no longer suffice. By the end of the 19th century, even electricity had lost its power to amaze: ‘We cannot now hang up a drum with electrical tappers concealed inside it, and expect an audience to be mystified by hearing that drum played invisibly. The day for that sort of thing has gone by, never to return’ (Maskelyne and Devant, 1992 [1911]: 145).

Curiously, given the public turn against scientific spectacles, the essence of the modern style was for the magician to assume the manner of the scientist-lecturer and to perform with elegant pieces of apparatus that were as simple and ordinary as the unusual circumstances of stage magic would allow. Robert-Houdin wrote of his shift away from the earlier theatricality of conjuring, epitomized by the robe, the conical hat and apparatus shaped and decorated to appear mysterious and potent. ⁸ His novel stage set in Paris in the 1840s became emblematic of the new approach: a naturalistic drawing room containing minimal and relatively minimalist items of Louis XV furniture, well-lit and without drapery (see Figure 1). As fitting for this scene, Robert-Houdin himself appeared in an evening suit. Promoting ‘propriety and good taste’, he eradicated the persona of the showman. He spurned anything that might suggest artifice: disclaimers (‘nothing up my sleeve’), ‘pedantic and affected’ patter, the use of confederates in the audience, pretended accidents or fast gestures to cover sleight-of-hand (Robert-Houdin, 2006 [1868]: 31–33). In their place, he promoted simple natural actions.

OPEN IN VIEWER

Figure 1.

Robert-Houdin’s stage set, Paris 1840s.

It is unlikely that there was a single point of origin of the modern style. The wearing of evening suits rather than costumes, for example, happened sporadically before Robert-Houdin (Clarke, 1926–1928). But from the 1840s onwards, leading performers strived to produce more elegant and natural performances, including the famous Austrian Compars Herrmann (Cook, 2001) and Prussian-born Wiljalba Frikell. Frikell was an extreme case, performing with a bare stage supposedly after a fire had destroyed his equipment. The English conjuror Joseph Hartz also pushed the style far by developing an act around transparent glass props (Ridgely Evans, 1928). By Maskelyne and Devant’s (1992 [1911]) treatise, the new thinking was expressed through 24 rules of performance that emphasized clarity, simplicity and ordinariness, and the genuineness of the performer’s manner: ‘Let each magical act represent a complete, distinct, and separate entity; comprising nothing beyond one continuous chain of essential details, leading to one definite effect’ (p. 23).

Some cultural theorists have described the scientific manner of Robert-Houdin and others as an affectation motivated by a yearning for bourgeois respectability (e.g. Cook, 2001; During, 2002). Cook (2001) puts it: ‘First and foremost, they brought the magician fully into the more respectable confines of the urban middle-class theater’ (p. 167). For During (2002), ‘Robert-Houdin negated the triviality and cultural nullity of magic by bringing to the stage the prestige of the inventor and scientist’ (p. 120). But the modern style also marked a profound shift in the experience of spectators and brought a new modus operandi of deception that sought to erase any clue to the machinic basis of the trickery. The modern style removed any sense of dramatic pretence by presenting the performance as something real. The audience of the Light and Heavy Chest illusion did not see an actor pretend to lift a heavy object, but saw a real person actually lifting a wooden box that other real people were unable to lift. The effect of magic tricks was thus enhanced by setting them not in a dramatic story-world but in the present reality inhabited by spectators. This new format of realist theatre created a deeply contradictory experience for spectators, who now confronted something simultaneously real and unreal. Maskelyne and Devant, quoted above, pinpointed this contradiction, stating that the impossible events of a magic trick are ‘only interesting from the fact that they occur’. And in Sharpe’s line above, the contradiction was expressed from the magician’s perspective as the need to produce ‘magical effects’ yet ‘not to deceive’. The contradictory experience is also expressed as the juxtaposition of the natural and the supernatural in Sharpe’s (2000 [1932]) definition of conjuring: ‘Conjuring is the Art of pretended Magic. It consists in producing supernatural effects by natural means, the details of which are obscure to the general public’ (p. 15).

While the magicians of the modern style likened their craft to science, we might also see them as working in what Andrew Pickering (2010) calls the ‘performative idiom’ (p. 19). Like the cyberneticians studied by Pickering, magicians built special machines that probed audiences’ understanding of artefacts and played with a sense of agency around the human–machine boundary. ⁹ More generally, conjuring can be seen as an early form of simulation in which impossible sequences of events were simulated on stage. For some, this interpretation might seem to perpetrate a muddle of history. Evelyn Fox Keller (2003) has argued that the term ‘simulation’ did not have its current meaning of imitative modelling until the era of electronic computing following World War II (WWII) and that before this time ‘it invariably implied deceit’ (p. 198); she cites earlier definitions of simulation from the Oxford English Dictionary (OED) including ‘A Deceiving by Actions, Gestures, or Behaviour (1692)’ and ‘a Pretense of what is not (1711)’. Curiously, the word and its derivatives make occasional but prominent appearances in the pre-WWII writing of magicians. Maskelyne and Devant (1992 [1911]), for example, gave ‘simulation’ prominence in their treatise, defining it as ‘giving apparent existence to things that do not exist, or presence to things that are absent’ (pp. 122–123). ¹⁰ While magicians’ use of the term indeed referred to a form of deception, it was closer to a sense of deceiving the senses rather than a moral deceit. The premise of stage magic, well established by Robert-Houdin’s time, was that audiences were aware that the magic was a pretence. Rather than deceit, Maskelyne and Devant’s use of ‘simulation’ referred specifically to techniques of acting as if something was true, like pretending to hold an egg in a hand that was actually empty. These techniques bear at least some resemblance to the computer-age notion of imitative modelling in the sense that they imitated a sequence of events that were realistic but not real.

Alongside its simulations, the naturalism of the modern style’s realist theatre can be seen as enacting complementary techniques of dissimulation that erased the presence of the methods and mechanisms that lay behind the simulated effects. Maskelyne and Devant (1992 [1911]) did not use the term ‘dissimulation’, but expressed it as ‘distraction’ and ‘disguise’ (p. 117). Distraction meant leading spectators’ attention away from secret mechanisms or actions, while disguise meant hiding those things within the structure of apparatus or performance. They stressed the need to go beyond simple concealment: ‘magical appliances should be so constructed that their inner devices are not concealed by a mere covering of some sort, but are disguised by blending with the general structure’ (Maskelyne and Devant, 1992 [1911]: 121). Again, this echoes Francis Bacon on dissimulation and his point that ‘closeness, reservation, and secrecy’ were not enough to hide one’s true self from others, ‘for men are too cunning to suffer a man to keep an indifferent carriage’ (quoted in Pitcher, 1985: 77). It was similarly never enough for spectators simply to fail to see how the trick was done; they had to be confident that they had seen all they needed to see. How this dissimulation was achieved on the magic stage is the focus of the next two sections, first to see how it was inscribed in apparatus and next to see how it was enacted through the performer’s manner.

Magical apparatus: dissimulation through the empty box

It is easy enough, no doubt, to play the conjuror without possessing either dexterity or mental ability. It is only necessary to lay in a stock of apparatus of that kind which of itself works the trick. This is what may be called the ‘false bottom’ school of conjuring. Cleverness at this sort of work is of the same order as that of the musician who produces a tune by turning the handle of a barrel-organ. Such performers will never merit the title of skilled artists and can never hope to obtain any real success. (Jean Eugene Robert-Houdin, 2006 [1868]: 29)

When appliances are so designed as to show that they are mere covers for mechanical trickery, a spectator’s only source of interest is in wondering how the machinery is constructed. (Nevil Maskelyne [Maskelyne and Devant, 1992 (1911): 120])

To produce convincing magic there must be a total absence of visible trickery and skill. There is no more magic in the average conjuring performance than in a box of wire puzzles, for the results are obviously produced by trickery. (Samuel Sharpe, 2000 [1932]: 15)

How did modern conjuring work its format of technological dissimulation? How could it render methods and mechanisms absent, when audiences knew that the magic was artificial? A key site of dissimulation was the design of apparatus. As seen above, the conjuring theorists poured scorn on performers who used overtly deceptive objects: bulky furniture and baggy costumes that declared themselves as hiding places. ¹¹ Instead, they advocated a form of apparatus that I will describe through the metaphor of an empty box, a cousin of the familiar black box. Being an empty box implies that the internal workings of an apparatus are not just forgotten about or obscured but are clearly seen to be absent. It also implies, as with Robert-Houdin’s transparent stage set, that items of apparatus were clearly seen to be isolated and not components of larger machinic arrangements.

To see how dissimulation was achieved through apparatus designed and deployed in the form of the empty box, three tricks that illustrate the modern style are now examined. In all three, apparent supernatural agency was manifest partly through the reconfigured agency of an assembly including a performer, visible and concealed apparatus and assistants, and magical inventors and designers. First, consider again Robert-Houdin’s Light and Heavy Chest illusion, which he performed in the early part of his short stage career from 1845 to 1853 (see Figure 2). He had a smaller version of the illusion involving a cigar-box at his home in Blois where he used it to entertain dinner guests (Klein, 1988). ¹² In both places, the effect was brought about by the force of a hidden electromagnet that could be secretly turned on or off. In the theatre version, it was concealed within the board that Robert-Houdin used to walk from the stage to the pit, on which the chest was placed. An off-stage assistant, who could hear the action on stage, switched the current on and off at the right moments (Robert-Houdin, 2006 [1877]). The illusion is a clear example of one of science’s ‘surprising results’ being brought to the service of magic.

OPEN IN VIEWER

Figure 2.

The Light and Heavy Chest illusion, Paris, 1840s.

Electricity offered great possibilities for dissimulation, as expressed later by Maskelyne and Devant (1992 [1911]) through a principle of the ‘Conveyance of Power through Supports’ (p. 146). Electrical power could be fed to a piece of apparatus through a floor and a stand, without any visible motion. The electromagnet was supreme in this regard, stored safely out of sight, its magnetic field reaching invisibly to objects on display. The result for the Light and Heavy Chest was a near-perfect and near-literal instance of the empty box form: a simple chest seen to be without workings and seen to be isolated from other possible mechanisms. However, it was possible that at least some people in the audience of Robert-Houdin’s shows would be aware of the electromagnet, following its invention by Sturgeon in 1825 and the many attempts to find new applications (Schiffer, 2008). So even with its perfect concealment, the Light and Heavy Chest demanded a deeper form of dissimulation. Most importantly, the presentation of the trick carefully disassociated the method from the effect. An electromagnetic force was used to simulate a gravitational force. The use of a wooden box dispelled any connection with magnetism; in reality, the bottom had a metal core. Also important was that, for a mid-19th-century audience, two of the electromagnet’s significant properties – its great force and its sudden on/offness (Schiffer, 2008) – were not fully grasped.

The form of dissimulation used in the Light and Heavy Chest can be seen in terms of Suchman’s notion of human–machine reconfiguration. The effect was produced by an assembly including the performer, the hidden assistant, the inventors and builders of the apparatus, the chest, its metal core, the electromagnetic force and concealed wiring. By neutralizing the chest and rendering its mechanism absent, the act disconnected the agencies of the assembly. The result was a simulated effect of a weight-changing object cut free from the network of labours that produced it. An important aspect of the dissimulation was the manipulation of what the audience took to be necessary elements of performance and what appeared to be incidental. The chest was clearly a necessary and focal item, while the board, where the chest ‘happened’ to be put, appeared as a contingent and almost accidental circumstance. From the magician’s back-stage view, the board was a pivotal component of the machinic arrangement, providing the platform above the electromagnet on which the chest must be placed. Rendering the method absent rested partly on this vital element disappearing behind a fabricated ordinariness.

For our next trick, we move forward to 1865 when an English magician, Colonel Stodare, presented a much awaited and successful Sphinx Head mystery at the Egyptian Hall in London, a venue for the public display of natural and scientific curiosities. The trick is well described in magicians’ own histories (e.g. Lamb, 1976; Steinmeyer, 2001, 2003). Stodare entered the stage carrying a box inside which, it had been promised through notices appearing in The Times newspaper, was the dismembered head of an ancient Egyptian. The front of the box was opened to reveal the head, which suddenly came alive and answered questions from the magician. The head was that of a heavily made-up actor, and the method rested on the box being placed on a small table, standing in a curtained alcove, in which the body of the actor was concealed (see Figure 3). The secret of this concealment, still used by magicians today, is based on familiar everyday objects arranged to produce a surprising optical result. While this method and its careful calculations drew on techniques from laboratory science, its detailed form was developed within the world of magicians and science entertainers.

OPEN IN VIEWER

Figure 3.

The Sphinx Head, Egyptian Hall, London, circa 1865.

The Sphinx Head enacted a similar kind of human–machine reconfiguration as the Light and Heavy Chest. An isolated living head emerged from an arrangement of agencies expressed by the performer, the concealed actor, the designers of the box and the table and its mechanism of concealment. By initially entering the stage carrying only the box, the performance established the prop as an isolated locus of effect. An empty box form was created, and the attached human body was rendered absent through its unsuspected concealment in an apparently transparent setting. The Sphinx Head again illustrates how dissimulation rested on the fabricated ordinariness of apparently incidental elements of apparatus. The table in the curtained alcove, where the box happened to be put down, was cast simply as a means of display, leaving the focal element of the box as an isolated prop.

Also important to the Sphinx Head was the dissociation of method from effect. It is significant to compare an earlier, less successful, trick based on the same optical method of concealment. The Proteus Cabinet had been presented in the same year at the Royal Polytechnic Institute in London, devised by Thomas Tobin, a Secretary and lecturer at the Institute and also a trained architect. In that illusion, people were seen to appear, disappear or change costume as they went in and out of a large cabinet. This was puzzling but did not create the sensation of the Sphinx Head. The problem lay in overtly presenting the cabinet as a special machine to make things disappear or change (Steinmeyer, 2001, 2003). It was a failure to dissimulate an apparatus of the sort castigated by the conjuring theorists, as seen in the lead quotes of this section. Once spectators had made an attribution to the cabinet as a special apparatus, the experience of magic was lost although they did not understand the mechanism. In contrast, Colonel Stodare’s approach with the Sphinx Head was to transform the same mechanism into the effect of a live dismembered head, just as the Light and Heavy Chest transformed electromagnetism into gravity. This was dissimulation in action: dissociating method from effect and shifting the mechanism into the incidental margins of performance.

For our third and final trick, we move forward again to the early years of the 20th century and David Devant’s Educated Fish illusion, described in Maskelyne and Devant (1992 [1911]). ¹³ Here again, the secret method was an ingenious machine, although this time it involved no formal scientific techniques or knowledge. An audience was shown four goldfish in a large glass bowl resting on a metal stand (see Figure 4). The conjuror dropped 26 wooden tablets, marked with the letters of the alphabet, into the bowl where they sank to the bottom. An audience member was invited to select a word from a newspaper, which the fish then apparently spelled out by causing its constitutive letters to rise to the surface of the water in sequence. In reality, the rising letters were from another set of tablets made of buoyant paraffin wax. They were hidden behind a shiny metal plate which reflected the surrounding water and so became invisible. The buoyant tablets were held down in metal boxes, but each could be released by pulling on a fine thread connected to the lid of its box. Together, these fine and imperceptible threads passed up out of the bowl, down the back of the stand and through a slot in the stage floor to where a hidden assistant held them in readiness.

OPEN IN VIEWER

Figure 4.

The Educated Fish, London, St George’s Hall, circa 1908.

Devant’s Educated Fish further illustrates the peculiar human–machine reconfiguration of stage magic. The agency manifest in the goldfish was produced by reconfiguring an assembly involving the performer, the hidden assistant, the special goldfish bowl of visible and concealed letter tablets and its system of boxes and fine threads. An empty box form was created in the shape of a fully transparent goldfish bowl and stand, apparently devoid of mechanism and effectively detached from other elements in the set. As with the other two tricks, the illusion rested on certain vital elements disappearing behind their incidental ordinariness. The fish, bowl and tablets were cast as focal to the effect, while the metal stand, where the fish happened to be displayed, receded as an incidental contingency. Without spectator awareness of all the components, the machinic arrangement as a whole was again rendered absent, leaving the focal elements as isolated and inert.

Magical performance: dissimulation through acted naturalness

Some conjurors use an excessive amount of gesture in order to conceal their manipulations. This is wrong. Genuine conjuring demands perfect simplicity of execution. The more simple and natural the movements of the performer, the less likely is the spectator to detect the trick. (Jean Eugene Robert-Houdin, 2006 [1868]: 32–33)

A young performer often imagines that the ease of manner and ready flow of language possessed by his seniors are more or less spontaneous in origin … But the fact is that practically every word and action has been mostly carefully rehearsed, before the presentation was ever put before the public. (Nevil Maskelyne [Maskelyne and Devant, 1992 (1911): 78])

The actions which appear to be the most natural on stage are, in reality, the least natural ones: i.e., those which have been most rehearsed. (Samuel Sharpe, 2000 [1932]: 138) in my opinion, too many magicians present their effects in a commonplace way on stage … They ‘do their stuff’ without a vestige of dramatic presentation and do not try to create an atmosphere of suspense and mystery; in fact, they are too off-hand. (Samuel Sharpe, 2000 [1938]: 293))

In tandem with the design of minimal and apparently transparent apparatus and stage sets, there was something else going on in the modern style of conjuring that was perhaps even more deceptive. But now the target was the performer who, as operator of the apparatus, was also part of the arrangement that worked the trick. Just as agencies captured in apparatus had to be carefully erased, so too the instrumental role of the magician was in need of dissimulation. This need extended to on-stage assistants, as the magician Edwin Sachs (1885) wrote, ‘The worst is the one who conveys by his appearance and actions that he “knows all about it.” The spectators at once attribute the greater portion of the results to his agency – not incorrectly perhaps’ (p. 394).

This second kind of dissimulation was, and still is, talked about by magicians as naturalness. The performer was to act naturally and be natural. One thing this entailed was modesty, rediscovering the manner of Robert Boyle and the experimental philosophers of the 17th century as described by Steven Shapin and Simon Schaffer (1985). An ‘earnest desire to please’ wrote Maskelyne and Devant (1992 [1911]) of the required disposition (pp. 96); ‘conjuring should seem more magical than clever’, followed Sharpe (2000 [1932]: 41).

More significantly, the naturalness of the performer was another move in the human–machine reconfiguration of conjuring. The approach was to assume a persona that clearly disentangled itself from the secret work of operating the trick. As with the dissimulation of apparatus, but to an even greater extent, this relied on creating a sense of incidental ordinariness. A key step was for magicians to act as themselves on stage. Maskelyne and Devant (1992 [1911]) wrote of the need for the public performance to resemble something done in private and formalized this approach in their principles: ‘Never attempt, in public, anything that cannot be performed with the utmost ease in private’ (p. 67).

By being themselves, magicians were able to perform actions on stage that were simple and ordinary. This might seem at odds with the extraordinary and even bizarre elements of conjuring tricks. However, within their extraordinariness, most tricks are largely taken up with what appears as ‘articulation work’ (Star and Strauss, 1999: 10), referring to those incidental actions that are outside of a plan but are needed to keep things on track. Leading up to the brief magical ‘climax’ of the trick, these occur as part of the long ‘complication’ or ‘preparation’ phase (Sharpe, 2000 [1932]: 65): picking up a saw, arranging cups, ropes, card or coins on a table to be inspected, or getting a grip on a hat to show it as empty. The evident simplicity and ordinariness of these actions reinforced the modern style’s format of realist theatre and gave spectators a sense of seeing all they needed to see. The conjuring theorists further emphasized the need to ensure that this transparency did not escape the attention of the audience:

While the magician must use all his art to disguise and cover up what he does not require to be seen, he is equally bound to make sure that every moment and every detail that ought to be seen shall be seen. (Maskelyne and Devant, 1992 [1911]: 122)

Sharpe continued this line, calling for ‘the Preparation to be made deliberately so that there is no chance of the audience missing or forgetting an incident’ (Sharpe, 2000 [1932]: 65).

To manufacture a sense of ordinariness in every performance, the apparently natural behaviour of modern magicians was as tightly scripted and produced as for any conventional play. The articulation work around the deployment of props was exactingly choreographed to seem of the moment, while providing perfectly synchronized covers for secret actions and avoiding revelations or clues about the method. Some props were included simply to contrive apparent reasons for the articulation work to unfold as it must. On the value of using a magic wand, for example, Angelo Lewis (1876) wrote, ‘If it is necessary, as frequently happens, to turn his back upon the audience for an instant, the momentary turn to the table, in order to take up or lay down the wand, affords the required opportunity’ (p. 5). In the Sphinx Head illusion, to take another example, the casual and seemingly incidental movement of the performer across the stage followed a carefully pre-arranged path that avoided entering regions that would disturb the optical effect. In emphasizing the need to script the un-scripted, in the context of distractions, Maskelyne and Devant (1992 [1911]) wrote, ‘The more haphazard the distraction appears to be, the less likely it is to arouse suspicion as to its true purpose’ (p. 118). Herein lay the great emphasis on practice and rehearsal in conjuring, including all ‘incidental patter and business’ (Maskelyne and Devant, 1992 [1911]: 77).

Something that was known to blow the cover off the natural performance was the repetition of a trick still fresh in spectators’ memories. What had seemed incidental on first viewing was seen as machinic second time round. This is part of the rationale behind the magicians’ rule never to repeat a trick twice on a single occasion and to some extent the rule never to say in advance what will happen during a performance – both rules being advocated, with various qualifications, by Robert-Houdin, Maskelyne and Devant, and Sharpe. This need to avoid repetition captures the contradiction at the heart of the modern style: exactingly designed and constructed performances, repeated with machinic reliability show after show, but each time appearing effortlessly of the moment.

The acted naturalness of the performer was carefully crafted in concert with the design of apparatus to appear transparent. Together they created a powerful dissimulation of methods and machinery, and gave rise to the new elegance and minimalism of the modern style. This drew large audiences and many followers who attempted with varied success to emulate the widely acclaimed performances of Robert-Houdin, Devant and their peers. But by the 1910s, the golden age of conjuring began to recede, and by the 1920s it was into decline as film emerged and became popular. It was not simply that film displaced magic, a point made by Solomon (2010), because conjuring was already losing out to other theatre productions (Steinmeyer, 2003). Rather, conjuring lost its power to entertain in a similar way perhaps that public displays of science had in the previous century. Samuel Sharpe, the most prolific writer on conjuring theory, was an admirer of the modern style but saw how its minimalism was no longer working; he was critical, for example, of the dry ‘lecture-room atmosphere’ created by John Maskelyne, father of Nevil, (Sharpe (2000 [1932]:86) and he endorsed a negative review in The Observer newspaper of the Zanzigs performing at the Alhambra in London in 1923: ‘… he looks more like an electrician at a power station than an accredited follower of Cagliostro… Theirs is the cold, clear light of the arc lamp, rather than the spectre-haunted gloom of the wizard’s dip …’ (Sharpe, 2000 [1932]: 58–60). Sharpe called for stage magic to recover a dramatic atmosphere and to reverse what he saw as the ‘off-hand’ style of many performances. But his call was not easily accommodated within the modern style’s format of realist theatre, creating a dilemma that marked the end of the golden age.

Stage magic and other performances of technology: a comparison with computerized life forms

An experiment is transparent when the apparatus and procedures appear to contribute nothing to what the experiment shows. Like the preparations and manipulations of a magician, they are unobtrusive, though not necessarily invisible. This is how Faraday persuaded his audiences to perceive the great lecture theatre at the Royal Institution as Nature’s school – not Faraday’s. (David Gooding, 1985: 107)

Cog had a face, it made eye contact, and it followed my movements … although I knew Cog to be a machine, I had to fight my instinct to react to ‘him’ as a person. (Sherry Turkle, 2011: 84)

Although still an imposing figure of a robot, what struck me most powerfully about Cog was the remainder of its ‘body’ not visible in media portrayals. The base of Cog’s torso was a heavy cabinet from which came an extraordinarily thick sheaf of connecting cables, running centaur like to a ceiling-high bank of processors that provided the computational power required to bring Cog to life. (Lucy Suchman, 2007: 246)

These early-morning conversations were quirky and entertaining, and there were moments when Phil seemed truly alive. Indeed, it was the offbeat and unpredictable nature of his response that gave him personality. Without them, without the vestiges of a real man’s mind threaded through, he would be another bland chatterbox. He would be, well, robotic. (David Dufty, 2012: 184)

A picture has been presented of 19th and early 20th-century stage magic as a human–machine reconfiguration. Its apparent effects of supernatural agency are seen to have resulted from translations of various other agencies within an assembly of performers, visible and concealed apparatus and assistants, apparatus builders, magical inventors and spectators. This section now considers this picture of stage magic in relation to other kinds of technology display and performance. As a specific focus, a comparison is made with recent productions of computerized life forms, these being the focal concern in Suchman’s account of human–machine reconfigurations. But before turning to that specific case, a more general parallel between stage magic and technoscientific displays will be sketched.

It has been seen how modern conjuring’s techniques of dissimulation were deliberate manipulations of audience experience that created apparently spontaneous effects on stage by erasing an extended history of painstaking back-stage labours, their orchestration and associated machinery. This careful separation of front- and back-stage experience echoes foundational scholarship in STS on public presentations of science, including Steven Shapin and Simon Schaffer (1985) on Boyle’s air-pump displays, Bruno Latour (1988) on Pasteur’s anthrax demonstrations and David Gooding (1985) on Faraday’s public lectures at the Royal Institution. In all of these cases, unrealized by audiences, an extended network of helpers and a long history of trial and error lay behind the production of perceptually discernable and often stimulating effects for public consumption. A parallel can also be drawn with more recent forms of public science, as in Hélène Mialet’s (2003) study of the invisibility of assistants and assistive technologies around the work of physicist Stephen Hawking.

Not only do these studies chart the extent of back-stage labours, but also they show how care was taken during performance to draw attention to particular elements while minimizing or obscuring others. Most famous, perhaps, is the hiding of Boyle’s assistants who worked the air-pump (Shapin, 1989), so avoiding the perception that the effects of nature were somehow a result of their agency. Along with the assumed modesty of the scientist presenters, these techniques resemble those of magicians to remove, or at least disentangle, the presence of agents that might be suspected of working the trick. Similarly, scientists’ efforts to cast apparatus as a transparent window onto nature resemble the shift of the modern style of conjuring towards minimal props, procedures and sets. Although such connections to stage magic do not figure largely in these works of STS, they have not gone entirely unnoticed, as in the passing comment by David Gooding above.

Studies of public technology demonstrations, as distinct from science shows, reveal a similar ‘erasure of human labors’ (Suchman, 2007: 238). Taking the resemblance to conjuring further, they also sometimes point to the possibilities of audiences being deceived. For example, Harry Collins (1988) described a televised demonstration in the United Kingdom of nuclear waste containers withstanding a high-speed train crash. He argued that this was a kind of ‘trick’ inasmuch as it was presented as an open experiment when in fact it was carefully designed by the British Central Electricity Generating Board to produce a particular outcome (Collins, 1988: 730). Critics claimed that the result depended on favourable conditions of which the public remained unaware. Reinforcing the sense of an open experiment, the screening on live television allowed the witnessing of events as they happened, a technique that resembles the shift of modern stage magic away from drama to events experienced in the present. More recently, STS researchers have examined demonstrations of computer-based technologies, including the development of medical imaging prototypes (Coopmans, 2011), product launches for radio-frequency identification (RFID) devices (Simakova, 2010), and the selling of business IT systems (Smith, 2009). These studies further document the carefully planned nature of such performances and some of their potentially deceptive forms, as with Coopmans’ observations of carefully choreographed revelations and concealments.

Fundamental differences remain, of course, between stage magic and technoscientific display. In another rare reference to conjuring in STS, Collins (1988) reminds us that while authorized displays of science and engineering are taken as evidence of reality, the public does not take conjuring to be real and that ‘illusionists, though they may be able to produce identical effects … are not licensed to convince’ (p. 729). Nevertheless, the claim here is that similarities exist across this divide in the way performances are crafted and conducted. The origins of these similarities are partly interpretive, given that stage magic has been viewed here through the lens of the human–machine reconfiguration which is grounded in recent technology productions. But equally, the similarities are likely to be the result of an extended mutual influence. Conjuring and public displays of science, from the enlightenment onwards at least, have both contributed to, and been influenced by, a common ‘theatre of marvels’ that includes circus, juggling and many other forms.

To explore the parallel in more detail, recent displays of computerized life forms, such as robots and simple ‘intelligent’ functions like the iPhone’s Siri, provide a useful point of comparison. Of all technologies, they might be expected to bear a close resemblance to conjuring. Not only do they express the same aim to simulate an appearance of agency in matter, but more significantly they hold deeper historical connections. Stage magic in the 18th and 19th centuries, including the shows of Robert-Houdin and the Maskelyne family, had a close association with the construction and display of mechanical automata of lifelike figures that are now considered by many as part of the prehistory of computerized life forms (e.g. Riskin, 2007). ¹⁴ Automata are therefore an important reference point for the present comparison with significant cross-overs of style, technologies and methods with stage magic. Some conjuring tricks, like the Educated Fish, might now be seen as quasi-simulations of intelligence in a surprising context. Conversely, many automata involved trickery and relied on secret human inputs. For example, Jacques Vaucanson’s celebrated mechanical duck, displayed around 1738, could rise up, lie down, move its wings and tail and eat corn, but its simulation of digestion was faked through secretly pre-loading the machine with real faeces (Riskin, 2003). And Baron Von Kempelen’s famous chess-playing ‘Turk’, displayed from 1770 to 1838 and capable of beating human opponents, was actually operated by a human chess player cleverly concealed within its faked mechanical workings (Standage, 2002; Sussman, 1999). ¹⁵

Through these strong links with automata, stage magic might itself be seen as an element in the prehistory of computerized life forms. As Jessica Riskin (2007) carefully argues, in relation to studies of automata, this is not to assert a straightforward continuity from the mechanical to the digital, driven by ‘timeless compulsions, transcendent forces, or inevitable trajectories’ (p. 3). But rather, the intention is to trace both continuities and discontinuities between computerized life and stage magic, to deepen the understanding of both forms and to contribute to a larger picture of technological performance. The first part of this is to consider how displays of simulated living things exhibit the techniques of stage magic, particularly how overt simulation is complemented by covert techniques of dissimulation. The second part is to develop further the present technological picture of stage magic itself, by considering how it reflects elements from the world of simulated life forms, both computerized and mechanical.

First, then, how might displays of computerized life be seen as enacting a pattern of simulation and dissimulation in common with stage magic? Take modern conjuring’s shift towards minimal props and the empty-box form, exemplified by the wooden chest that changed weight, the free-standing container of the Sphinx Head and the transparent goldfish bowl and stand. The rendering absent of inner workings and back-stage machinery in these cases finds resemblance in the way robot performances often dissimulate their ‘affiliated technologies’ (Suchman, 2007: 246), those surrounding and connected components that are necessary to a production but not considered part of the display. Suchman (2007) describes, for example, how a vast scale of offboard electronics and connective cables are considered peripheral to Cog, MIT’s most famous social robot, and are not visible in widely seen demonstration videos. A more extensive illustration appears in David Dufty’s (2012) account of the ambitious project to build a life-sized replica of the cult science fiction writer Philip K. Dick, capable of actively engaging with humans through conversation and related movements of head, eyes and body. Eerily resembling Robert-Houdin’s drawing room stage set of 1840’s Paris, the talkative android was often presented at tradeshows sitting on a sofa in a room-size capsule that recreated the living room of a California Bungalow of the early 1970s. The simplicity and naturalness of both stage sets worked to dissimulate the presence of simulative machinery hidden in the furniture and built fabric.

Similarly, the dissimulation of the magician’s instrumental agency in working the trick by acting out a different persona, and likewise for assistants, is mirrored in the erasure of human efforts and intentions around robot displays through the downplaying of the vital work of technicians. This was true for the Philip K. Dick replica, which ostensibly performed alone but in practice depended on a demonstrator to manage interactions with spectators, wiping memory buffers that got overloaded and intervening when the android’s mental machinery was failing. Morana Alač et al. (2011) extend this observation in their study of social robots and preschool children, pointing to the influence of not only human presenters but also the physical setting in the production of robotic agency.

A further aspect of the common pattern across conjuring and computerized life forms, is the translation of elements that are a necessary part of a machinic production into apparently incidental or even accidental circumstances of the display. In stage magic, this has been seen in the subtle framing of apparatus, as with the casting of Robert-Houdin’s temporary walking board to reach the stage as an incidental spot to perform the Light and Heavy Chest, when it was actually the vital platform containing the hidden electromagnet. It was also evident in the assumed naturalness of the modern magician that disguised a reliance on rote patter and choreography. These techniques find resemblance in the way performances of intelligent machines sometimes blur the boundary between the scripted and the incidental, as suggested for Philip K. Dick in the excerpt above. Another example is seen in Suchman’s (2007) classic study of people using the intelligent help function of a photocopier, a simulated being of more modest and practical intent. Breakdowns in human–machine interaction often occurred where users appeared to assume that the machine’s instructions were tailored to the particularities of their unfolding situation, when in reality it was scripted to serve more generic situations as envisaged by a now distant designer.

The common pattern between these two worlds also extends to the prescribed manner of audience engagement, in the form of an invitation to be playfully deceived. In the case of computerized life forms, this stems from Alan Turing’s (1950) seminal test for what counts as an intelligent machine: one that can fool a human into believing that it, too, is human. It is often said of the various talk-bots that have attempted the Turing test, that they use tricks to produce their effects of being uncannily life-like and entertaining us for stretches of lively conversation. While most robot displays, like those of Philip K Dick, are not set up as formal Turing tests, they still carry an invitation for the watcher to decide how real the encounter feels and, implicitly perhaps, whether they or others might be deceived. This connection has also been noted in the case of automata displays. In tracing Charles Babbage’s fascination with automata, including the faked chess-playing Turk, Simon Schaffer (1996) pointed to the ‘tempting contemporary resonance’ with the Turing test and its shared reliance on ‘concealment and detection’ (pp. 77–78).

Related to this format of playful deception, there is another common element in audience experience. The modern style evoked a deeply contradictory experience for spectators who, while knowing that the magic was not real, were faced with a magician in ordinary clothes using simple props and speaking and acting in ways that seemed natural and of the moment. This mirrors some reports of the reception of robot simulations, such as Sherry Turkle’s encounter with the precocious Cog, above, in which she felt it to be alive while knowing it to be only a machine. The contradictory experience in both cases can be seen to be a product of the combination of simulation and dissimulation: creating an effect known by all to be contrived, while simultaneously erasing signs of its contrivance in machinery and method. Again, automata are thought to have elicited a similar contradiction for watchers ‘because they dramatized two contradictory claims at once: that living creatures were essentially machines and that living creatures were the antithesis of machines’ (Riskin, 2003: 612).

The productions of stage magic and recent computerized life forms, then, appear to share at least superficial similarities in the way various agencies within their assemblies are dissimulated. This might account in part for an ‘often heard’ comparison between social roboticists and magicians, as reported by Alač et al. (2011: 918). A significant question is whether the appearance of conjuring-like dissimulation in productions of computerized life is to be seen as deceptive, playful or otherwise. On this question, it is important to see that computerized life forms are often, in other ways, deliberately non-dissimulative. Many social robots, including Cog, have other-worldly appearances that openly declare themselves as machines, and those resembling human form often sport a fashionable hole in the body or head through which electronics can be glimpsed, as was the case with the Philip K. Dick android. This deliberately non-dissimulative strategy is held in common with automata of the Enlightenment onwards. Vaucanson’s duck, a leading example of this approach, broke with the earlier tradition of hiding mechanisms in a supporting plinth and instead carried all of its machinery onboard, made visible through transparent feathers (Riskin, 2003).

Computerized life forms therefore present a mixture of the strategies of automaton display and stage magic, openly presenting themselves as technology but hiding away their affiliated components and assistive human labours. Under this mixed strategy, it is clear that the intention of technologists in using dissimulations is not always to deny the presence of technology but often to construct a particular witnessing of their current design and approach. The dissimulations around Cog and Philip K. Dick, for example, might be seen as clarifying techniques that allow public spectators to see the significant possibilities of these technologies without being distracted by mundane practicalities. Equally, however, where these dissimulations accumulate, and render invisible in the articulation, work around public displays, they hold the potential to deceive watchers including the technologists themselves, about the nature and viability of the technological achievement. The partial and mysterious glimpses into internal workings may constitute only an apparent transparency that reinforces a larger dissimulation.

The second part of comparing conjuring with computerized life forms is to see how it might enhance the present picture of stage magic as a technological accomplishment. Reflecting the discussion back onto conjuring, we see that, strangely, on the magic stage the human performer became a kind of automaton, executing an exactingly choreographed and scripted version of itself. The conjuror Edwin Sachs (1885) wrote of his technique in 1885: ‘Each hand and foot had its proper position at a given moment’ (p. 153). And in a revealing line, Robert-Houdin pointed out the dangers of letting audiences glimpse through the cracks of performance:

Some artists commit, when performing, a fault which cannot be too carefully avoided: They lay aside their animated and genial expression the moment the trick is over, as if they were mere smiling machines, set in motion and stopped at the touch of a spring. (Robert-Houdin, 2006 [1868]: 31)

The problem was not that they had performed exactingly to script but that they had given themselves away.

In resembling programmed automata, modern magicians might further be seen as having entered themselves and their audiences into a strange variant of the Turing test. To succeed, the magician-automaton sought to pass off its behaviour as natural and occurring in the moment and to conceal its origins in a carefully pre-programmed script. Any sense of machinic repetition was the death of the performance – hence the rules never to repeat tricks on the same occasion and never to say in advance what was intended to take place. This mirrors conventional Turing tests in which, as noted by Collins (1990), an experience of encountering an intelligent being is broken if the machine noticeably and exactly repeats an extended sequence of behaviour.

The terms set by the Turing test are for the humble computer to do its best before privileged human judges. In conjuring shows of the modern style, a more deceptive machine took to the stage, one comprising an arrangement of visible and hidden components and scripted operators. For the audience, the status of this machine was obscure: its inert props seemed to lack any inner workings, and its main operator seemed to be simply a presenter behaving naturally in the moment. Unwittingly, spectators became, in Steve Woolgar’s (1991) term, ‘configured’ into this machine (p. 61) – its transparency and minimalism creating a false sense of them being in control of their own comprehension of events. Unlike contenders in a conventional Turing test, the strange machines of stage magic swiftly confronted the audience with inexplicable impossibilities. The human watchers were no longer arbiters of what counts as intelligence, but instead their own capacity for sense-making was brought into question.

What conjuring reveals about agency: everywhere and nowhere

What has been revealed about translations of agency by peering inside stage magic’s peculiar human–machine reconfigurations? To consider this, a particular card trick, ‘Everywhere and Nowhere’, provides a useful metaphor. It was invented around the middle of the 19th century by Johann Hofzinser, an Austrian magician who through a minimalism and elegance can also be considered to be part of the modern style. In this trick, a chosen card is seen to be at many different locations, but each time it disappears. By analogy, if we try to follow agency through the performance of a magic performance, at each moment it shifts between presence and absence, between being real and being imagined. ¹⁶

First consider the trick’s magical ending, the apparent production of a supernatural effect. It is known by all to be a fiction. Yet strangely, as emphasized by the modern style, it depends on being grounded in a material reality of witnessed events – events that are ‘only interesting from the fact that they occur’. So begins a pattern of agency being somehow both real and unreal. Modern spectators, accepting that there can be no supernatural forces at work, wonder where the agency comes from that secretly works the trick. Look next at the magician. The agency of the performer also pales under closer inspection. During performance, it is obliterated by a fabricated stage persona designed to appear as spontaneous but in reality reproducing a rote script of action and patter. Similarly, the agency of assistants, which lies directly behind the magical effects like the Light and Heavy Chest, the Sphinx Head and the Educated Fish, is also enlisted in the machine of production, following strict instructions to act on cues from the magician.

Look next at the apparatus. It has been seen how it was designed to appear in the form of an empty box: inert props that neither housed a mechanism nor existed as a component of a larger machinic arrangement. The agencies of the designers and makers of the apparatus were thereby denied on stage. But what of the preparations for the show? Look next at the inventors and builders of the apparatus, and the producers of the show, including the off-stage magician. Surely the agency behind the supernatural effect resided in these back-stage designs and arrangements. Here too, though, it fades on closer inspection. In the history of conjuring there are but few essential magic tricks, performed in many stylistic variations: ‘The difficulty of producing a new magical effect’, wrote Nevil Maskelyne, ‘is about equivalent to that of inventing a new proposition in Euclid’ (Maskelyne and Devant, 1992 [1911]: 43). Graham Jones (2008) has recently described the great extent to which even Robert-Houdin deceived others and possibly himself about his inventive role in creating new tricks that were plainly copied from other performers. Drawing a parallel with Latour’s (2005) points about constructivist views of science, the view of conjuring tricks as simulating a form of agency does not imply that magical effects were created as a matter of whim, or even as a direct result of the deliberative intentions of the performer. Rather, the productions of stage magic emerged through an accumulated multitude of intentions, of very many performers, inventors and apparatus builders over an extended period, carried forward through gradually evolving apparatus and trick procedures, and shaped further by the receptions of successive audiences.