Return Home Contact Links Return Home
Technology and Patents Case Studies Press and Publications About Ross-Hime Designs, Inc.
Press and Publications


Leonardo's Programmable Automaton

A Reconstruction

(Codex Atlanticus, f. 812 r, ex 296 v-a)

By Mark Elling Rosheim
© 2000, Ross-Hime Designs, Inc.


Rendering of Leonardo's Programmable Automaton

Click the image for a larger view

See Video

The "Lettura Vinciana," traditionally, should have a Leonardo text as the subject for a critical analysis or interpretation. However, drawing for Leonardo is superior to words in communicating an idea. It is enough to quote his famous statement on a sheet of anatomical studies of 1513 next to a large drawing of the heart shown in its context of arteries and veins:

With what words, O Writer, will you describe with like perfection the entire configuration which the drawings here does?…1

Therefore it is fully justified that a "Lettura" be applied to a drawing such as the one in the Codex Atlanticus2, f. 812 r [296 v-a], which shows the formation of a technological idea as early as c. 1478, when Leonardo was about twenty-six years old. This drawing, where not a single word appears, can now be "read," as I intend to do, as representing a programmable automaton. As such, it was to have a great impact on Leonardo's future views on biomechanics, all the way to his latest achievements with the famous and still enigmatic mechanical lion of about forty years later.

This paper builds on the fundamental discovery of Carlo Pedretti, who in 1975 was the first to recognize "Leonardo's so-called automobile" in the Codex Atlanticus for what it is-an automaton.3 Featuring nothing less than front wheel drive and rack-and-pinion control! He has gracisouly resumed his research to edit this "Lettura" for which I am indebted.

Leonardo developed a means of locomotion and control for robotic devices, called automata, which would later become common in the festivals and court masques of the sixteenth and seventeenth centuries. In 1478, while under the patronage of the Medici, he designed a programmable, mechanical computer-controlled automaton. This automaton was a precursor to mobile robots and was perhaps the earliest "computer" in western civilization. By reinterpreting material from Leonardo's notebooks, as well as the works of a Japanese artisan of the eighteenth century, I have been able to reconstruct Leonardo's intentions for the programmable automaton. I also examined the work of Giovanni Alfonso Borelli and how they may help in interpreting Leonardo's concepts from the first of the Madrid manuscripts. Finally, anonymous copies of his technological drawings are discussed.

This basic design was perhaps dusted off some thirty-six years later, in 1515 France for use as a platform for a self-propelled mechanical lion, again under a commission from the Medici. Lomazzo's 1584 retrospective account of the allegory of friendship between the Medici and Francis I at the latter's accession to the throne of France tells us that "once in front of Francis I, King of France, [Leonardo] caused a lion, constructed with marvelous artifice, to walk from its place in a room and then stop, opening its breast which was full of lilies and different flowers." The triumphal entry in 1515 of Francis I into Lyons was the occasion. The Lion was an old symbol of Florence, and the lilies referred to the fleur-de-lis that Louis XI of France had given to Florence as a token of friendship. Thus the two were combined symbolically in a dramatic presentation by one fantastic machine.

All this has been ascertained only recently by Carlo Pedretti. Michelangelo Buonarroti the Younger, nephew of Leonardo's great rival, in his description of the wedding of Maria de' Medici, Queen of France and Navarre, a booklet published in Florence in 1600, mentions an automaton which was presented at a banquet, a mechanical lion that walked a few steps and then rose on its hind quarters, opening its breast to show that it was full of fleur-de-lis, a concept, conclude the younger Michelangelo, "similar to that which Leonardo da Vinci realized for the Florentine Nation on the occasion of Francis I's entry into Lyons."4

The lion for Lyons was preceded in Leonardo's works by the early version discussed here, his robot knight and his reconstruction of ancient automata. The first was a continuation of a tradition of robot design in the Verrocchio bottega. As Vasari relates, Verrocchio designed an automaton clock for the New Market place in Florence, "a device that was then much admired as a beautiful and whimsical thing."5 One can draw the conclusion that Leonardo was following in his master's footsteps and perhaps attempting to exceed his abilities.

In his first robot design of about 1478, the programmable automaton reconstructed here took the form of a wheeled tripod. Leonardo seemed to have been making a literal interpretation of a stanza from the Iliad, Book 18, which apparently describes an entire flock of robots in the house of Hephaistos:

…since he was working on twenty tripods which were to stand against the wall of his strong-founded dwelling. And he had set golden wheels underneath the base of each one so that of their own motion they could wheel into the immortal gathering, and return to his house: a wonder to look at. 6

Evidence can be found in Leonardo's drawings that he may have been striving toward a literal reconstruction of the automaton described in this stanza. Tripod support is evident in Leonardo's three wheels: perhaps the "golden wheels" of Homer. The phrase, "of their own motion," implies autonomous power and control. Similarly, the phrase "wheel into the immortal gathering, and return to his house dining room, alone . . ." describes the unaided travel of a device with a rudimentary, programmed navigational ability. All of the behavior of the machines from the Iliad is consistent with Lomazzo's description of Leonardo's lion.

Leonardo returned to the theme of ancient automata with his water powered bell-ringer, c 1508, Windsor, RL 12716 & 12688 (Fig. 1). This hydraulic automaton stands on a cylindrical base containing twelve interconnected chambers which, by filling in a sequence, actuate float valves.7 The figure swings its arm and trunk to strike the bell with a mace. The bell-ringer's position and pose are as if it had just walked up to the bell to strike it. The posture recalls Pindar's seventh Olympic Ode (c. 520 B.C.), which describes the animated images of Rhodes and Crete:

And their ways teemed with sculptured forms like to beings
that live and walk:
And high waxed their renown.
But in the man who has the craft-lore even the greater
skill putteth itself forth without the guile of magic.8

From this brief survey, we can see that automata of various descriptions were very much on Leonardo's mind up to and during the design of the programmable automaton. However, the Codex Atlanticus and Uffizi drawings discussed below do not show Leonardo's complete intentions for the automaton. These intentions had to be reconstructed by reinterpreting other materials from Leonardo's oeuvre.

Armed with the critical guidance of Carlo Pedretti that I was facing a automaton and not an unworkable anticipation of the modern automobile, I looked to robotic designs of subsequent centuries and diverse cultures to form my design and operation hypothesis of Leonardo's automaton. I then reinterpreted the fragments based on this hypothesis. My guess was that traces of the programmable automaton would be found in later designs and accounts of ancient automata. This information would aid in interpreting and reconstructing the nuts and bolts of Leonardo's design as well as its performance characteristics and application. At the same time, I drew upon my lifelong experience as a student of Leonardo's legacy.

As I will show, several seemingly unrelated pieces of evidence, when reinterpreted, reveal the intended working of Leonardo's programmable automaton. These are, first, an eighteenth-century Japanese tea-carrying robot, which betrays Spanish origins. To interpret Leonardo's research in the first of the Madrid manuscripts we will examine the work of the seventeenth century author of De motu animalium, Giovanni Alfonso Borelli (1608-1679). A final piece of the puzzle comes from an early technological sixteenth-century copy of Leonardo's drawings.

I    Eighteenth-Century Japanese Tea-Carrying Automaton

To further understand the design and operation of Leonardo's programmable automaton, I looked at the late eighteenth-century Japanese "Karakuri," or tea carrier.9 In this design, separated by several centuries and two continents from Leonardo, I found an autonomous mobile robot of equal complexity, construction, materials and performance characteristics as those of the lion that delivered flowers to Francis I.

The fourteen-inches tall (35.56 cm) Tea Carrier (Fig. 2) has a cherry wood frame, and composite gears of laminated oak and Japanese cedar for strength held together by wooden pins. The only metal components are in the governor which regulates a coiled baleen (whale-bone) spring and is steered by means of programmable cams mounted on the spring's hub. A cam follower actuates the front steering wheel.

Cams are revolving machines such as a wheel so shaped to impart a preprogrammed motion in another piece engaging it. By changing the shape of the cam one changes the "program," the motion that is imparted to another piece engaging it. Seeing programmable cams in the Japanese design, I reexamined Leonardo's programmable automaton. Indeed, these do exist and will be discussed in the reconstruction.

The autonomous robot designs would also reappear in Spain in the work of the Juanelo Torrianus, a master watchmaker in the service of Charles V. Such designs would become increasingly common in the eighteenth century, with the development of water-powered grottos and spring-driven dining table servers. Such an automaton, built in Japan in the late eighteenth century, constitutes the second piece in the puzzle of Leonardo's programmable automaton. Most significantly, the Karakuri, or tea-carrying automaton, was not likely native Japanese design. Rather, its origins can be traced to Spain, where Madrid I and II were kept since their transfer to the royal library in 1642. The missing section of the first manuscript could well have disseminated Leonardo's ideas otherwise unknown to us today. The design for the tea-carrier may have been transferred to Japan perhaps by Spanish Jesuit missionaries during the eighteenth century, where it was reinterpreted as a geisha.

II    Platform for Automata in the Madrid Manuscripts

Although there are now no longer any complete designs for automata in Madrid MS I, that manuscript contains numerous examples of potential components, such as compact spring drives with imaginative integral fusee mechanisms that could have powered them. Apparently designed for clocks,10 they could well have been employed in mobile automaton applications where weight and size need to be optimal. They consist of a drum containing the spring often referred to as the going barrel or tambour. The fusee mounted on the top rotates and automatically compensates for the loss of torque as the spring unwinds. Various means are used to transfer the power from the moving fusee to a stationary rotating power output shaft. These could have been used as automaton power sources for successors of the programmable automaton.

Folio 4 recto of Madrid MS I shows the first of a series of ingenious, lightweight, compact fusee-regulated power supplies (Fig. 3). Spiral arrays of driver pins connected to the barrel form the fusee. A sliding lantern gear transfers the barrel's rotation to the output. Greater amounts of torque may be transmitted using these gears over conventional cables and chains. Madrid MS I, f.16 r (Fig. 4) uses a conical lantern gear to compensate for the difference between the pinion path and the fusee pins. The often reproduced design on Madrid MS I, f. 45 r (Fig. 5), if laid on its side, could constitute a drive element for an automaton. The disk on the output appears to be a drive wheel. All of these designs feature a significant gear ratio increase which would be useful for propulsion.

Several theoretical studies comparing the various spring morphologies for use as a power source are shown in Madrid MS I, f. 85 r (Fig. 6). A "T" shaped key is shown for winding. It is possible that a key like this was used to wind the programmable automaton. The depth of Leonardo's interest in springs is indicated by the fact that he even shows manufacturing techniques illustrating automated drawing machines for strips of spring steel.

Numerous gear studies ranging from low-cost sheet metal gears to elaborate helical and involute gear forms are shown. Theoretical studies of gear geometry are also present. Cams in a myriad of forms are shown beginning on Madrid MS I, f. 1 v (Fig. 7) and 24 r (Fig. 8). Leonardo engages the curved surfaces on the inside, outside and mixes gears with cams to produce hybrids.

The practical demonstrations of sophisticated automata with the exception of the astronomical clock are missing. It is my opinion that these were once part of the eight folios that were removed at some point in time before the Madrid codices were published. As shown in the Reti edition, they are ff. 37-42 and 55-56.11

III    Borelli's "De motu animalium"

Giovanni Alfonso Borelli was an important mathematician of his time who became interested, as Leonardo before him, in modeling the movements of animals. Born in Naples on 28 January 1608, he was the son of a Spanish infantryman and his Italian wife. In 1635, through Castelli's recommendation, Borelli obtained the public lectureship in mathematics in Messina, Sicily, then ruled by Spain. In 1658 he accepted the chair of mathematics at Pisa. Malpighi is credited with sparking Borelli's interest in the movements of living creatures. Around 1675, Borelli created De motu animalium in hopes of being received and accepted into the Academie Royal de Science newly established by Louis XIV in Paris.12

The similarity between Borelli and Leonardo has not gone unnoticed by modern researches. V.P. Zubov, makes the general conclusion that Borelli over simplifies.13 This may be true in some cases but in many other cases he is now viewed as being highly advanced for his time, anticipating technology that did not occur until well into the twentieth century.14 Zubov also criticizes either the absence or incorrectness of Leonardo theories of flight. And yet, as shown by Madrid MS I Leonardo's theories were adequate to design in 1495 a hang-glider capable of lifting a man. This has been proven through modern reconstructions15

Only a very few of the scholars who have dealt with Leonardo's scientific and technological endeavors have called attention on Borelli's work for possible reflections of Leonardo's ideas. None have ever raised the question whether such ideas could have reached Borelli directly or through the mediation of those who had access to Leonardo's manuscripts.

As is well known, those manuscripts were brought back to Italy from France by Francesco Melzi at Leonardo's death in 1519. Other books, of course, could have already circulated when Leonardo was still alive, and there is evidence that an autograph manuscript on the movements of the human body analyzed geometrically -- precisely a subject extensively treated by Borelli -- was seen by Federico Zuccaro either in Rome or in Turin at the end of the sixteenth century.16 And since the mechanical lion for Francis I is now ascertained to have been a Medici commission of which Leonardo worked either in Florence or in Rome about 1515, he very well may have gathered all the pertinent information in a book left to his patrons and then lost, as was the case with that of unspecified contents given to Messer Battista dell, Aquila, steward-in-waiting to the Pope -"cameriere segreto del Papa."17

Leonardo's invention of diver devices and his studies for the submarine were shown by Mario Baratta in 1903 as part of a vast historical context that includes Borelli's comparable studies.18

Borelli's studies on the flight of birds were first mentioned by Gustavo Uzielli in 1884 in connection with Leonardo's Codex on the Flight of Birds, in particular for a note on f. 7 r (and again on f. 10 r), in which Leonardo, long before Borelli, formulates the theory that the wind acts as a wedge in lifting the bird, "il vento fa ofitio di cuneo."19 Scholars such as Giuseppe Boffito in 1919 and Raffaele Giacomelli in 1935 systematically approached Leonardo's and Borelli's studies on the flight of birds on a comparative basis, and this is also the subject of a paper published by G. Pezzi in 1972.20

It is indeed surprising that recent scholars such as Martin Kemp (1982), Kenneth Keele (1983) and Kim H. Veltman (1986), who have greatly contributed to place Leonardo's scientific studies in their proper historical context, make no reference to Borelli. And yet in the Keele-Pedretti edition of the Corpus of Leonardo's Anatomical Studies at Windsor (1980), Keele points out21 that the complex mechanism of the wood-pecker's tongue, on which Leonardo intended to write22 is first explained by Ulisse Aldrovandi and again by Lorenzo Bellini and by Borelli, who gives a beautiful illustration of it in plate V, fig. 11 (Fig. 9).23 Finally, Leonardo's extensive and innovative studies on mechanics are shown by Roberto Marcolongo in 1939 to find reflections only in Borelli's work of nearly two centuries later.24 Unfortunately, this interesting pointer was not followed-up by Arturo Uccelli (1940) with his monumental edition of those studies.25

Of the eighteen plates in De motu animalium, almost every one offers a figure that is similar in theme or style to Leonardo's. This is the case not only with drawings; coincidentally, several reference letters match. For example, the Borelli reference letters on plate III, fig. 2, H, C, and A (Fig. 10) match those of Codex Urbinas, f. 120 v (Fig. 11). Interestingly, Borelli appears to follow the sequence of illustrations in Madrid MS I: he starts with a figure that has elements in common with the page before the first section of missing pages in Madrid MS I. On Madrid MS I, f. 36 r (Fig. 12), Leonardo shows a pulley design in which the number of pulleys is the same as the number of muscle fibers in Borelli's first illustration on plate I fig. 1 (Fig. 13). Also plate I, fig. 3 (Fig. 14), has the same number of muscle fibers as cable convolutions in Madrid MS I, f. 36 r.

Understanding of Leonardo's biomechanics may be enhanced through the comparative study of Borelli. Because of the great similarity of theme and organization, Borelli may be used as an aid in interpreting drawings such as CA, f. 966 r [349 r-b] (Fig. 15) which represent the human body schematically and are similar to the diagrams in Borelli's plate V, fig. 1 (Fig. 16). Madrid MS I, f. 90 r-v (Fig. 17) bears a striking resemblance to Borelli's plate V, fig. 6 (Fig.18). This relates to the legs lifting capacity in retraction. Also in Borelli's plate XII, fig. 4 (Fig. 19), we see a graphic similarity to Leonardo's MS L, f. 28 v (Fig. 20), and CA, f. 444 r [164 r-a], (Fig. 21) showing a centerline though the body's limbs. The above studies lead directly to Leonardo's practical demonstration piece, the fabulous robot knight of about 1495. Indeed, Borelli's plate V, fig. 8, could very well represent Leonardo's intention for the robot knight's leg (Fig. 22).

IV    Sixteenth-Century Copy of Leonardo's Drawings

Another piece of the puzzle came from part of an early technological sixteenth century copy of Leonardo drawings, Uffizi, GDS, no. 4085 A r, of a large sheet of technological drawings including manufacturing and weapon designs (Fig. 23).26 The top left and second to the bottom far left clearly show a similarity to the plan view of CA, f. 812 [296 v-a] i.e. a mechanism having two large gears and arbalest springs. Although the form is different--a rocker arm acting on the two separate gears to form an escapement--the basic concept is the same. A rocker arm held in tension by cables connected to the arbalest springs, creates an escapement to regulate the speed of the gears oscillating back and forth like a teeter-totter. Two wheels located on the bottom of the machine could be for locomotion although it is difficult to interpret them at this time.

It is this element, the escapement acting on the barrels that confirmed what I had already suspected -- that the arbalest springs engaging the cogs comprised an escapement mechanism. Equipped with this last piece of the intellectual puzzle, I was able to begin the reconstruction.

V    Reconstruction

Since 1929, when it was first recognized by Guido Semenza as a self-propelled vehicle, the machine represented in CA, f. 812 r [296 v-a], has had all interpreters convinced that the arbalest springs were the source of motive power.27 This way of reading Leonardo's technological graphic notation involved, of course, arbitrary alterations and modifications of his design, primarily the addition of cables transmitting power from the arbalest springs to the barrel gears. Scholars such as Giovanni Canestrini, Arturo Uccelli and Jotti da Badia Polesine argued and even quarreled over irrelevant points of details, all concurring, however, in praising Leonardo's ingenuity as the inventor of what they considered to be the first example of a differential gear.28 It did not matter, therefore, whether the machine, as reconstructed by them for the Leonardo exhibition of 1939, would work or not. Like Leonardo's helicopter, his so-called "automobile" was just another idea for modern interpreters to integrate and develop. So did Canestrini as he prepared a set of blue-prints for the model presented in the 1939 exhibition. Inexplicably, he concluded that what he had so painstakingly reconstructed could never work:

These schemes [i.e. blue-prints] which have provided the basis for our reconstruction of Leonardo's car, along with the hypotheses that one may speculate about, confirm our conviction that this vehicle could never function the way it was designed by Leonardo.29

The programmable automaton as depicted in the Codex Atlanticus consists of a square mortised and dovetailed wooden frame approximately 20 by 20 inches (50.8 cm by 50.8 cm). These dimensions are based on several measurements taken from the Codex Atlanticus drawings which are assumed to be full-size preliminary fabrication drawings.30 This is because in Leonardo's time, standardized measurements small enough for mechanisms did not exist. Producing full sized drawings provided a simple, accurate and direct guide for the craftsman. It should be noted that the dimensions of the automaton provide a perfect stage on which to mount a figure such as a life-sized lion seated in an upright posture but would seem small for a lion walking on all fours, as would have been necessary for the revelation of the flowers at Francis I's entry into Lyons in 1515.

Leonardo went to pains to secure the joints with fasteners, no doubt for protection from vibration. A second, lower frame is attached to the top via angled brackets as shown in CA, f. 812 r [296 v-a] (Fig. 24). The frames contain two interdependent subsystems for propulsion and guidance. Using the top view figure as a guide I believe the right side is for steering. This is indicated by the right side unit having several cams drawn on the barrel gear. The unit on the left is for propulsion. This is indicated by the left-hand unit cam follower return spring having a screw which may be tightened with a nut, as shown in CA, f. 878 v [320 v-a] (Fig. 25). Controlling the tension of the cam followers is the means of controlling their speed. This is also supported by Leonardo selection of the left side of the automaton to be illustrated in his perspective sketch in CA, f. 812 r [296 v-a] (Fig. 24) to show a rough idea for propulsion.

Each escapement arbalest consists of two springs that merge to engage the corner cog CA, f. 812 r [296 v-a] (Fig. 24]. The two large barrel gears of the propulsion and steering systems mesh thus phasing the subsystems together. The interaction of these two subsystems forms a complete escapement that enables speed regulation of the large springs and hence the large gears. The gear ratio from the springs to the spoked drive wheels increases the springs revolutions per minute to maximize the travel of the programmable automaton.

Petal-like cams located on the top surface of the large gears drive the two scissors-like cam followers which are kept in contact with the cams via the arbalest return springs. These springs have traditionally been mistaken for a power source but they are actually return springs for the cam followers and probably would have been fashioned out of bent wood. This material is indicated because large amounts of force are not needed to keep the cam followers in contact with the cams. The scissorlike cam follower arms are roughly shown in perspective engaging the cam lobes for left power and right steering, left of the top view of CA, f. 812 r [296 v-a] (Fig. 24). A smaller sketch above this rough perspective sketch shows perhaps a partial top view of the scissors or geometrical study of same.

The top perspective sketch is only preliminary, and Leonardo seems to be exploring two steering options. The first shows a tiller connected to the wheel with the opposite end driven by the linkage. The second seems to be a wheel directly mounted underneath the platform, with the fork's shaft passing up into the automaton mechanism. The shaft connects to the upper cam follower arm.

The bottom three figures may relate to an additional control mechanism. In the center of the drawing directly above the main top view is a rack-and-pinion mechanism with the pinion axle terminating in a crank. This, I believe, is the pinion mounted in the left return spring shown in the bottom left hand corner of the folio. The spool and cable version above it is a simpler lower-cost alternative embodiment. The spool would clip-into the cam follower in-place of the gear. Based on the lower left figure this subassembly is mounted on the upper left return spring. This is the reverse location of the rack in the top view which is near the right return spring. Two functions could be accomplished with this arrangement. First, the automaton could be slowed as it encountered resistance of a dedicated cam. Secondly, upon slowing the rack-and-pinion mechanism could trigger a special effect such as opening a door.

Featuring front-wheel drive both spoked wheels are driven under power; this interpretation is based on the escapement arbalest springs bias in relationship to the corner cogs caused by the gear hubs creating a ratchet i.e. a single direction gear-train. Through Carlo Pedretti's insight that the leaf springs could not produce enough revolutions per minute to make the platform move a useful distance, and by using the Japanese Tea Carrier as a model, I am now convinced that Pedretti was right in suspecting that there are large coil springs located below the large gears.31 I suggest that a group of slight arcs shown within a figure on the bottom center of CA, f. 878 r [320 r-a] (Fig. 26), may represent just such a spring. Winding of the automaton may have been via one of the corner cogs, taking advantage of the gear reduction. The force generated by these springs would be substantial, perhaps in the fractional horsepower range.

Additional folio's show alternative possibly earlier embodiments. A second level of gearing of the going barrel gear is shown on the lower right of CA, f. 878 v [320 v-a] (Fig. 25). In addition a parallelogram linkage smaller preliminary sketches at the center left. This folio also provides an important reconstruction detail i.e. the right corner cog and its smaller gear meshing with the barrel gear.

Similar to the above a second level of gearing for the going barrel is shown on the lower right of CA, f. 878 r [320 r-a]. A parallelogram linkage is also sketched on the center of same. Near the cut-out section of Uffizi, GDS, no. 446 E-r (Fig. 27), shows another parallelogram mechanism similar to the above. These linkages may be indexing concepts. Finally, yet another second level of gearing is also shown on CA, f. 926 r [339 r-a] (Fig. 28) and CA, f. 956 r [347 r-b] (Fig. 29).

Codex Atlanticus folio 956 v [347 r-b] (Fig. 30) is interesting as it shows elements in other drawings such as the multi-grooved sinusoidal cam mechanism shown in "A Draped Figure and Studies of Machinery," 447 E Uffizi r (Fig. 31). See also the verso of the same Uffizi sheet (Fig.32) for very early fragmentary studies of the automaton. In Ashmolean Museum P. 18 (VII) (Fig.33) a small square sketch at the top center may be related to the triggering mechanism shown at the bottom of f. 812 r, [296v-a].32

Although the above may simply represent early iterations they may also represent additional or alternative drivetrains. Perhaps rotating vertical shafts went up to power doors or other devices for various theatrical effects, such as rearing of the lion on its hindquarters and upon opening its breast, revealing its heart full of Florentine lilies.

VI Operation: Propulsion and Control

The coil springs mounted within the going barrels are regulated by a unique verge-and-foliot clocklike mechanism (Fig. 34). This is achieved through interaction of the corner cogs and escapement arbalests. The escapement arbalests are equivalent to the verge albeit split, the pallets of the verge being the arbalest tips. Both corner cogs equate to the crown wheel although this function is also split between the two. Drive wheels and their respective drivetrains equate to the foliot or mass of the system. Dampening of the system is aided by the springingness of the escapement arbalests (Fig. 35).

In operation the corner cogs rotate towards their arbalest springs phased one-half step apart from each other. As the systems oscillates one would hear a "click click click" as the corner cogs incrementally release the escapement arbalests. The automaton is wound by a corner cog in a clockwise motion overcoming the greater resistant of the biased escapement arbalests. Perhaps generating a loud "clack clack clack" as it is wound.

The impetus for the system comes from the two springs housed within their barrels (Fig. 36). They are directly connected to the large gears which drive the smaller gears attached to the corner cog arbors. These arbors drive the lantern gears and thus the spoked wheels for forward motion.

Direction and velocity of the programmable automaton is controlled by the array of cams attached to the top of the large barrel gears (Fig. 37-38). As they rotate they force the scissors-like cam followers to pivot. The cam followers are forced against the cams by the arbalest return springs via the criss-crossed cables. The steering cam follower connects to the clevis and its steering wheel below the pivot.

The propulsion cams on the left large gear control speed, perhaps even momentarily stopping the automaton. Its basic operation is similar to the stackfreed used to regulate the earliest spring clocks. In the stackfreed a cam attached to the top of the barrel has a spring loaded arm pressing against it -the resistance matching the torque as the spring unwinds. Perhaps a few petal shaped cams were used to slow or momentarily stop the automaton. This cam follower could also drive the rack-and-pinion subassembly. As the return spring pivots the pinion moves up-and-down the rack perhaps triggering a mechanism for ancillary devices such as a door.

Starting from a home position the automaton begins to move forward and follows a preprogrammed course stopping and starting, turning left and right. Special effects at any point could be introduced. This program is determined by the number, shape and location of cams on top of the two large cam gears. Each cam representing an individual instruction or "line of code." The "program" is latent in these cams and is predetermined by the programmer who might "debug" or modify his "program" with a file!


Leonardo's programmable automaton is the first record of a programmable analog computer in the history of civilization. We see in it Leonardo's first design effort in planning automata culminating in his fabulous robot knight33, of about 1495, a practical demonstration piece based on his pioneering study of biomechanics (Fig. 39-44). Leonardo's sophisticated use of mechanisms at a very early age further highlights his talent. His ability to produce a design that simultaneously combines marvelous compact packaging, complex guidance and control can only bring to mind his own words:

When you wish to produce a result by means of an instrument do not allow yourself to complicate it by introducing many subsidiary parts but follow the briefest way possible, and do not act as those do who when they do not know how to express a thing in its own proper vocabulary proceed by a method of circumlocution and with great prolixity and confusion.34

The enormous losses Leonardo's legacy suffered, of which Madrid MS I is but one example, has given Leonardo the undeserved reputation as brilliant but disorganized compiler. The correct reconstruction of this work will continue to demand expert knowledge in several and widely diverse fields.

The seed of Leonardo's biomechanics was to come to full fruition nearly two centuries later with Giovanni Borelli's De motu animalium. Zubov appropriately refers to Leonardo as Borelli's "spiritual father."35 Curiously, Pierre Duhem, a fervent advocate of the theory that much of Leonardo's legacy was indeed available to, and taken advantage of by his successors, never mentions Borelli.36 That Borelli learned from Leonardo may never be proved. But it would seem too much of a coincidence that his approach to biomechanics should be so strikingly similar to Leonardo's. One may once more ponder over the testimony of Giovan Paolo Lomazzo who in the later part of the sixteenth century was to record that Leonardo had found the way "to have lions move by the power of wheels" (andar I leoni per forza di ruote), showing the impact of his technological ideas on his successors through the dissemination of his manuscripts in every part of the world.

Above all writers," concludes Lomazzo, "great praise deserves Leonardo da Vinci, who taught the anatomy of the human body and of the horse, which I have seen in the possession of Francesco Melzi designed divinely by his hand. He also illustrated every aspect of the human proportions, wrote of perspective and of light and shade, and of a way of representing figures larger than life, thus compiling many other books to show how many movements and effects fall under mathematical principles, while considering easy ways of moving and lifting weights. All this in books of which the whole of Europe is full, and which are held in great esteem by the experts, as they judge that nothing more could be done than what Leonardo did.37


1 Windsor, RL 19071 r (C.11.1 r), c. 1513. Further down, in the same sheet, Leonardo writes: "How in words can you describe this heart without filling a whole book?…". For the italian text see Scritti scelti di Leonardo in Vinci. A cura di Anna Maria Brizio, Turin, 1952, p. 508: "O scrittore, con quali lettere scriverai tu con tal perfezione la intera figurazione, qual fa qui il disegno?… Con quali lettere descriverai questo core che tu non empia un libro?…"

2 D'ora in poi citato anche come CA.

3 Carlo Pedretti, "Eccetera: perche' la minestra si fredda" (Codice Arundel, fol. 245 recto). XV Lettura Vinciana…, Firenze, 1975, p. 13, note 9, where the so-called "automobile" represented in CA, f. 812 r [296 v-a], is considered more likely "a cart for festivals moved by springs and planned to cover brief tracts as from one side of a piazza to another" ("non era altro che un carro per feste azionato da molle e destinato a percorrere brevi tragitti come da un punto a un altro di una piazza." In his Leonardo architetto (Milan, 1973), pp. 32-22, the interpretation is further elaborated and the Leonardo drawings shown in relation to other early studies in the Codex Atlanticus at the Uffizi.

4 Nothing is left of Leonardo's project for the mechanical lion and what used to be known of it was only through the mentions by Vasari and Lomazzo, who did not indicate the precise occasion of the event nor its symbolism. The missing information is supplied by the Descrizione delle felicissime nozze della Cristianissima Maesta' di Madama Maria Medici Regina di Francia e di Navarra by Michelangelo Buonarroti the Younger (Florence, 1600), p. 10, as first discussed and reproduced by Pedretti, Leonardo architetto, cit., p. 322. See also, by the same author, "Leonardo at Lyon," in Raccolta Vinciana, XIX, 1962, pp. 267-72, and Leonardo. A Study in Chronology and Style, London, 1973 (New York, 1982). p. 172.

5 Cf. C.P., "Leonardo's Robot," in Achademia Leonardi Vinci, X, 1997, pp. 273-4, quoting Vasari, III. 375, as first noted by Simona Cremante, who suggests a relation with Leonardo's later project for a bell ringer (Windsor, RL 12716 & 12688): "E' anco di mano del medesimo [Verrocchio] il putto dell' oriuolo di Mercato Nuovo, che ha le braccia schiodate in modo che, alzandole, suona l'ore con un martello che tiene in mano: il che fu tenuto in que' tempi cosa molto bella e capricciosa."

6 Homer, Iliad, XVIII. 372-376. Cf. Iliad of Homer. Translated with an introduction by Richmond Lattimore. Chicago, 1950 pp. 385.

7 In addition to a fragmentary sheet at Windsor reconstructed by Carlo Pedretti with fragments W. 12716 & 12688, the bell ringer project has survived in several sheets scattered throughout the Codex Atlanticus, e.g. f. 65 v [20 v-b], which in turn is the parent sheet of Windsor fragments 12480 and 12718. Cf. Leonardo da Vinci, Fragments at Windsor Castle from the Codex Atlanticus. Edited by Carlo Pedretti, London, 1957, pp. 38 and 41, and pl. 2. See also my paper "Leonardo's Lost Robot," in Achademia Leonardi Vinci, IX, 1996, pp. 99-110, in particular p. 100 and fig. 2.

8 Pindar, Olympic Ode, VII, 51-54. Cf. L. Farnell, The Works of Pindar, London, 1930, p. 35.

9 This is fully discussed and illustrated in my book Robot Evolution: The Development of Anthrobotics, New York 1994, pp. 27-29.

10 See The Unknown Leonardo. Edited by Ladislao Reti, New York, 1974, in particular the chapter by Silvio A. Bedini and Ladislao Reti, "Horology," pp. 240-63, with various kinds of springs discussed on pp. 250-3.

11 Leonardo da Vinci. The Madrid Codices. Volume III. Commentary by Ladislao Reti, New York, 1974, pp. 23-26. According to Carlo Pedretti, Leonardo architello, cit., p. 322, those missing sheets could well have contained "most accurate and spectacular studies for the robot."

12 The best account of Borelli's work, in particular his innovative studies on the flight of birds, is still the one in Giuseppe Boffito, Il volo in italia, Florence, 1921, pp. 137-42 (with full bibliography). For a comparable account, see Galileo Venturini, S.J., Da Icaro a Mongolfier, Rome, 1928, Parte Prima, pp. 243-5, which concludes with a reference to Leonardo: Nella pare che riguarda il volo, chi volesse fare un accurato confronto, troverebbe le stesse lines maestre, tracciate da Leonardo da Vinci: con questa differenza pero', che mentre Leonardo ci da' (ne' poteva essere altrimenti) un ingegnoso trattatello, dove non se sa se piu' ammirare la intuizione o la succosa brevita' del poderoso autodidatta, il Borelli, che ha potuto far tesoro delle osservazioni sagaci di tanti predecessori, e che in quella materia se sente appieno in casa sua, ci presenta un completo trattato scientifico". See also the preface to Paul Maquet's English edition of Borelli's De motu animalium (On the Movement of Animals), Berlin, 1989, pp. v-ix. Borelli was one of Galileo's most prominent followers, not only as a member of the celebrated Accademia dell Cimento in Florence and as a friend and a colleague of Evangelista Torricelli, but above all as a pupil of Benedetto Castelli, whose treatise Della misura dell'acque correnti (1628) was at one time believed to have been based in part of Leonardo's writings on the subject. Cf. Filippo Arredi, "Intorno al trattato 'Della misura dell 'acque correnti' di Benedetto Castelli", in Annali dei Lavori Pubblici, 1933, fasc. 2, pp. 1-24, and L'idraulica di Galileo e della sua scuola, Rome, 1942, in particular p. 16. Borelli's writings on hydraulics are included in the Raccolta d'autori italiani che trattano del moto delle acque, Bologna, 1822, Vol. III, pp. 289-336. One of his treatises, a report on the Pisa and Livorno swamps ("Stagno di Pisa"), is yet to be examined in connection with Leonardo's previous studies on the subject. Cf. Siro Taviani, Il moto umano in Lionardo da Vinci, Florence, 1942, pp. I-LXIV, in particular p. VI for the reference to Leonardo as having recognized before Borelli the general physiological laws of the muscular system.

13 V.P. Zubov. Leonardo da Vinci. Translated by David H. Kraus, Cambridge. Mass., 1968. Pp. 184-185. A comparable, modern assessment of Borelli's work comes from the author of a perceptive and well informed essay on "The mechanics of Leonardo da Vinci", Clifford A. Truesdell, in his Essays in the History of Mechanics, New York, 1968, pp. 324-25: "In the seventeenth century, statics was a well developed subject, and it was applied in a way then acceptable to many persons in many cases where any modern engineer would require laws of motion, then unknown, For example, we may cite Borelli's book, On the Motion of Animals (1685), where the parallelogram of forces seems to be the only quantitative basis for two volumes on the subject named, and where, despite the title, we look in vain for any laws of motion. I do not mean at all to ridicule the book; it is not only truly scientific but also ingenious in many places; I adduce it as an example to show the work both intelligent and extensive can be done on a wobbly foundation, and that the existence of serious literature in a domain, leading to some measure of success, does not necessarily imply that the structure is sound."

14 Paul Maquet, as cited in note 11 above.

15 See Michael Pidock, "The Hang Glider," in Achademia Leonardi Vinci, VI, 1993, pp. 222-25, with an editorial introductory note and reproductions, figs 1 and 2, of photographs of a first test flight of the reconstructed hang glider (Sussex Downs, England, 20 October 1993). A version more faithful to Leonardo's drawings has recently been built at Sigillo in Umbria by a local association of hang glider pilots. See Carlo Pedretti, Leonardo. The Machines, Florence, 1999, p. 29.

16 Federico Zuccari's account of the lost Leonardo manuscript of the Codex Huygens type of kinesiology studies is given in his Idea, Turin, 1607, p. 31, as fully discussed and reproduced in Leonardo da Vinci. Libro di pittura. Edizione in facsimile del Codice Urbinate lat. 1270 nella Biblioteca Apostolica Vaticana acura di Carlo Pedretti. Trascrizione critica di Carlo Vecce, Florence, 1995, pp. 42-43.

17 Leonardo's memorandum in CA, f. 780 r [287 r-a], c. 1514-15, used to be quoted, as Richter does, SS 7A, with the addition of the words "de vocie" (On Acoustics) taken to be the title of Leonardo's book, when they are, instead, Leonardo's "label" to an adjacent diagram. This is explained by Carlo Pedretti in his Commentary to the Richter anthology (Oxford, 1977), Vol. I, p. 107.

18 Mario Baratta, Curiosita' Vinciane, Turin, 1905, pp. 179-84. See also Francesco Savorgnan di Brassa', Da Leonardo a Marconi. Invenzioni e scoperte italiane, Milan, 1941, pp. 78-79, for the mention of Borelli's project of a submarine as well.

19 Gustavo Uzielli, Ricerche intorno a Leonardo da Vinci. Serie seconda. Rome, 1884, p. 403.

2020The importance and originality of Borelli's studies on the flight of birds were first recognized by E.J. Marey, La machine animale, Paris, 1873, and again, in a context that includes Leonardo's comparable studies, in roman Le vol des oiseaux, Paris, 1890, pp. 234-5, a classic publication on the subject which is not recorded in Verga's Bibliografia vinciana (1931). See also; Modestino Del Caizo, Studi di Giovanni Alfonso Borelli sulla pressione atmosferica, Naples, 1886, and, by the same author, Giovanni Alfonso Borelli e la sua opea De motu animalium, Naples, 1908…Raffaello Caverni, Storia del metodo sperimentale in Italia, Florence, 1891-1900, 6 vols, in particular Vol. III, p. 402; Giuseppe Boffito, Il volo in Italia, cit. (as in note 11 above), pp. 137-42 (in comparison with Leonardo); Raffaele Giacomelli, Gli scritti di Leonardo da Vinci sul volo, Rome, 1935, pp. 206-7, and, by the same author, "Il De volatu di Borelli," in L'Aeronautica, XIV, fasc. 3, 1934, pp. 1-15. For the wedge theory in both Leonardo and Borelli, cf. G.B. De Toni, Le piante egli animali in Leonardo da Vinci, Bologna, 1922, p. 137. In 1900 G.B. De Toni ("Osservazioni di Leonardo intorno si fenomeni di capillarita'", in Frammenti Vinciani, I-IV, Padua, 1900, pp. 53-61) had already mentioned Borelli in connection with Leonardo's experiments on capillarity. G. Pezzi, "La meccanica del volo nell'opera di Leonardo da Vinci e nel De motu animalium de Gian Alfonso Borelli", in Minerva medica, LXIII, 1972, pp. 2184-8, and Annali di medicina navale e coloniale, LXXVI, 1971, pp. 2750-82. And finally: Useful information on the life and work of Borelli is still to be found in Giammaria Mazzuchelli, Gli scrittori d' Italia, Brescia, 1762, Vol. II, Part III, pp. 1709-14. See also Pietro Riccardi, Biblioteca matematica italiana, Modena, 1970, sub voce, and Le opere dei discepoli di Galileo Galilei. Volume Primo. L' Accademia del Cimento. Parte Prima, ed. By Pietro Pagnini, Florence, 1942, pp. 21-22.

21 Leonardo da Vinci. Corpus of the Anatomical Studies in the Collection of Her Majesty The Queen at Windsor Castle. By Kenneth D. Keele and Carlo Pedretti, London and New York, 1979 and 1980, 3 vols, Vol. I, p. 362. For other aspects of Borelli's biological studies in relation Leonardo's, see F.S. Bodenheimer, "Leonard de Vinci, biologiste", in Le'nard de Vinci et l'espe'ience scientifique aux XVIe sie'cle, Paris, 1953, pp. 172-88, in particular p. 175 (flight of birds), 179 (Leonardo's studies on animal locomotion as compared with Borelli's principles of "iatophysics", i.e. the application of physics to medicine), 182 (Leonardo as precursor of Malpighi, Redi and Borelli). Cf. In the same volume the "Rapport final" by Alexandre Koyre' (p. 244).

22 Cfr. Windsor, RL 19070 v: "scrivi la lingua del pichio." See also Windsor, RL 19115 r: "fa il moto della lingua del picchio."

23 Cfr. Guglielmo Bilancioni, "Leonardo da Vinci e la lingua del picchio," in Rivista di storia delle scienze mediche e naturali, XVII, 1926, pp. 1-12. Bilancioni does not mention Borelli, but for the explanation of the mechansim of the wood-pecker's tongue he gives full credit to Borelli's pupil Lorenzo Bellini. The illustration in pl. V, fig. 11 (Figu. 9) is not based on that given by Ulisse Aldrovandi, Ornithologiae…, Bolgna, 1599, p. 838. According to Carlo Pedretti (oral communication), the way the complex mechanism of the bird's tongue is shown in an overall view of the bird's head, as seen three quarters to the right from above, is well in keeping with the type and character of Leonardo's anatomical illustrations of c. 1510 e.g. the sheet with studies of the palate, tongue and larynx, and hyoid bone, in Windsor, RL 19002 r (A. 3).

24 Roberto Marcolongo, Leonardo artista-scienziato, Milan, 1939, pp. 197 and 294.

25 Leonardo da Vinci, I libri di meccanica nella ricostruzione ordinate da roman Uccelli, Milan, 1940.

26 Reproduced in facsimile in I Disegni di Leonardo da Vinci e della sua Cerchia nel Gabinetto dei Disegni e delle Stampe della Galleria degli Uffizi a Firenze ordinati e presentati da Carlo Pedretti. Catalogo di Gigetta Dalli Regoli, Florence, 1985, pp. 92-93, no.s 43 [4084A]. See also my paper on "Leonardo's Lost Robot," (as in note 6 above), p. 109, fig. 32.

27 Cf. Carlo Pedretti, The Codex Atlanticus of Leonardo da Vinci. A Catalogue of Its Newly Restored Sheets, New York, 1978-1979, Vol. II, pp. 125-26, new folio number 812 r. The studies on the subject are as follows (in chronological order): Guido Semenza, 'L'automobile di Leonardo', in Archeion, IX,i, 1928, pp. 98-104; Arturo Uccelli, "L'automobile a molle e Leonardo da Vinci", in La lettura, no. 3, March 1936, pp. 7-8; Id., 'Leonardo e l'automobile', in Raccolta Vinciana, XV-XVI, 1935-1939, pp. 191-9; Giovanni Canestrini, Leonardo costruttore di macchine e di veicoli, Rome, 1939, in particular pp. 67-129 (section reprinted from the author's L'automobile: il contributo italiano all' avvento dell' autoveicolo, Rome, 1939 See also the interpretation by another engineer, Enrico Gigli, later published in the book by Marialuisa Angiolillo, Leonardo. Feste e teatri. Presentazione di Carlo Pedretti, Naples, 1979, one page of text accompanying pl. 6. This too with arbitrary modifications to, or distortions of Leonardo's design, in such a way that resulting machine would never work. The latest studies on the subject are as follows: Mario Loria, "Ruota trascinata e ruota motrice: L' "automobile' di Leonardo," in Leonardo nella scienza e nella tecnica. Atti del Simposio internazionale di Storia della Scienza, Firenze-Vinci, 23-26 giugno 1969, Florence, 1975, pp. 101-3; Augusto Marinoni, Leonardo da Vinci: L' automobile e la bicicletta, Milan, 1981, and, by the same author, "Leonardo's Impossible Machines," in Leonardo da Vinci Engineer and Architect. Edited by Paolo Galluzzi, Montreal, 1987, pp. 111-30, in particular pp. 124-25. See finally my paper "Leonardo's Lost Robot," cit. (as in note 6 above), for the Appendix on pp. 109-10: "Leonardo's 'Automobile' and Hans Burgkmair's 'Gala Carriages'."

28 The interpretation of Leonardo's drawings as showing a differential gear in unfounded. A differential gear is a complex bevel gear mechanism to permit differential motion of a pair of drive wheels. This is to accommodate fewer revolutions of the inside wheel versus the outside wheel during turning. The two front wheels of the platform for automata rotate together at a constant velocity.

29 Canestrini, op. cit. (as in note 25 above), p. 128: "Questi schemi che ci hanno servito di base per la ricostruzione del carro, che abbiamo eseguita, confermano, con le ipotesi che abbiamo avanzato, la convinzione che questo veicolo cosi come e' stato disegnato non puo' aver funzionato."

30 A scale drawing or at least a drawing begun as such and then turned into a rough sketch, is shown on CA, f. 878 r [320 v-a], c. 1478, which I consider as part of the studies for the programmable automaton. As in other sheets (e.g. CA, f. 956 r [347 r-b]), the wheels are shown with a 4.2 inches (106 mm) radius. Leonardo's drafting procedures are shown by the draft of a letter to his patron Giuliano de' Medci with which, about 1515, he was to complain about the behavior of one of his two German assistants, who was apparently prepared to steal his inventions. See CA, f. 671 r (247 v-b), Richter, SS 1315: "Afterwards he wanted to have the models made in wood, just as they were to be in iron, and wished to take them away to his own country. But this I refused him, telling him that I would give him, in drawing, the breadth, length, height, and form of what he had to do; and so we remained in ill will," Cf. Carlo Pedretti, Introduction to Leonardo da Vinci Engineer and Architect, cit. (as in note 26), pp. 12-13, and Achademia Leonardi Vinci, VI, 1993, p. 184.

31 This was already discussed in the Appendix to my paper "Leonardo's Lost Robot," cit. (as in note 6 above), p. 109, note 7. Carlo Pedretti has shown that a comparable application of a coil spring is found in Leonardo's so-called "helicopter" sketched and described in Paris MS. B, f. 83 v, c. 1487-90. See his Studi Vinciani, Geneva, 1957, pp. 125-29, and his Leonardo. The Machines, cit. (as in note 14 above), pp. 8-10 and 29. See also Giovanni P. Galdi, "Leonardo's Helicopter and Archimedes' Screw. The Principle of Action and Reaction," in Achademia Leonardi Vinci, IV, 1991, pp. 193-5.

32 The Uffizi and Oxford sheets contain studies for an Adoration of the Shepherds on which Leonardo was working in 1478, thus confirming the proposed date for the programmable automaton. Cf. A. E. Popham. The Drawings of Leonardo da Vinci, London, 1945. pls 50 and 51.

33 This is to correct my interpretation of CA, f. 812 r [216 v-b, c.] from "Leonardo's Lost Robot" cit. (as in note 6 above). The middle figure is a combination of both the drive and idler wheels not a mechanism for additional figures. The idler wheels are shown top left, and the drive wheels and cable in perspective, bottom center. See my reconstruction of Leonardo's robot knight in the compact disc Mechanical Marvels. Invention in the age of Leonardo. Finmeccanica Istituto E Museo Di Storia Della Scienza. Giunti Multimedia. 1997.

34 CA, f. 549 v [206 v-a], c. 1497: Quando voi fare uno effetto per istrumento, non ti allungare in confusione di molti membri, ma cerca il piu brieve modo; e non fare come quelli che non sapendo dire una cosa per lo suo proprio vocabulo, vanno per via di circuizione e per molte lunghezze confuse."

35 Zubov, op. cit. (as in note 12 above), p. 184.

36 Pierre Duhem, E e' tudes sur Leonard de Vinci. Ceux qu' il a lus et ceux qui l'ont lu, Paris, 1906, 1909 and 1913, 3 vols.

37 Giovan Paolo Lomazzo, Idea del Tempio della Pittura, Milan, 1590, p. 17. Cfr. Luca Beltrami, Documenti e memorie riguardanti la vita e le opere di Leonardo da Vinci in ordine cronologico, Milan, 1919, p. 206.


1.  Windsor, RL 12716 & 12688, c. 1508-10

2.  Eighteenth century Japanese Tea Carrier

3.  Spring driven power source Madrid MS I, f. 4 r

4.  Spring driven power source Madrid MS I, f. 16 r

5.  Spring driven power source Madrid MS I, f. 45 r

6.  Spring morphologies Madrid MS I, f. 85 r

7.  Cams. Madrid MS I, f. 24 r

8.  Cams. Madrid MS I, f. 28 v

9.  Borelli, plate V, fig. 11.

10.  Borelli, plate III, fig. 2.

11.  Codex Urbinas, f. 120 v

12.  Madrid MS I, f. 36 r

13.  Borelli, plate I, fig. 1.

14.  Borelli, plate I, fig 3.

15.  CA, f. 966 r [349 r-b]

16.  Borelli, plate V, figure 1.

17.  Madrid MS I, f. 90 r-v.

18.  Borelli, plate V, fig. 6.

19.  Borelli, plate XII, fig. 4.

20.  MS L, f. 28 v-c

21.  CA, f. 164 r-a

22.  Borelli, plate V, fig. 7

23.  Uffizi, GDS, no. 4085 A r

24.  CA, f. 812 r [296 v-a]

25.  CA, f. 878 v [320 v-a]

26.  CA, f. 878 r [320 r-a]

27.  Ufffizi, GDS, no. 446 E-r

28.  CA, f. 926 r [339 r-a]

29.  CA, f. 956 r [347 r-b]

30.  CA, f. 956 v [347 r-b]

31.  447 E Uffizi r

32.  447 E Uffizi v

33.  Ashmolean Museum P. 18 (VII) v

34.  Front view of programmable automaton

35.  Rear view of programmable automaton

36.  Bottom view of programmable automaton

37.  Top view of programmable automaton

38.  Three-quarters view programmable automaton

39-40.  Robot Knight head/neck Forster MS. II, 2 f. 65 r, 65 v

41.  Leonardo da Vinci, Il Codice Arundel 263.

42.  Robot Knight cable studies. CA 330a r [120 r-b].

43.  Robot Knight cable studies. CA, f. 330b r [120 r-c].

44.  Robot Knight cable studies. CA, f. 369 r [134 r-b].