The Real da Vinci CodeIs his mysterious three-wheeled cart a proto automobile? A remote-controlled robot? A rolling Renaissance computer? The quest to rebuild Leonardo's "impossible machine."By Tom Vanderbilt
SourceI imagined the road to unraveling a 500-year-old Leonardo da Vinci mystery would take me down rain-slicked flagstones in the crepuscular shadow of a glowering Tuscan cathedral, or perhaps through the mote-strewn catacombs of a Florentine palazzo. Instead, my first stop is a prim brick colonial on a broad, verdant thoroughfare in suburban St. Paul, Minnesota. I pull into the driveway and come to a stop in front of a tin-can robot standing astride the porch.
"My mother made it for me," explains Mark Rosheim, a roboticist who has produced designs for NASA and Lockheed Martin. His living room is dominated by two hulking cabinets, each filled with oversize editions of da Vinci codices. It is, the owner suggests with the slightest bravado, "the largest collection of Vinciana in the Midwest." He points to one set, a dozen volumes of the Codex Atlanticus, the thousand-page collection of drawings that is da Vinci's best-known work. "I got that one from Christie's in London through a telephone bid," he says. "That was before eBay. The auction was at 4 in the morning. It was very exciting."
On one wall, there is a family picture frame with a series of oval and square photos. "This is the whole Mark Rosheim saga," he says. His father's drugstore. His computer science-trained brother. His grandfather, a pioneering dentist who owned the first x-ray machine in Story City, Iowa. And Rosheim at a Cub Scout gathering at age 9. He is dressed as a robot.
As we tour the house, I get the feeling that Rosheim is not simply interested in studying da Vinci, but that he would like to be da Vinci. There are certain parallels. Da Vinci was self-taught and often referred to himself as an omo sanze lettere - a man without letters; Rosheim is a high school dropout. Da Vinci was apprenticed to Andrea del Verrocchio's workshop at age 15; Rosheim filed for his first patent - for a hydraulically powered servomechanism - at age 18. Da Vinci was determined to understand the architecture of the human body. By the time he was 65, he had dissected the corpses of more than 30 men and women of all ages. Rosheim is a student of kinesiology who has paid particular attention to the human wrist. In a basement workshop, he shows me a prototype of his Omniwrist, a joint that can move in any direction across a full hemisphere, without gears.
In the early 1990s, Rosheim's twin passions of da Vinci and robotics fatefully converged. After an Italian scholar showed Rosheim some recently recovered da Vinci drawings, Rosheim took a fresh look at what had been dubbed "Leonardo's automobile," a wooden three-wheeled cart. Da Vinci enthusiasts have reconstructed the automobile several times during the past century, but it's never worked. The device seemed destined to join the ranks of da Vinci's grandiose but flawed inventions - what one scholar called his "impossible machines."
To Rosheim, the machine was hardly impossible. Immersing himself in the minutiae of each sketch, gleaning inspiration from inventions that came later, he concluded that the device was not simply a spring-powered cart - as novel as that might be for 1478 - but something more radically innovative. Da Vinci's automobile, Rosheim maintains, is actually a robot with its own set of programmable instructions. This "precursor to mobile robots," Rosheim suggests, might even be "the first record of a programmable analog computer in the history of civilization."
The notion that da Vinci was some sort of proto-computer geek is not as far-fetched as it sounds. In a 1996 article in the journal Achademia Leonardi Vinci, Rosheim offered compelling historical and mechanical evidence that da Vinci had designed - and perhaps built - automata. Rosheim pointed to da Vinci's so-called Robot Knight, a cable-and-pulley-driven artificial man, which had been thought to be a simple suit of arms. Citing drawings discovered decades earlier by Italian scholar Carlo Pedretti, Rosheim explained how the figure "sat up, waved its arms, moved its head via a flexible neck, and opened and closed its anatomically correct jaw - possibly emitting sound while accompanied by automated musical instruments such as drums."
The robot, the theory goes, may have been commissioned by the Sforza rulers as court entertainment or an exhibit in a kind of mechanical sculpture garden. A finished drawing of the knight has never been recovered, but Rosheim, armed with mechanical aptitude and a strong knowledge of the history of robotics, was able to extrapolate its use from a patchwork of drawings. Paolo Galluzzi, director of Florence's Institute and Museum of the History of Science, described Rosheim's robot thesis as "absolutely convincing." Galluzzi included the knight in an exhibition and commissioned Rosheim to create a computer model. In 2002, Rosheim was invited by the BBC to build a prototype. His model was able to walk and wave - proving Rosheim's theory once and for all.
Vindicated, Rosheim revisited other da Vinci machines. His searching led to a 1975 article written by Pedretti, the same scholar who had done pivotal research on the knight. The article presented Pedretti's analysis of a new sheet of drawings discovered in a collection at Florence's Uffizi. They were sketched by an anonymous 16th-century draftsman but included copies of da Vinci's technological studies. Pedretti focused on one sketch that clearly outlined the function of the arbalest-like springs in the depiction of da Vinci's baffling three-wheeled cart. They were, he realized, not for power, as earlier scholars had thought, but for steering. Like an escapement mechanism for clocks, the springs retained movement but didn't generate it. He concluded that the movement must come from somewhere else. So Pedretti looked back at da Vinci's original drawing and noticed a faint circle in the center of one of the car's toothed gears. The little circle, he believed, was almost a suggestion to look for something transparent, something beneath the cart.
Perhaps there were larger coil springs, hidden inside the tambours, that would drive the cart.
I take all this in at Castel Vitoni, Pedretti's magisterial Italian estate, which commands a view of the Tuscan valley, including - not accidentally - the pastoral town of Vinci, Leonardo's birthplace. We sit in his office and pore over sketches of the cart on folio 812 recto of the Codex Atlanticus. I reach carefully for the espresso his wife has placed on the table, trying not to spill any on a nearby copy of the Italian mathematician Bernadino Baldi's 1589 translation of Heron of Alexandria's Automata. It is a first edition.
The sketch of the cart is not particularly impressive to look at. On the top of the page is a crudely drawn wagon with some sort of gear mechanism. The bulk of the page is dominated by a closer view of that mechanism, which combines a crossbow-like arbalest with the grooved gears and verge-and-foliot apparatus found in medieval clocks. On the periphery of the page, as on many Codex pages, there are details of component parts.
Though Pedretti had uncovered fragments of robot designs in da Vinci's sketchbooks, he couldn't figure out how they fit together. Rosheim, who had started corresponding with Pedretti after meeting him in 1993, began developing a CAD reconstruction and faxing documents to Pedretti at night. It was like a fill-in-the-blanks puzzle. "There's nothing saying, This is an automaton," Rosheim recalls, explaining how he contrived a robot. "I'm working with napkin sketches. It's very fragmentary stuff - otherwise it would have been done centuries ago." To divine what the artist envisioned for the cart's undercarriage, Rosheim tried to internalize the da Vinci method, studying myriad other drawings "to load it up into my subconscious" and inventing "an internal calculus to try and figure out everything."
One of the biggest breakthroughs, strangely enough, came not from da Vinci's own work but from a drawing Rosheim had of a karakuri, an 18th-century Japanese tea-carrying automaton (often resembling a geisha) - the Sony Qrio of shogunate Japan. The movement of the karakuri was determined by the placement of cams, small appendages on a wheel or shaft that engage a lever and convert rotary power to linear power. (Cams are still found in today's car engines.) Looking at the karakuri, Rosheim thought that da Vinci's cart might contain a similar arrangement. Sure enough, he found small camlike protrusions attached to one of the toothed wheels in da Vinci's drawing. The karakuri seemed to provide the missing link to understanding the cart's undercarriage - a perspective not shown in the sketches.
Rosheim's epiphany answered questions he'd been unable to resolve: How did the escapement work? How did you regulate the speed - in other words, the clock of the computer? How did that connect to the rest of the drivetrain? Once you understand the cams, the faint circles underneath the middle of the frame of the perspective view suddenly make sense, he says. "Obviously, they connect to one of those levers that's cam-controlled." The inspiration may have come from 18th-century Japan, but Rosheim says his ideas - unlike previous reconstructions - mesh perfectly with da Vinci's original design.
So here you had a small, front-wheel-drive cart no more than 20 inches square - many Codex illustrations are one-to-one scale fabrication drawings - that could, on the basis of spring-loaded power, be triggered via remote control and run a specific course, turning in a programmed direction at a certain point and perhaps even executing a "special effect" or two. What on earth was it for?
If Rosheim was able to supply the how of da Vinci's robot cart, Pedretti could offer a why: court entertainment. Da Vinci, he says, would have been 26 when he built the cart. It was 1478, and Florence was especially volatile: The Pazzis were conspiring against the reigning Medici family (da Vinci sketched the hanged Bernardo Bandini, who murdered Giuliano de Medici during the plot). The historical record offers no mention of da Vinci having built a cart. Pedretti, however, unearthed a potential clue. "I found a fantastic document, date 1600," Pedretti says. "It's a description of a banquet held in Paris to honor the new queen of France, who was a Medici. On that occasion, Michelangelo Buonarroti the Younger observed the presentation of a mechanical lion. It walked, opened its chest, and in place of a heart it had fleurs-de-lis." Pedretti pauses, gathering more papers. "This document, which was totally unknown, says this was a concept similar to one that Leonardo carried out in Lyons on the occasion of Francis I." It appears da Vinci had engaged in high tech diplomacy circa 1515.
The cart, suggests Pedretti, may have been an early study in an emerging da Vinci sideline. Leonardo, he believes, created animated spectacles centuries before the great age of the European automata of Jacques de Vaucansan and Wolfgang von Kempelen. "The irony of the whole thing is that there is not a single hint in Leonardo's manuscripts of this greatest technological invention," Pedretti says. "Imagine to have a lion walk and stand on its legs and open up its chest - this is top technology!" What happened to those pages of drawings that would have revealed the inner workings of these wondrous devices? Perhaps they lie misfiled in some lost archive; perhaps they were destroyed by some church authority in the manner of Albertus Magnus' mechanical woman, smashed by Thomas Aquinas as a work of the devil.
Half a millennium on, the cart could, says Rosheim, not only rewrite the history of robotics but also bring another da Vinci to light: da Vinci the roboticist. "If it was simply a spring-powered cart, it would not be that big a deal," he says. "What's significant is that you can replace or change these cams and alter how it goes about its path - in other words, it's programmable in an analog, mechanical sense. It's the Disney animatronics of its day." The individual parts, interestingly, are not original to da Vinci - gears, cams, and the verge-and-foliot mechanism were all familiar concepts, particularly to clockmaking, the nanotech of da Vinci's day. Indeed, as the historian Otto Mayr has noted, "clocks and automata, in short, tended to be very much the same thing"; clocks, in 16th-century dictionaries, were considered just one type of automata. But the possibility is that da Vinci married two ideas and created, in essence, a clock on wheels - turning the segmenting of time into the traversing of space - well before anyone else had thought of such a thing. No one could have done it as elegantly, in so compact a package, says Rosheim. "The robot cart is one of the most significant missing links in studying Leonardo. Suddenly, many drawings are making sense."
Just down the road from Pedretti's villa, I sit for an hour staring at two wooden models of da Vinci's three-wheeled machine at the Vinci's Leonardo Museum. The reconstructions were built by the Milanese design firm Studio DDM, working with a Florentine carpentry shop. After weeks of peering at the faded filigree of ancient manuscripts, I find it strange to see da Vinci's drawings in three dimensions. The models look at once primitive and complex, like out-of-time machines, steampunk for the Middle Ages.
"A lot of people say that Leonardo's machines will not work," says DDM's Mario Taddei the next day, as we gaze at his laptop. We are sitting in a Florence café overlooking the Arno, a body of water da Vinci had once proposed rerouting to the sea. "Half of them are perfect, half are not so perfect." Taddei, who's interested in blending the world of historical museums and videogames, is showing me CAD drawings of the robotic cart. "The design was so perfect," Taddei says, "that the first time we built the machine and charged the spring engine, it worked perfectly - something very strange in the world of Leonardo reconstructions." Taddei credits Rosheim for the central idea of cam-driven programmability and says his team used Rosheim's drawings, making subtle changes along the way.
Rosheim had only one comment on the reconstruction: "They apparently didn't figure out how the escapement mechanism works, because theirs just kind of runs really fast and then runs out of steam." When I speak with him several weeks later, he is nearing completion on his own reconstruction, which he has been building in his basement with his own money. The model, along with another "top secret" reconstruction, will accompany his book, Leonardo's Lost Robots. He tells me his model backs up the theory of his original drawing. "As you see in Codex Atlanticus folio 812, Leonardo has one half of the right large gear with cams and the other half with none. This generates a left-right zigzag motion."
Whatever the minor technical differences between Rosheim's vision and the handiwork of the builders at DDM, the cart provides further evidence that da Vinci was a Renaissance roboticist. Of course, absent the complete drawings of the cart, we will never know exactly what da Vinci had in mind. The cart will remain among his many inventions and artworks that plague scholars the world over. It's ultimately part of that grand guessing game: Who was Mona Lisa? Why did da Vinci leave Florence for Milan? Why did he not complete The Adoration of the Magi or any number of other major commissions? Was his glider ever launched?
"Leonardo is the Hamlet of art history," says art historian Kenneth Clark, "whom each of us must re-create for ourself." Da Vinci has been credited with inventing just about everything but the Internet. Now, from the faintest of sources, an American roboticist and an Italian Renaissance scholar have discovered another Leonardo: The creator of Hollywood-style special effects, perhaps even a lost progenitor of the programmable computer, is coming into frame.
Tom Vanderbilt (tvanderbilt@nyc.rr.com) is the author of Survival City: Adventures Among the Ruins of Atomic America.