In 1969 the Congress of the Comité International d'Histoire de l'Art was held in Budapest. The event was accompanied by a small dossier exhibition in the Museum of Fine Arts centred on a famous small bronze in the museum's collection, a horse and rider attributed to Leonardo da Vinci (fig. 1). The Budapest Horse and Rider was exhibited together with two close variants of the rearing horse, one from The Metropolitan Museum of Art in New York, and another which is now in The Hunt Museum in Limerick [1]. A brief report about the exhibition has survived for posterity: a short documentary film entitled Horse Race in the Museum put forward the arguments of Mária Aggházy, curator of the exhibition and keeper of the collection of sculptures, who strove to demonstrate that the Budapest small bronze was the only extant sculptural work by Leonardo's own hand. Given the set-up, it will come as no surprise to learn that the Budapest Horse and Rider turned out to be the winner of the race, and in the closing frames of the film, under the watchful eyes of the curator, a technician replaced the label on the statue's pediment, changing the legend "Follower of Leonardo da Vinci" to the infinitely more favourable "Leonardo da Vinci". This was more than a dramaturgical device; it was a genuine act of symbolism, the opening move in Aggházy's lifelong quest to have the Budapest Horse and Rider accepted as an autograph work by the great master. Aggházy's attribution proved fragile, however. At the start of the new millennium, I had the misfortune of being present when the small bronze, which had hitherto been an emblematic highlight of the collection at the Museum of Fine Arts, was mournfully ushered away from pride of place in one of the central cabinets of the Old Masters Gallery to a hidden corner of a nearby corridor. The statue's exile was the direct consequence of rumours that had long echoed through the corridors that the Budapest Horse and Rider had been cast centuries after Leonardo's time, just like its New York counterpart, which had recently been unmasked as a product of the nineteenth century. The story took a more fortuitous turn in 2009, when the statue was subjected to a thorough technical examination at the National Gallery of Art in Washington DC. The results dissipated all the doubts surrounding the date of the Budapest Horse and Rider, and the small bronze went on display that same year at the High Museum in Atlanta, and the following year at the Getty Museum in Los Angeles, as part of the most important exhibitions to date dealing with the sculptural activity of Leonardo da Vinci. Nowadays, the Budapest Horse and Rider is more popular than ever: it is one of our most frequently requested and exhibited masterpieces, and it is once again a central protagonist in the discourse on Leonardo as a sculptor. So why do I still nurse nagging doubts about the recent success of the small bronze? The way I see it, the Budapest Horse and Rider has a tendency to rise and fall in Leonardo's œuvre, following a pattern of zeniths and nadirs that is somewhat inevitable and almost predictable. A glance through the hundred-year history of the statue shows that opinions have always been divided between outright rejection and more optimistic standpoints. If our judgments are so uncertain, why is the authorship of the Budapest Horse and Rider, or indeed of any other work, such a defining question in art history? It is all part and parcel of our generally accepted practice of attribution, of trying to match artists whose names we know with artworks that survive from the times when those artists were around. The oft questioned, yet nevertheless widespread view regarding the authority of an individual artist as the central feature of any artwork goes back a long way and is deeply rooted in art historical thinking. In consequence of the notion that unique and unrepeatable artworks were brought into existence by the mind of a genius, by the end of the nineteenth century, determining the authorship of a work had become a central concern of art history, and this is reflected in the approach commonly taken by museums, in which the history of art is presented with emphatic reference to the names of the artists.

How is it possible to determine the creator of an artwork from a distance of several centuries? Our options are severely restricted. Artworks in those days were rarely signed, and even if a name or monogram can be made out, it was more often than not placed there by a later collector or art dealer, and should not be accorded much credit. If we want to achieve a degree of certainty, then we need to rely on documentary evidence (mostly in the form of contracts, contemporary accounts or collection inventories), which in some cases may even allow us to trace the work back to the master's workshop. The only problem in the case of the early modern period is that precious few documents have survived, and most of the extant descriptions are impossible to associate with any specific artworks. In the absence of written documents, we have no choice but to pass our verdict exclusively on stylistic grounds. This method is based on the idea that works are imbued with unique and clearly recognisable stylistic distinguishing features that enable the well-trained eye to identify the author of a work, or at least the time and place of its creation, with a greater or lesser degree of reliability. Although this practice is as old as the first art collections, connoisseurship in the modern sense of the term only began around a century and a half ago, thanks to the efforts of an Italian amateur art critic named Giovanni Morelli. Morelli decried the fact that the art historians and connoisseurs of his times mostly relied on their instincts when determining the author of a work, so their judgments were passed not in accordance with their observations but with their preconceptions. Instead of basing his assessments of paintings on the most obvious aspects of the works, such as their composition, the proportions of their figures and their distinctive colouration, Morelli focused his attention on tiny details which, at first sight, would appear to be insignificant. He was convinced that the true hand of the artist could be revealed through such incidental hallmarks as the shape of the bridge of a nose, the outline of a fingernail or the curve of an earlobe. Just like Sherlock Holmes, the popular literary detective who could solve the most complex mysteries by studying small clues that others overlooked, Morelli used dispassionate, methodical analysis to deduce the true identity of the artists behind the masterpieces that hung on the walls of picture galleries throughout Europe. In the meantime he was vehemently opposed to the idealist concept of art history that pervaded his day and age, trusting instead that there had to be a precise and scientific method of understanding art. His belief that correct conclusions could only be drawn by making observations that were free from all prejudice and from all aesthetic preference stemmed from the positivist thinking of the nineteenth century. Like the empiricists, Morelli opined that knowledge could be obtained exclusively through experience. In his view, the only evidence a connoisseur should use to judge an artwork was its style; he therefore gave the systematic comparative analysis of form supremacy over all other methods, including the interpretation of relevant documentary sources and art historical knowledge. Truth be told, however, Morelli often acted no differently from the contemporaries he so roundly criticised, and the majority of his attributions rested on his instincts rather than on rigorous scientific foundations. Unsurprisingly, a few decades after Morelli's death, his main detractor, the German art historian Max Jakob Friedländer, two generations younger and at least as successful a connoisseur as Morelli, reverted to the earlier contention that the only thing that could lead to a true attribution was the art expert's intuition. Whereas Morelli's method is usually compared with empirical science, based on exact observations, in Friedländer's estimation, connoisseurship constituted an intellectual and intuitive endeavour, in which the leading role was played by the overall impression generated by the artwork at hand. It is easy to spot the problem with the "scientific connoisseurship" advocated by Morelli. What we see, after all, is substantially more than the image projected onto our retinas, for it is influenced as much by our prior expectations of the image presented before our eyes as it is by our cultural environment and our background and training. Despite the fact that one of the simplest and most widely followed ideas within scientific cognisance, namely that conclusions of a generally applicable nature can be reached only via objective observations, has been called into question in the last century both by perceptual psychology and by the philosophy of science, this idea continues to occupy a prominent position in art history. Similarly, Morelli's "scientific" method left an indelible imprint on modern connoisseurship. As alluded to in the 1970s by the American art historian and theorist James S. Ackerman, "…without knowing it, my colleagues have grounded their method in the tradition of nineteenth-century positivism conceived to justify scientific empiricism". The problem with regarding empiricism as the sole valid path to cognisance is that it imposes limitations on the options available to us. The situation with the Budapest Horse and Rider is a case in point. Since the method of connoisseurship is based on the comparative analysis of artworks, before we can say anything at all about the authorship of a work in question we need to have at our disposal at least some works that can be unquestionably attributed to the given artist. No documented statue by Leonardo survives, however, nor even one that is widely accepted as his. And here is the stumbling block: if we do not know a single statue that was certainly by Leonardo, on what basis can we claim that the Budapest Horse and Rider was by his hand? Is it sufficient to look at the similarities in form between the small bronze and drawings by the master? Are we capable of conceiving, simply by studying his drawings and paintings and by reading what his contemporaries wrote about them, what Leonardo's statues would have looked like? Statues that in most instances never came to fruition? Due to the lack of proven Leonardo sculptures, from a strictly methodological point of view, the debate about the authorship of the Budapest Horse and Rider may seem entirely futile; an exercise on which it would be better not to waste our time. I am of the opinion, however, that the value of an attribution does not lie exclusively in its "objective" correctness, but rather in its capacity to inspire new ideas and discussion. An attribution as such is not primarily a fact, but rather a necessary hypothesis that depends for its validity on context and interpretation. Ideally, as an intellectual tool it initiates new directions leading to fresh inquiries rather than posing as a final affirmation. A new attribution with its hypothetical truth—as opposed to the certainty of truth—may usefully serve as a means for introducing new theories into the wider field of our understanding. This was expressed by Friedländer as "an attribution cannot be proved or disproved" and therefore "the only criterion of the truth is that it should prove fruitful". If we accept this more flexible and permissive definition of an attribution, then it may be worthwhile taking a step back from the traditional practice of attempting to identify the author of the Budapest Horse and Rider within the confines of the scientific rigour of positivism; instead, it may be "fruitful" to investigate the most important considerations that have played a role in the evaluation of the small bronze over the last hundred or so years.

The Budapest Horse and Rider, together with several other small bronzes, was purchased in Rome some time between 1818 and 1824 by the Hungarian sculptor István Ferenczy; its earlier provenance is now lost in time. When he bought it, nobody considered it to be by Leonardo. In 1846, when Ferenczy offered his collection for sale to the Hungarian National Museum, he listed the Budapest Horse and Rider as an "Athenian Greek statue". The transaction never took place, and the collection lay forgotten for many years, packed in sealed crates in an annex of Ferenczy's home in Rimaszombat (now Rimavská Sobota, Slovakia). According to a family tradition, the sculptor gave instructions that the crates should remain unopened for fifty years after his death. Credence must be given to the legend, for when Ferenczy's estate entered the possession of the Museum of Fine Arts in 1903, there was no mention of his collection of small bronzes. Even the sculptor's monographer, Simon Meller, who was later responsible for the first publication of the Budapest Horse and Rider, knew nothing of its whereabouts at this time. The next we hear of the statues is in 1913, when the Museum of Fine Arts decided to purchase the entire collection of small bronzes. The following year, the Budapest Horse and Rider was inventoried as a sixteenth-century Florentine sculpture. Simon Meller, keeper of the collections at the Museum of Fine Arts, however, may have suspected even then that this was no ordinary "Florentine piece", for next to the entry, albeit with a question mark, was written the name "Leonardo". It did not take long for the first press reports to appear suggesting that the equestrian small bronze in the Ferenczy collection might just be the sole surviving statue by Leonardo's own hand. In 1916 Meller finally published the Budapest Horse and Rider as an autograph work by Leonardo da Vinci. An attribution is informed by numerous sources, so our stylistic conjectures should ideally be backed up with written sources and evidence pertaining to the origin of the work. As not a single document has survived that mentions the Budapest Horse and Rider, and its provenance can be traced back no further than the Ferenczy collection, only visual clues are available to us to support its interpretation. As we have already mentioned, however, there is not a single sculpture that can be securely attributed to Leonardo. Meller overcame this not by seeking the master's hand in specific details of the small bronze, as would be the case with conventional connoisseurship, but by concentrating on the general formal appearance of the work. The motif of the rearing horse in the Budapest Horse and Rider can be seen in several of Leonardo's drawings, and Meller concluded that comparing the statue with these drawings, in particular with the designs the Italian master made for his equestrian monuments, would lead to a satisfactory result. As Meller believed that the solution to the conundrum lay in Leonardo's drawings, a good two thirds of his lengthy essay was dedicated to the horse studies, and only the last few pages referred to the small bronze. Like his German and Austrian contemporaries, Meller believed that the development of art was governed by processes similar to the laws of nature, which were of such regularity that they helped one understand not only the history of style in general but also the evolution of an individual œuvre. His contention that "the section of Leonardo's designs for his compositions often show a regularity that recalls the developmental law of organic nature" harked back to the ancient trope of the organic growth of art. In his study, he presented Leonardo's equestrian drawings as links in a connected chain, in which each consecutive work was the result of continuous, uninterrupted development, implying that his life's work progressed from the simple to the complex, from the rudimentary towards ever-increasing perfection. The Budapest Horse and Rider was, in Meller's view, the culmination of this development, "in which the old motifs, now further developed, are adapted to the new purpose and appear merged from the inside out into a fully integrated unity". From this point on, it was only natural for Meller to identify this small bronze, the pinnacle of Leonardo's depictions of horses, with the master's last known sculptural commission, the planned equestrian monument to the condottiere Gian Giacomo Trivulzio. Meller's attribution was received favourably, but not without a host of reservations. Though many questioned his attribution of the small bronze, it was his new chronology of the Leonardo drawings that provoked the sharpest criticism. At the start of the twentieth century, Leonardo's draughtsmanship, especially his horse studies, still constituted a relatively fresh topic. The starting point was a book published in 1889 by the German art historian Jean Paul Richter, an anthology of the written legacy and notes of Leonardo. Thanks to its large colour facsimiles, this volume brought many previously unfamiliar Leonardo drawings to wider audiences, including the majority of sheets in the Royal Collection at Windsor, which contain almost all his studies of horses. Richter still believed that all of Leonardo's equestrian drawings were made for a single project, the Sforza Monument in Milan. A few years afterwards, however, the brilliant Leonardo expert Paul Müller-Walde, deducing that some of the drawings were significantly later than the Sforza Monument, concluded that they may have been intended for the Trivulzio Monument; a decade and a half after this, Meller drew a parallel between the latter commission and the Budapest small bronze. This was the first step towards a more thorough understanding of Leonardo's equestrian drawings, and was the dominant view up until 1935, when Kenneth Clark published his classic catalogue of Leonardo's drawings at Windsor.

Horses played an important part in Leonardo's œuvre. In the 1470s, he spent his years as an apprentice in the workshop of Andrea del Verrocchio, the most successful sculptor in Florence at the time, who was working on the equestrian statue of Bartolomeo Colleoni. The project left an impression on the young Leonardo. His first drawings feature mounted riders fighting dragons or griffins (fig. 2). Among them is the earliest instance of a rearing horse, the motif immortalised in the Budapest small bronze. Furthermore, although no known painting or mural on the subject exists by Leonardo, we do know that the legend of Saint George and the dragon was a topic the artist returned to several times during his career (fig. 15). The background of the unfinished Adoration of the Magi (Uffizi, Florence), which he began to paint for the monks of San Donato a Scopeto in March 1481, was populated with similar mounted riders. One of the drawings on display at the Budapest exhibition was probably made as a study for this painting (fig. 3). Leonardo soon moved from Florence and settled in the Milanese court of Ludovico Sforza. Shortly after his arrival, the artist began work on his design for the seven-and-a-half-metre-tall equestrian statue of Ludovico's father, Francesco Sforza, the condottiere who rose to the rank of Duke of Milan and founded the Sforza dynasty. Leonardo's earliest sketches, dating from the end of the 1480s, indicate that he originally intended to produce a rearing horse, modelled on works from antiquity, especially coins, medals, gems and small bronzes (figs 4, 6 and 7). At the duke's insistence, however, and almost certainly because the artist also realised that his ambitious plan was impossible to cast, Leonardo soon opted for a more conventional pacing horse (fig. 11). Though work progressed slowly, by 1493 Leonardo was in the middle of preparations for his enormous clay model of the horse to be cast in bronze. When the French army invaded Lombardy, however, the work was abandoned. First, the bronze that had been set aside for the monument was sent to Ferrara to make cannons; then in 1499, after the fall of the duke, Leonardo also had to flee the city. The clay horse erected in the castle courtyard was used by Gascon archers for target practice. In 1501 it was transported to Ferrara, where it eventually fell to pieces. Though the clay model was destroyed, numerous drawings Leonardo produced for the statue have survived for posterity: rapid sketches of the horses from the famous Milanese stables, intricately detailed anatomical and proportional studies, and technical sketches and notes explaining how the enormous model should be transported and cast (fig. 5). Judging from these drawings, it seems that while the Sforza Monument owed much to Verrocchio's statue of Colleoni, Leonardo nevertheless regarded Donatello's more classicising monument to Gattamelata in Padua as his primary model. The final version, however, seems to have been inspired by equestrian statues from antiquity, above all the Marcus Aurelius in Rome, the horse and rider in Pavia that is known as Regisole, and the bronze horses of Saint Mark's Cathedral in Venice. After the occupation of Milan, Leonardo was soon back in Florence, where he received the most prestigious commission of his life: to paint a fresco cycle on the wall of the Hall of the Great Council in the Palazzo della Signoria (now Palazzo Vecchio). The subject for the work was the Battle of Anghiari of 1440, a resounding victory of the Florentine Republic over their eternal rivals, Milan; Leonardo set about planning a scene of horsemen coming to blows. His drawings, as well as copies made by his contemporaries, seem to prove that the dynamic clash between groups of rearing horses offered Leonardo an opportunity to rephrase the composition he had drafted two decades earlier for the background to the Adoration of the Magi (figs 3 and 8). Stopping only for brief intervals, he worked on the fresco from October 1503 to May 1506, but he had only completed the central episode of the battle, the Fight for the Standard, when he was invited back to Milan. From then until spring 1508, he divided most of his time between the two cities, but he never returned to the abandoned fresco. Although his work, protected by a wooden railing, was long regarded as one of Florence's main sights, the poorly heated and damp hall caused the fresco to deteriorate; its final traces were removed in 1563 when the hall was renovated. In summer 1508 Leonardo returned to Milan for good. Shortly afterwards he agreed to make a funeral monument for the condottiere Gian Giacomo Trivulzio, who had toppled the rule of the artist's former patron, Ludovico Sforza, and became Governor of Milan. This life-size equestrian statue, far more modest than the one envisaged for Sforza, never even reached a serious planning stage. Very little is known about the project, apart from an autograph cost estimate and a few drawings that are reminiscent of the preparatory studies Leonardo produced for the Sforza Monument a decade and a half earlier (figs 9 and 10). It is possible that the series of marvellous sketches he drew in France in the final years of his life were developments of his design for the Trivulzio Monument, although they could also have been plans for yet another equestrian statue (figs 11 and 12). Despite the presence of the pacing or rearing horse throughout his career, Leonardo used a relatively narrow repertoire of the motif, which recurs in his œuvre with varying degrees of modification. This prompted Clark to contend that the chronology of the drawings could not be established merely by observing their subject matter, but relied on such additional details as the style and technique of the drawings, the watermark on the paper, or Leonardo's handwriting. Meller's formalist approach, however, ignored these subtle differences, although we should concede that he wrote his study during the First World War, when he had no opportunity to study the original drawings and had to rely exclusively on reproductions. This may explain why he preferred to approach the small bronze from the already obsolescent evolutionary development perspective. Nothing, however, justifies his dogmatic rejection of almost all the earlier findings published by Müller-Walde. Unlike Müller-Walde, Meller had never dealt with Leonardo before, and although he was well informed, he was by no means an expert on the subject. Whereas the observations of his German colleague were based on direct and thorough knowledge of the drawings and written documents, Meller's study was restricted to secondary sources, mostly to the literature and the reproductions found therein. In 1935, when Clark published a new, more convincing chronology, largely based on Müller-Walde, Meller's study faded into the periphery.

Although Meller no longer played a key role in the discourse on Leonardo, he had firmly placed the position of the Budapest Horse and Rider at the point where the master's equestrian drawings and sculptural œuvre meet. The fact that not a single statue by Leonardo has survived did not deter art historians from trying to fit all manner of works into this gap. Over the last century and a half, numerous small bronzes as well as statuettes in terracotta and marble have been attributed to Leonardo and then almost immediately forgotten again, while the master's sculptural œuvre has turned into a haven for guesswork. There are numerous documents and contemporary accounts which prove that Leonardo was active as a sculptor, but none can be associated with a single specific sculptural work. At the same time, there are quite a few sculptures which, on the basis of their style, could justifiably be regarded as works by Leonardo, although none of these are corroborated by the written sources. Paradoxically, therefore, while we have good reason to assume that Leonardo did indeed make sculptures, we suffer from a paucity of evidence to back up any attribution. In the absence of extant sculptures, the comparative method of connoisseurship remainsrather fruitless, and therefore—like Meller—we are forced to make comparisons between Leonardo's possible sculptures and his proven paintings and drawings, in particular the designs he made for his equestrian statues. Our starting point, meanwhile, has to be the widely held presumption that Leonardo's paintings and drawings contain sufficient distinguishing features for us to be able to imagine how his sculptures would have looked. This means that all we are basing our decisions on is the quality of the works: if a statue seems to be good enough when compared with Leonardo's drawings and paintings, then we lean towards accepting it as an autograph work, while in the case of a weaker correlation, the attribution is easily rejected. Of course, what exactly we understand by good enough not only fluctuates with each new age, but also depends on our own personal tastes and beliefs. Beyond the subjective nature of our judgments, several other problems may be encountered when attempting to determine the authorship of artworks solely on the basis of their quality. First and foremost is the risk of regarding an artwork as the accomplishment of a single individual at a time when works were mostly produced collectively in workshops whose members were numerous and constantly changing. Moreover, we may ignore the fact that the quality of a work not only depends on the artist's talent and ambition, but also on the amount of time devoted to the piece and the quality of the materials used, in other words, the amount of money the person commissioning the work was prepared to spend. In most cases, the client specified the subject matter and sometimes even insisted on a particular model that the artist should follow. On top of all this, if the only criterion to which we resort is the quality of the artworks, then we have to assume that all the works Leonardo produced, in all the genres of the arts, were consistently of the exact same standard—and that includes his first youthful trials, the masterworks from his mature period, and the works made in his old age. The central question is how far to spread the net that embraces not only the painter of the Mona Lisa and the Last Supper, but also the creator of the early Florentine Madonnas, while at the same time excluding the works of his pupils and followers. To put it another way, which works by Leonardo provide adequately solid foundations on which to base judgments of any kind?

When it comes to the works now attributed to Leonardo, we are inclined to treat them as though they have always appeared with his name attached to them. However, there are just four paintings by Leonardo whose authenticity and authorship have never been disputed: the unfinished Adoration of the Magi, the likewise incomplete Saint Jerome (Rome, Vatican Museum), the severely damaged and repeatedly restored Last Supper (Milan, Santa Maria delle Grazie), and the Mona Lisa (Paris, Musée du Louvre). The Adoration of the Magi is the only documented work from Leonardo's early period: the archives of the Monastery of San Donato a Scopeto house a contract from March 1481, commissioning the altarpiece, and a payment receipt. The Last Supper, painted on the wall of the refectory in the convent of Santa Maria delle Grazie, appears frequently in contemporary documents. Moreover, it is still in its original location, although it is only a pale shadow of its erstwhile glorious self. By contrast, the Saint Jerome in the Vatican is not mentioned in any account or document until the nineteenth century, and yet no doubts have ever been voiced concerning its author. The situation with the Mona Lisa is different again, for attempts have been made to connect the painting with countless written sources. The earliest is a travel account from October 1517, whose author visited Leonardo in the Château de Cloux (Clos Lucé), where he saw, alongside several other paintings, a picture of "a certain Florentine woman"; many believe this to be the first written reference to the Mona Lisa. Another oft cited source is the Codex Magliabechiano by an anonymous author, written around 1530–40, which contains brief biographies of Florentine artists, including Leonardo; here mention is made of a "portrait of Francesco del Giocondo", which is either a slip of the pen, for the Mona Lisa is a female portrait, or it refers to a different painting. The most important source of Italian Renaissance art, The Lives of the Most Eminent Painters, Sculptors, and Architects by Giorgio Vasari (first edition 1550, revised second edition 1568), provides a detailed description of the painting. It is likely, however, that the author only knew about the painting from hearsay, for while he makes no mention of the work's idiosyncratic composition or the landscape stretching into the background, he waxes lyrically and at length on the eyebrows, a detail that cannot even be discerned in the painting. Although the above sources are somewhat confusing, and in one instance we cannot even be sure that it actually pertains to the painting in the Louvre, the authenticity of the Mona Lisa has never been called into doubt. The assertion that the Mona Lisa was painted by Leonardo, and that it is the very same painting as the female portrait exhibited in the Louvre, is the result of several centuries of consensus, rooted in art historical tradition, which has exerted a defining influence on European culture ever since. And while we attempt to formulate an opinion about the identity of the author of a painting, sculpture or drawing on the basis of relevant written sources, the style of the work, and our own personal aesthetic judgments, the attribution of the majority of artworks rests not on incontrovertible facts, but on consensus among scholars. One such consensus came about in the 1930s, when Kenneth Clark proposed a connection between the Budapest Horse and Rider and a lost clay or wax model by Leonardo. Small sculptural models were often used in Renaissance workshops in order to work out compositions for multifigural scenes or to register figures in unusual poses. Such easily variable small wax figures were used by Michelangelo in his preparations for the Last Judgment in the Sistine Chapel, and Leonardo recommended using similar miniature models on one of the Windsor sheets containing his studies for the Battle of Anghiari [2]. This sheet formed the basis for Clark's notion that the Budapest Horse and Rider might have been modelled on one of these tiny wax or clay horses, which was later enlarged and cast in bronze by an apprentice or follower of Leonardo. In Clark's opinion, this change in scale would also explain why the statue was lacking some of the dynamism that is so palpable in Leonardo's drawings. Clark's hypothesis gained wide acceptance, for it represented a compromise in the debate surrounding the authorship of the small bronze. At the same time, by not associating the small bronze directly with Leonardo, but merely with the Leonardesque idea, he was returning to an earlier tradition of connoisseurship. Modern connoisseurship presupposes that artworks carry the unique signature hallmarks of the artist's hand, consisting of tiny—but crucial—stylistic and technical details; in this instance, however, the Budapest small bronze does not bear the imprint of Leonardo himself, only his idea, as conveyed by an unknown master. This is exactly the idealist strain of connoisseurship against which Morelli and his countless followers took such a strong stand.

Due to the lack of statues, Leonardo's sculptural activity continues to be an area of research that is fraught with uncertainties; despite this, the Sforza and Trivulzio monuments have long garnered particular interest. The equestrian statue of Francesco Sforza is the best documented of all Leonardo's works. Dated references and contemporary commentaries, as well as a wealth of Leonardo's horse studies and notes devoted exclusively to the problem of casting, allow us to reconstruct the evolution of the project (figs 4 and 5). All that is left of the documents pertaining to the Trivulzio Monument, by contrast, are Leonardo's cost estimate and a few drawings; his contemporaries also had nothing to say about it, at least, not as far as we know (figs 9 and 10). As neither work was actually executed, Leonardo's drawings represent the only clue we have regarding the possible appearance of his equestrian statues. Leonardo's drawings of horses, however, are more or less modified reiterations of the same motifs, so it is far from certain which drawing was made for which commission. In the early twentieth century, as the vociferous debate rumbled on around the equestrian drawings, the position in Leonardo's œuvre accorded to the Budapest Horse and Rider varied depending on one's position with regard to the drawings associated with it: Meller considered the small bronze to be a model for the Trivulzio Monument, while others believed it to date from the period when Leonardo was designing the Sforza statue; a third camp linked the work to his preparations for the Battle of Anghiari. This increasingly circular and unfruitful discourse was given a new twist in the 1970s by Mária Aggházy, who suggested that the Budapest Horse and Rider was a "model cast" produced by the elderly Leonardo as a study for his planned equestrian statue of King Francis I of France. The rider, in her view, was none other than the ruler cast in the guise of King Arthur. Aggházy supported her hypothesis by pointing out the similarities between the facial features of the rider and those seen in contemporaneous portraits of the king, and by referring to a brief comment found in a treatise by the sixteenth-century Milanese painter and art writer Gian Paolo Lomazzo, which mentioned a small horse statuette, derived from Leonardo's hand, belonging to the collection of Leone Leoni. Aggházy's assertion, resting on a hypothetical commission for a statue, on literary sources with unverifiable origins, and on some convoluted and abstruse iconographic arguments, was greeted with polite restraint. Sometimes, of course, words that go unspoken are the most powerful words of all. Although there can be no doubt that Aggházy's efforts were aimed at having the Budapest Horse and Rider accepted as an autograph work by Leonardo, her high-handed attribution unwittingly contributed to the gradual reduction of interest in the small bronze. Common wisdom states that an inconsistent theory is useless since it entails everything and hence explains nothing, but the repercussions of Aggházy's conjectures about the statue of the French king turned out to have surprising staying power. In January 1517 Leonardo worked on plans for a canal in the Loire valley, and in May 1517 he settled for good in Amboise. How he spent the last two years of his life there is clouded in mystery, and the only surviving documents from this period are a few drawings and some notes. The same cloud hangs over his late equestrian designs, with the different style and technique of the late drawings associated with the Trivulzio Monument raising doubts among a number of scholars. The drawings in question include the two sheets from Windsor on display at the present exhibition, in which the pose of the horse bears the closest resemblance of all Leonardo's drawings to the Budapest Horse and Rider (figs 11 and 12). When Carlo Pedretti, rightly the most celebrated Leonardo expert of the twentieth century, re-examined the drawings in the collection at Windsor four decades after Clark's catalogue was published, he expressed his concerns about the dating of the two sheets. The drawing featuring both the more classical pacing horse and the more dynamic rearing horse was already deemed by Clark to be "very late"; Pedretti went even further, claiming that all the signs indicated that it dated from Leonardo's time in France (fig. 11). As for the other study sketch, made around the same time, Pedretti opined that it was more likely to have been executed for a wax model of a small bronze than for a monumental equestrian statue (fig. 12). When Pedretti made his assertions, the dominant view was that Leonardo completely gave up designing the Trivulzio Monument after leaving Milan in 1517; moreover, there was no indication at the time that the master had ever embarked on any other equestrian statue. Consequently, no explanation for the apparent contradiction in the style of the drawings was forthcoming. This situation lasted until 1992, when the decades-long process of conservation and technical research on the Leonardo drawings in the collection at Windsor was finally concluded and experts caught glimpses of numerous new French watermarks among the sheets, including on equestrian studies that had earlier been held to be drawings for the Trivulzio Monument. Before the nineteenth century, paper was made by hand: the pulp was spread onto a mould, pressed together, and slowly dried. In order to distinguish between the products of different paper mills, a simple emblem made of thin wire was placed onto the mould, which left a slight indentation in the thickness of the paper. When the sheet was held up to the light, the faint outline of the watermark could be made out, and this functioned as a kind of trademark for the paper manufacturers. Paper mills strove to use distinctive watermarks, and they changed from region to region, so when a watermark is found on a sheet of paper, its place of origin can be determined with varying degrees of certainty. Furthermore, when a watermark is found on a sheet of paper that was used for a purpose that can be dated precisely, such as archive documents, printed books or engravings, other sheets with a similar watermark can be dated to more or less the same time. Watermarks have therefore long played a key role in determining when and where a given drawing was made. In 1996, the French watermarks found on the Windsor sheets helped Martin Clayton, Head of Prints and Drawings at the Royal Collection Trust, to solve the long-lasting riddle surrounding the style of Leonardo's late equestrian drawings. Like Pedretti before him, Clayton came to the conclusion that some of the drawings executed in a different style were later than the Trivulzio Monument, and must have been made while Leonardo was in France (figs 11 and 12). Now, however, the stylistically based opinion could be corroborated by the recently discovered watermarks. At this point, Clayton revived Aggházy's hypothesis about the equestrian monument that Leonardo was alleged to have begun working on in France. While conceding that there was no documentary evidence in support of the claim that Leonardo may have designed an equestrian statue for Francis I, meaning that the whole idea was pure speculation, he was nevertheless inclined to consider the attribution of the Budapest Horse and Rider from a stylistic point of view, given its comparative similarity to the newly re-dated drawings (fig. 12). Thus three decades later, Aggházy's widely discarded conjecture, albeit indirectly, once more rejoined the discourse on Leonardo's equestrian statues.

Leonardo's name has been both a blessing and a curse in the life of the Budapest Horse and Rider. On the one hand it has guaranteed continuous interest in the small bronze, but on the other it has locked dialogue about the piece into a kind of quarantine. While many rejected Leonardo's authorship from the outset, and even regarded the small bronze as a cast from a later age, for a long time no other name was put forward in connection with the work. Anthony Radcliffe, in his 1966 monograph on Renaissance small bronzes, was the first to tentatively attribute the Budapest Horse and Rider to Giovanni Francesco Rustici. Leonardo was closely connected to Rustici, who was a generation younger than the great master. When Leonardo was working on the Battle of Anghiari, they not only lived in the same house, but—according to Vasari—they also collaborated on the creation of the three monumental bronze figures of Saint John the Baptist Preaching to a Levite and a Pharisee towering above the northern gate of the Baptistery in Florence. It should be borne in mind that when Leonardo left Florence permanently in 1508, the work was still only in its initial stage: the terracotta figures were made in autumn 1509, and they were cast in bronze in summer 1511. Nevertheless, it has been generally concluded that the œuvre of Rustici is the best echo of Leonardo's lost sculpting activity (fig. 13). Leonardo's Battle of Anghiari was one of the most important artistic undertakings of the age. Though it was never completed, the unfinished fresco was a source of inspiration to generations of young painters and sculptors. Rustici was one of those ineluctably drawn under its spell: two terracotta sculptures by him, which are now in the Bargello in Florence and in the Louvre in Paris, were inspired by Leonardo's battle scene; it was for this reason that, for a brief period at the start of the last century, both works were attributed to Leonardo (fig. 14). Radcliffe's proposal was taken up again in the early 1990s by Pietro C. Marani, the noted Leonardo expert from Milan. In his view, Leonardo continued his designs for the Trivulzio Monument even after his move to Amboise; at the end of the 1530s, Rustici may have made use of these sketches when designing his equestrian statue for King Francis I. Marani based his supposition on a late drawing in which Leonardo, alongside numerous studies of horses and lions, sketched out variations of a horse and rider fighting a dragon, viewed from different angles (fig. 15). As Trivulzio was made a Knight of the Order of Saint George in 1504 by Emperor Maximilian I, it is possible—in Marani's view—that Leonardo's intention for the equestrian statue of the condottiere was to portray him in the guise of the heroic saint vanquishing the fire-breathing monster. Rustici's equestrian statue has not been handed down to us, but unlike the pure speculation of Leonardo's hypothetical French monument, it is mentioned in numerous contemporary sources, and what is more, it would appear that the younger sculptor not only borrowed Leonardo's drawings, but was also very familiar with his notes on bronze casting. The attribution of the Budapest small bronze to Rustici was accepted by few scholars, but the mere appearance on the scene of a new potential candidate played a great part in expanding the discourse on the Horse and Rider to a wider context than ever before.

Problems with an attribution begin when our judgment is based on stylistic arguments alone, without further evidence to back them up. As connoisseurship is rarely considered adequately convincing on its own, we often turn to scientific technologies in order to support assertions based on the observation of the naked eye. The increasingly sophisticated methods of technical examinations have in many cases contributed to our knowledge of the circumstances under which a work was created and about workshop practices at the time. The notion that science is infallible, however, is rooted in the same philosophical soil as the positivism from which modern connoisseurship sprang; it declares that science is the only source of knowledge. The ever more baleful reputation of the Budapest Horse and Rider reached pit bottom around the turn of the millennium. This was mostly to do with the fact that a close variant of the small bronze, the Rearing Horse in The Metropolitan Museum of Art in New York, was shown to have been cast no earlier than the nineteenth century. The revelations concerning the New York small bronze soon helped to cast doubt on the authenticity of the Budapest Horse and Rider; despite the fact that no similar technical investigations were carried out on it at the time, the conviction that this was also a modern cast continued to spread inexorably. In 2009, therefore, tense anticipation abounded when the Budapest small bronze was transported to the National Gallery of Art in Washington DC for detailed technical examinations. Forty years after Aggházy's exhibition, the three works—the Budapest Horse and Rider and the rearing horses from New York and Limerick—could once again be compared with one another in a single setting. They were joined this time by a fourth statuette, the Cowering Warrior from the Museo Poldi Pezzoli in Milan, a small bronze figure, recently regarded as originating from Leonardo's workshop (fig. 16). When a small bronze is subjected to technical examination, it is customary to analyse the casting technique, the composition of the alloy, and the details of the surface. The results are then compared with the small bronzes that are believed to be by the sculptor's own hand, and if the similarity is high enough, and the style also matches, then this is regarded as sufficient verification that the works in question were produced by the same master. In the present case, however, the circumstances were somewhat extraordinary, for there are no known autograph small bronzes by Leonardo still in existence, nor indeed are there any sculptures by him in any other material. The question which Shelley Sturman and her colleagues in Washington wanted to answer was not whether the Budapest small bronze was a work by Leonardo, but whether—on the basis of casting technique and alloy composition—it was possible to determine that the statuette might be contemporaneous with Leonardo. All the small bronzes examined in Washington at this time, including the one from Budapest, were produced using lost-wax casting. In this process, a wax model of the statue is formed around a core made of plaster or clay; the model is then invested with a mould made of heat-resistant material, likewise usually plaster or clay. During casting, the wax melts and flows away, and its place is taken by molten bronze. When the cast has cooled down, the outer covering is removed, traces of casting and any flaws are repaired, and the surface of the bronze is often chased to bring out the contours. Finally, the statue is coated with a dark patina, which not only protects the surface of the bronze, but also significantly alters the appearance of the finished work. In the case of the Budapest small bronze, however, very little attention was paid to the work's final finish, and several tiny holes are visible in the surface of both the horse and the rider. Such holes arise due to the iron nails, known as core pins, that fix the inner core of the form to the outer coating of wax; the core pins ensure that the inner and outer form do not move in relation to one another. When the core pins are removed, the holes are usually covered with metal plaques, and during the process of patination the joins are completely concealed. This was not carried out on the Budapest small bronze, so the holes caused by the core pins are still intact. Finally, the statuette was given an artificial green patina, which was rare in the Renaissance, and which greatly contributed to the nineteenth-century belief that it was a work from antiquity. The New York statuette was cast using a similar process to the Budapest Horse and Rider, but the surface treatment of the bronze was completely different: more attention was paid to finishing the surface and making the core pin holes disappear. The Rearing Horse from The Hunt Museum, meanwhile, is completely different from both the Budapest and the New York small bronzes in terms of appearance and working; it is also a few centimetres smaller. Traditional X-ray imaging allows the method of casting to be deduced, while the results of X-ray fluorescence spectroscopy can reveal the composition of the bronze alloy. Although X-ray images of the Budapest small bronze had already been taken in the past, the technology had improved to such a level of precision in the intervening period that even the tiniest traits of the casting could now be discerned. This revealed that the Budapest small bronze was cast using the indirect technique. In this case, the sculptor's model was not used directly for casting, but a temporary copy—an intermodel—was made from it; molten wax was then poured over it and the inner cavity was filled with a heatproof material, in this case gypsum plaster. The advantage of the indirect method is that the artist's model is not destroyed during the casting process, so it can be used several times. The simple fact that this technique was common in the Renaissance is not a decisive factor when it comes to dating the Budapest Horse and Rider. It does, however, indicate strongly that every effort was made during the casting process to preserve the model of the small bronze. The walls of the New York small bronze are thinner and more even, and follow the external form of the statue more precisely, which in itself suggests that it was cast later than the sixteenth century. What is more, X-ray images clearly show that two of the metal plaques covering the core pin holes terminate in modern screw threads. The horse's tail was not part of the casting but a later addition—as was the case with the Budapest horse—, but the welding material used for the New York small bronze is of a composition with a very high melting point, implying a soldering technique that has only been around since the nineteenth century. The technical execution of the small bronze from The Hunt Museum, meanwhile, is so different from its Budapest and New York counterparts that it cannot possibly have originated from the same workshop. The results of the X-ray fluorescence spectroscopy proved decisive, for they not only led to conclusions about the technical details of preparation and casting, but also about the age of the bronze, as revealed by its composition. The National Gallery of Art has built up a substantial bronze database over the years, so the expert team had a rich set of reference data at their fingertips. Compared with small bronzes made in the Italian Renaissance, the alloy of the Budapest Horse and Rider falls well within the expected range, so there is no reason to assume that it was made at a later time. On the other hand, the New York and Limerick small bronzes are made of a far purer composition, which is typical of modern casts. The Cowering Warrior from Milan is linked to the Budapest Horse and Rider by only the thinnest of threads (fig. 16). This small bronze is sometimes associated with an early engraving which reproduces some of the initial sketches Leonardo made for the Sforza Monument [3]. As a way of providing stability to the group statue, Leonardo initially took inspiration from classical prototypes by placing a warrior, protecting himself with his shield, crouching beneath the raised front legs of the rearing horse (fig. 7). This figure was cast in bronze by the master of the Milanese statuette. Although the alloy used for the Milan small bronze is closer in composition to that of the New York Rearing Horse than to the Budapest Horse and Rider, Sturman and her colleagues concluded that the Cowering Warrior was made in Leonardo's workshop in the 1490s. The technical examinations finally put to rest all the doubts that had surfaced concerning the Renaissance origins of the Budapest small bronze. Whereas the Rearing Horse statuettes now held in New York and Limerick were shown to be no older than the nineteenth century, there is no reason to believe that the Budapest Horse and Rider was made later than the sixteenth century. What is more, comparative images made using a 3-D laser scanner indicate that the New York small bronze was cast directly from the Budapest horse. The most important finding made by Sturman and her colleagues, namely that "no scientific data was discovered that disputes an early casting date" of the Budapest small bronze, was expressed in the dry, reserved style of natural science. Their further assumption, however, that the horse and the rider were not modelled together, and that the warrior was only added to the equine immediately before the statuette was cast, rested on less precise foundations. The idea that the horse and rider were not originally together dates back to the beginning of the twentieth century, when it was widely commented that the rider was, by classical standards, disproportionately small in comparison with the horse. It was not by chance that the first critic to point out this observation was the classical archaeologist Antal Hekler. Sturman and her team argued that the lack of a saddle and the unusual method of removing the excess clay from the inside of the bronze horse after casting—not through a slit cut in the horse's back or belly, but through its tail—demonstrated that the horse was originally designed without a rider. Moreover, in The State Hermitage Museum in Saint Petersburg, there is a close variant of the rider from the Budapest small bronze, which was later added to a horse originating from a different workshop [4]. The above facts would all seem to give sufficient cause to question whether the horse and rider making up the Budapest small bronze actually belong together, but this supposition is disproven by the results of the technical examinations: "the use of the same alloy for both horse and rider indicates that they were cast at the same time". This does not of course rule out the possibility that the two models were made separately, but it is important to stress that this particular suspicion arose not from the technical examinations but from the apparently unusual proportions of the small bronze. Decisions on what Leonardo would have regarded as acceptable, classical proportions depend exclusively on our aesthetic judgment of the work. However, by placing their aesthetic deliberations within the context of their technical results, their hypothesis was disguised as a fact supported by scientific results. Consequently, the latest literature seems to regard their theory as having been confirmed and authenticated by technical examinations. We should proceed with caution, however, for while scientific methods only exert a limited influence on our aesthetic considerations, the latter have the potential to determine how the results of technical examinations are interpreted.

Names often play a more important role in art history than we might like. Furthermore, as in the case of the Budapest Horse and Rider, our ability to provide a resolute and satisfactory answer to the question of the authorship of a work is extremely limited. In the mid-nineteenth century, when art history entered the faculties of universities as a new branch of study, the aim was to eliminate these uncertainties. It was therefore natural for art historians to borrow terms from more established disciplines, in particular the natural sciences, as a way of augmenting their own prestige. This was the golden age of "art history without names", which was permeated with unbridled optimism that the development of artistic forms could be described in similarly precise terms to those used by the natural sciences, thus opening the way towards a historical explanation for why and how changes in style take place. Although art history started out with every intention of steering away from the question of artistic authorship, the names kept seeping back into the equation, and sooner or later something had to be done about them. We have seen how Morelli attempted to validate conclusions drawn using the more or less elusive criteria of conventional connoisseurship with a precisely defined terminology of stylistic analysis. It soon became apparent that the majority of attributions still remained unverifiable even using the precise terms of the natural sciences, exposing Morelli's "scientific connoisseurship" as mere fiction, so the one-sided relationship between art history and scientific thinking persisted as a multiple source of confusion. The simple and widespread view of the scientific method, which states that impartial and unbiased observations can lead us to generally valid statements, often seems to be the most pertinent solution in the natural sciences. Nature is dominated by large-scale regularity, so this method is not only suitable as a means of explaining the phenomena that come under investigation; as the laws of nature need to be provable, that is, consistently repeatable, no matter how many times they are tested, they can be used as a more or less reliable guide to predicting what would happen in the future, given identical circumstances. Art historians often incongruously apply scientific models to sets of circumstances completely unrelated to science. Like it or not, we have a longstanding, unrequited love for empiricism. According to empiricists, human disciplines, such as art history, which fail to predict the natural behaviour of social phenomena, are merely interpretative. In contrast to the natural sciences, in which factual description is considered admirably objective from an empirical point of view, the differing opinions in human disciplines are judged to be subjective and its scholarship is relegated to the status of an accumulation of factual or admittedly subjective descriptions. There is little to be gained, therefore, in taking art history to task for being subjective. Art history is nothing more than an artificial, socially, and culturally determined, constantly changing intellectual construct of our own making. Art existed in innumerable cultures and periods without anyone attempting to piece together its history, and until the mid-sixteenth century, Europeans also managed perfectly well without any construct of this kind. Further muddying the waters is the fact that the definition of art changes from time to time, so the methods we use in order to interpret artworks are only valid inasmuch as they are compatible with our prevailing ideas about art and art history. There is another way of thinking about science which differs drastically from the positivist approach. According to the American philosopher Thomas Kuhn, science does not move forward in a straight line interspersed with proofs and refutations, but occasionally undergoes a makeover, a fundamental rebirth. At such times, old theories are replaced with new ones, and radically new approaches, hitherto regarded as invalid, are suddenly made possible. Within this process, which he describes as a series of "paradigm shifts", the frames of interpretation are set not by objective criteria, but solely by consensus among the scientific community. The radically new "paradigms" challenge earlier scientific practice, and at the same time they propel new problems and interpretations up to the surface. Kuhn's theory of the philosophy of science has become a metaphor for a more dynamic, pluralistic approach to science. If we accept that the interpretation of artworks is not the kind of task for which a single solution, once found, ends the matter forever, then it follows, quite self-evidently, that not a single attribution can ever be totally definitive and unchallengeable. Just like a chess piece, whose strength varies continuously as the game develops, the validity of an attribution is also dependent on the past and present theories that surround it, as its own internal logic. When a new suggestion is put forward and turns out to be worth consideration, it may have the capacity to fundamentally rearrange our existing views of the artwork and the artist. Ideas that once languished by the wayside can suddenly be recast in a new, significant role, while long-held hypotheses can just as suddenly be shuffled off into the sidelines. Whilst I concede that the language of connoisseurship has a tendency at times to be irritatingly self-confident and even downright manipulative, coming to conclusions that are in the main highly debatable, I am not in favour of its banishment from art history. There is no single route to the truth, so it would be wrong to question the right of any method to exist; indeed, the more theories there are competing against one another, the more colourful and fruitful the discourse may become. As demonstrated by this brief history of the Budapest Horse and Rider, an attribution may be of value not only by gaining widespread acceptance, but also by leading us to new realisations, even though the attribution on its own may be less than persuasive.

1 Attributed to Béni Ferenczy, after Leonardo da Vinci, Rearing Horse, early 20th century, bronze, height: 19.7 cm, New York, Metropolitan Museum of Art, inv. no. 25.74 (online at www.metmuseum.org); after Leonardo da Vinci, Rearing Horse, early 20th century, bronze, height: 20.5 cm, Limerick, The Hunt Museum, inv. no. MG 037 (online at www.huntmuseum.com).arrow_upward

2 Leonardo da Vinci, Horses, Machinery and an Angel, c. 1503–4, black chalk and pen and ink on paper, 200 × 283 mm, Windsor, Royal Collection, inv. no. RL 12328r (online at www.royalcollection.org.uk).arrow_upward

3 After Leonardo da Vinci, Four Studies for an Equestrian Statue, engraving, 217 × 160 mm, London, British Museum, inv. no. 1895,0617.182 (online at www.britishmuseum.org).arrow_upward

4 Circle of Leone Leoni, Rider, early 16th century, bronze, height: 13.5 cm; Paduan Master, _Horse_,early 16th century, bronze, height: 22.3cm, Saint Petersburg, The State Hermitage Museum, inv. no. 2234 (online at www.hermitagemuseum.org).arrow_upward

Source: Zoltán Kárpáti, Leonardo da Vinci and the Budapest Horse and Rider, exhibition catalogue, Budapest, Museum of Fine Arts 2018