Concerning the Chronology of Cimabue ’ s Oeuvre and the Origin of Pictorial Depth in Italian Painting of the Later Middle Ages

The purpose of this study is to examine the reawakening of spatial realism in the art of central Italy during the latter part of the dugento and the early years of the trecento. It will focus particularly on the oeuvre of Cimabue, generally considered the most outstanding painter Florence produced in the generation preceding Giotto. Articulating developmental artistic patterns and their sources in this crucial period, when Italian painters gradually became aware of three-dimensional form and space, and the intricacies of nature and human appearance and behaviour, is a diffi cult task. Overriding fundamental differences in time and culture, the aesthetic power of some masterpieces then produced, including Cimabue’s Santa Trinità Madonna (Figure 1) and Giotto’s Ognissanti Madonna (Figure 2), seems as strong today as it ever was. Retrospectively considered, their appeal seems substantially rooted in the particular artists’ resolutions integrating fundamental aspects of the fl at and symbolic, spiritually oriented, pictorial vision of the Middle Ages, with an emergent realism which could not be denied. This emergent realism, as it applied to the representation of three-dimensional form and pictorial depth, progressed gradually and painfully from within the resisting force of the former. As will be examined, in the works of specifi c painters, especially Cimabue, this progress assumed the form of successive steps which, it is here assumed, can be reasonably defi ned, thus assisting in the establishment of a progressive chronological sequence. According to the criterion of pictorial space, the evolution of central Italian painting covering this period offers convenient chronological limits. Around mid century and before, the imagery highly fl at and reduced, an interest in pictorial depth and three-dimensional form virtually wholly absent, fi gures and objects are essentially pleated against the picture plane. This can be seen in the murals of the Life of Saint Sylvester in the church of the Santi Quattro Coronati in Rome, dated into the later fourties; Bonaventura Berlinghieri’s Saint Francis altarpiece in Pescia painted in 1235; and the mid-century Acton Collection Madonna by the Bigallo Master (Figure 3); etc. However, a half century later, in his Ognissanti Madonna (Figure 2), usually dated around 1310, Giotto constructs a substantial coherent depth reaching from the picture plane into a middle ground, including steps rising from the base line toward a platform which supports the Virgin’s substantial throne. Although beyond the throne the usual fl at Concerning the Chronology of Cimabue’s Oeuvre and the Origin of Pictorial Depth in Italian Painting of the Later Middle Ages


Background
The purpose of this study is to examine the reawakening of spatial realism in the art of central Italy during the latter part of the dugento and the early years of the trecento. It will focus particularly on the oeuvre of Cimabue, generally considered the most outstanding painter Florence produced in the generation preceding Giotto.
Articulating developmental artistic patterns and their sources in this crucial period, when Italian painters gradually became aware of three-dimensional form and space, and the intricacies of nature and human appearance and behaviour, is a diffi cult task. Overriding fundamental differences in time and culture, the aesthetic power of some masterpieces then produced, including Cimabue's Santa Trinità Madonna (Figure 1) and Giotto's Ognissanti Madonna (Figure 2), seems as strong today as it ever was.
Retrospectively considered, their appeal seems substantially rooted in the particular artists' resolutions integrating fundamental aspects of the fl at and symbolic, spiritually oriented, pictorial vision of the Middle Ages, with an emergent realism which could not be denied.
This emergent realism, as it applied to the representation of three-dimensional form and pictorial depth, progressed gradually and painfully from within the resisting force of the former. As will be examined, in the works of specifi c painters, especially Cimabue, this progress assumed the form of successive steps which, it is here assumed, can be reasonably defi ned, thus assisting in the establishment of a progressive chronological sequence.
According to the criterion of pictorial space, the evolution of central Italian painting covering this period offers convenient chronological limits. Around mid century and before, the imagery highly fl at and reduced, an interest in pictorial depth and three-dimensional form virtually wholly absent, fi gures and objects are essentially pleated against the picture plane. This can be seen in the murals of the Life of Saint Sylvester in the church of the Santi Quattro Coronati in Rome, dated into the later fourties; Bonaventura Berling-hieri's Saint Francis altarpiece in Pescia painted in 1235; and the mid-century Acton Collection Madonna by the Bigallo Master ( Figure 3); etc.
However, a half century later, in his Ognissanti Madonna (Figure 2), usually dated around 1310, Giotto constructs a substantial coherent depth reaching from the picture plane into a middle ground, including steps rising from the base line toward a platform which supports the Virgin's substantial throne. Although beyond the throne the usual fl at medieval golden background takes over, a vast difference in spatial thinking separates Giotto's Madonna from the one by the Bigallo Master. This comparison can be used for establishing the approximate limits within which the paintings here considered will fi nd their spatial niches.
In dealing with this revolutionary change in spatial parlance caution is advised. It has to be viewed from within the dialectic of its own late medieval period for its meaning to be properly understood. Here the later quattrocento developments in rational perspective by Brunelleschi and Alberti, including the concept of spatial infi nity, simply do not apply.
One must realize that certain reduced and "distorted" pictorial spatial devices, reverting to the Late Antique, were used throughout the Middle Ages, in Italy well into the later dugento. The most prominent of these, preferred in Byzantine art, is reversed perspective. 1 Contrary to actual visual experience, it applies to objects which get larger as they recede into distance. This convention conveys the idea that the objects so presented, and the fi gures connected to them, should project forward, in the direction of the beholder. It is widely used in the rendering of objects: thrones, chairs, tables and platforms supporting fi gures. Here only that part of the object closest to the observer really matters. Connected to the hierarchic medieval presentation of signifi cant beings, be they personifi cations, emperors, saints or Divinity, reverse perspective effectively converted spatial rendering, however abbreviated and arbitrary, toward serving spiritual ends.
Consider a Byzantine icon of the Madonna Enthroned at Mount Sinai (Figure 4), usually dated into the twelfth century, 2 on which her feet rest on two platforms, one set on top of the other. Both throne and platforms are spotted against a spatially neutral golden background. Although the throne itself is fl at, here the use of reverse perspective in the rendering of these two platforms serves to endow the Madonna and Child by association with a degree of thr ee-dime nsional substance, thus contributing to their hierarchic, that is, spiritual signifi cance. The same visual symbolism applies to Byzantine representations of signifi cant beings either standing or seated on such platforms. 3 Reverse perspective can be applied subtly to objects of great refi nement, such as Christ's ivory throne on the Harbaville Triptych in the Musée du Louvre. Reverse perspective is widely used in later dugento Italian painting. It is still found, subtly presented, in Duccio's Rucellai Madonna ( Figure 5), dating close after 1285. 4 However, at a certain point, as in Giotto's Ognissanti Madonna, usually placed around 1310, it is suppressed, its obvious diametric opposition to real visual experience causing its dismissal from artistic usage. 5 Medieval fi gures enthroned frequently offer another "distorted" spatial convention. It applies to the orientation of the foot stool (suppedaneum), which, sharply upward tilted, diverges often radically from that of the throne. This convention, already evident on a sixth century Byzantine ivory Madonna relief in Berlin, 6  . The small monk Abbas kneeling before her is spotted against the golden background. 4 Duccio's Rucellai Madonna will be treated in a separate study to follow. 5 The rejection of reverse perspective from representations of the Madonna enthroned around the beginning of the fourteenth century in central Italian painting is connected to the dismissal of her oblique throne. From that time onward the throne seen directly from the front, already appearing in Cimabue's Santa Trinità Madonna, to be discussed, takes over. See also J. Polzer, The "Byzantine" Kahn and Mellon Madonnas: Concerning their Chronology, Place of origin, and method of Analysis, Arte Cristiana XC, 813 (2002), 410. 6 Illustrated in Mother of God (n. 2 supra), 29, pl. 12. 7 The recently restored face of Christ closely resembles that of God the Father in the mural of the Creation of Adam and Eve painted in the bay next to the crossing on the upper tier of the northern nave wall in the upper church of San Francesco at Assisi. As is well known, this mural copies in large measure the early Christian mural of the same subject matter once in the basilica of Saint Paul fl m in Rome. Jacopo Torriti has been considered its Throughout the Middle Ages the exceptional art of Byzantium was bound to infl uence Western Europe, and especially Italy, given its proximity to the Near East. Its infl uence increased signifi cantly during the political disintegration of the Byzantine Empire following the Frankish conquest of Constantinople in 1204. The particular conservative spiritual refi nement of Byzantine art was diffi cult to resist. However, as will be examined, in the course of the later thirteenth century its infl uence in central Italy gradually diminished under pressure from the emergent wave of realism challenging its conservative medieval stance. This is the period when Cimabue held a leading position in central Italian art. Unquestionably, he was profoundly infl uenced by artistic conventions reverting to Byzantium. As will be seen, his artistic evolution demonstrates the diffi culties he faced in reconciling the conceptual and symbolic parlance of the Byzantine and local medieval artistic traditions, in which he was weaned, with the early stirrings of the new realism he partly also investigated.
The specifi c analytical criterion here used for measuring the later Italian medieval painter's awareness of pictorial depth simply involves the precise placement of an object's feet, should they be represented, on the ground. Let us assume that the artist places the beholder directly in front of a seated person, as is usually the case in representations of Christ or the Madonna enthroned. According to the beholder's angle of vision, if both the front and rear feet of a painted throne, or chair, come to rest at the same level, they will appear fl at, as well as the ground around them, conforming more or less to a vertical plane. However, if they connect at differing levels, the rear feet duly located above those in front, then some awareness of the object's three-dimensionality is introduced, extending to a ground receding into depth, even if other clues to its recession may not be present.

Concerning the Chronology of Cimabue's Oeuvre
Our understanding of Cimabue's art is based on a limited group of paintings sharing a distinctive dramatic fi gure style generally considered autograph. With one exception, a mosaic produced close to the time of his death, no work by him is precisely named and dated. Vital aspects of Cimabue's artistic progress are still under discussion. It is here assumed, as above indicated, that the stepped increase in his awareness of pictorial depth, and that of concrete form dependent on the latter, contribute to the understanding of his artistic evolution, and its place in contemporary central Italian art.
A document places Cimabue as a witness in Rome in 1272, by which time he was certainly an established master. 8 His only extant documented and dated work is the mosaic of Saint John the Evangelist of 1302 located in the eastern apse of Pisa Cathedral (Figure 7). 9 He died shortly after. That is all the documentary evidence remaining regarding both his life and work, beside what can be gleaned from his paintings. Assisted by Renaissance historiography and his distinctive dramatic fi gure style, modern scholarship is generally agreed on a nucleus of works constituting his autograph oeuvre. Most prominent is his monumental mural decoration of the eastern choir, crossing vault and much of the transept in the upper church of San Francesco at Assisi, assigned to him virtually unanimously since the end of the nineteenth century, although not attested by documentation from the time of its production. 10 No archival or documentary evidence informs directly on its chronology. Recent interpretation yields essentially two views. Majority opinion connects these murals to the papacy of Nicholas III covering the years 1277-1280. 11 However, recently it has also been linked to that of Nicholas IV during 1288-1292. 12 Here one has to consider the sequential position of Cimabue's murals within the interior decoration of the upper church, which extended from about the church's consecration by Innocent IV on May 23, 1253, to circa 1300, since by 1307-1308 its latest murals were already copied. 13 It began with the stained glass windows of the choir by a northern Gothic shop (Figure 8) whose sophisticated supple rhythmic  fi gure style stands in sharp contrast to anything then available on Italian soil. 14 A northern master was also active in the mural decoration of the northern transept. 15 The importation of the northern stained glass masters, considered the most "modern" available, would have refl ected the papacy's promotion of the sanctuary's international signifi cance. Some time after they left Cimabue took over, painting the interior of the upper choir, the remainder of the transept, and the crossing vault. 16 Considering the drastic differences in style, one would like to know what went on in the respective patrons' minds, whether representing the papal curia in Rome or the Franciscan community at Assisi. Were they motivated by questions of cost, or aesthetic preference? Compared to the sophisticated fi gure style of the northern stained glass masters, Cimabue's seems downright archaic. However, his byzantinising parlance and his sense for drama would have been familiar to the Franciscan community at Assisi. It strongly recalled the style of Giunta Pisano, probably central Italy's most prominent painter of the middle of the century, who had painted a cross at Assisi for friar Elias as early as 1236. This painted cross no longer exists, but a cross he signed is still in the sanctuary of S. Maria degli Angeli below Assisi, and another in the church of San Domenico in Bologna (Figure 9). 17 Although we hear no more of Giunta after the mid-fi fties, in Umbria the impact of his dramatic version of the crucifi ed Christ lasts into the seventies of the dugento as evident from the monumental cross by the so-called Saint Francis Master, which, significantly, is dated 1272. 18 The mural decoration of the life of Christ and Saint Francis in the lower nave of the church of San Francesco at Assisi has been assigned to the same master. His type of painted cross had a wider geographic currency, since it closely resembles the one in the monastic library in the church of San Francesco in Bologna, etc. 19 Of course, here there is much room left open for speculation. What counts is this: that Cimabue's artistic origin is rooted in an artistic current grounded in central Italian painting since the early part of the dugento. The possibility that an element of local familiarity contributed to his appointment for decorating the most sacred portion of the upper church must be taken seriously.
All this preceded the exceptional artistic ferment occurring in Rome and Assisi during the last decades of the du- The very intensity of this later dugento Roman artistic revival, involving a renewed awareness of late antique mural painting conventions, contributed to a profound questioning of established medieval and Byzantine artistic thought. Here the revival of sculpture headed in Rome by Arnolfo also played its part. 23 This questioning, and its consequences, is imprinted in the evolving decoration of the nave and vaulting of the upper church at Assisi, culminating in the murals of the Isaac Master and the later murals of the cycle of the Life of Saint Francis lining the lower nave walls. The masters, some surely themselves Roman, were fully aware of what was happening in Rome. Their artistic vision came to differ categorically from Cimabue's in the same church. In essence, the latter connects with earlier Italian painting. It follows that the later Cimabue's murals in Assisi are placed, say around 1290, as has been recently proposed, 24 the more retarditaire they would appear. city. 26 All this happened after King Charles of Naples left the Roman senate to which he had been previously appointed. The arguments recently advanced for placing Cimabue's mural decoration in the upper church, differently, into the papacy of Nicholas IV around 1290, make little sense. This would mean that the bulk of the mural decoration of the upper church would have to be compressed within about one decade or less. Compared to the most "progressive" murals of the Life of Saint Francis cycle, dating around the later nineties of the dugento, Cimabue's murals are obviously extremely conservative. This includes the signifi cant changes they respectively reveal in the treatment of pictorial depth. All this will be considered in some detail.
Would a Cimabue have been invited to work in Assisi so late in the dugento? Commenting on the uncertainties of fortune, in a well-known passage in his Purgatory, Dante wrote that Giotto's fame had eclipsed Cimabue's. 27  the Purgatory by the middle of the second decade of the trecento, placing his journey through the afterlife, however, into the year 1300. Dante surely knew Giotto in person, and he would also have known Cimabue. Given the exceptional contemporary speed of artistic change, around 1280 his murals in the upper church might still have had a leading artistic role, but hardly ten years after! In essence, the stylistic direction endorsed by the later masters of the Saint Francis cycle represented a clear rejection of Cimabue's style, and especially its byzantinising baggage. This seems confi rmed, in addition to the latest murals in the upper church, by Giotto's earliest certain dated work: the mural decoration of the Arena Chapel completed around 1305, and this would correspond to Dante's fi rst hand verdict! Substantial evidence challenging the dating of Cimabue's activity in the upper church around 1290 is given by two panel paintings closely resembling his Crucifi xion in the upper southern transept, one of the most dramatic murals of the period (Figure 10). 28 One is the Crucifi xion located on the left wing of a tabernacle in the National Gallery of Scotland in Edinburgh ( Figure 11) whose central panel offers the earliest known image of the death of Ephraim. Consider the similarities. Both Crucifi xions share the same pronounced S-curve of Christ's dead body, his head sunken deep into his right shoulder. They also share Christ's loin cloth windswept sharply to the right; the identical poses of the Virgin and Saint John holding hands as they stand beside the cross; John's gesture of presenting the open palm of his left hand to the beholder, thus in-forming him of his adoption of the Virgin; the angels catching the blood fl owing from Christ's wounds in their vessels; and the centurion who gestures sharply toward Christ as he stands in front of a group of soldiers. Here the painter even copied Cimabue's arbitrary superposed placement of the soldiers' tilted feet! The painter of the panel did make one signifi cant change. He substituted Stephaton, holding pail and staff topped with a sponge, for Cimabue's soldier clasping his shield and lance standing directly in back of the centurion. Here Cimabue's direct infl uence seems persuasive. Gertrud Coor-Achenbach placed the Edinburgh tabernacle close to Guido da Siena, dating it around 1285, 29 and more recently Miklos Boskovits has identifi ed the painter, according to his reading of a fragmentary inscription on the tabernacle's lower border, as the Florentine Grifo di Tancredi, who appears in a document of 1295. 30 Fig. 10 Assuming that Cimabue's Assisi Crucifi xion belongs from the time of Pope Nicholas IV, is it reasonable to date the Edinburgh tabernacle so late? Its style would place it considerably earlier. A late date might be explained if one were to consider the painter of the tabernacle unusually retarditaire, plausibly a provincial removed from the vital progressive artistic events then occurring in central Italy. On the other hand, the Byzantine image of the funeral of Ephraim at the center of the tabernacle would indicate a wider artistic awareness on the part of the painter and the patron. Such a late date is improbable.
A small Crucifi xion panel in the Pinacoteca Nazionale of Siena ( Figure 12) also closely resembles Cimabue's Assisi Crucifi xion in a number of details. These include the Virgin and Saint John holding hands and bowing their heads toward each other as they stand beside the cross; the centurion pointing toward Christ; and beside the latter the unusual soldier clasping his shield and spear before his body as he moves away from the cross. I should not be surprised if he were to represent the contrite Longinas. 31 Beyond this fi gure also appears the same thoughtful older man pensively stroking his beard. All appear in both paintings. In spite of the comparative smallness of the crucifi ed Christ, and the absence of his wind agitated loin cloth, in the small panel painting the sharp sideward curve of his body is also repeated. That there is a close connection can hardly be doubted! There remains the question of how it is to be interpreted.
In a recent exemplary investigation, Holger Manzke and Barbara John have established that the small Crucifi xion panel in Siena formed part of an elaborate altarpiece made for the high altar of the cathedral of Siena some time after the battle of Montaperti, which occurred in 1260. 32 It consisted of the central image of the half fi gure of the Virgin holding the Christ Child, fl anked by twelve scenes of the life of Christ, six to each side, with a coronation of the Virgin located in the gable. Sawn apart around the middle of the fi fteenth century, its principal image is no less than the Madonna del Voto who receives a cult in the cathedral to this day. The other dispersed members of the altarpiece have found their way into various museums the world over. Although the altarpiece is not dated, its production has been placed into the later sixties of the dugento when the construction of the choir and domed crossing of Siena Cathedral would have been completed, corresponding roughly to the production of Nicola's Siena pulpit made in the years 1265-1268. 33 Barbara John attributes the altarpiece to two masters, one anonymous, named the master of the Madonna del Voto, the other being Guido da Siena. 34 From its extended use in the Madonna del Voto and the Coronation of the Virgin, here chrysography-mania partly prevailed, as it did in many other contemporary Tuscan paintings, refl ecting a phase of strong Byzantine infl uence, as will be discussed. How does all this bear on the chronology of Cimabue's Assisi Crucifi xion? What are the possibilities? Could Cimabue's painting have copied the Madonna del Voto altarpiece Crucifi xion? Given the substantial differences in quality and in monumental and dramatic concept, this option can be excluded. Conversely, if the Sienese Crucifi xion panel copied Cimabue's mural the latter would have to be dated much earlier than has been thought. Of course, this chronology would be in confl ict with the connection of the Orsini arms on the façade of the senatorial palace of Rome in Cimabue's Assisi mural of the Evangelist Mark to the papacy of Nicholas III.
There remains the third possibility of a shared contemporary source. Considering all relevant factors, this may be, tentatively, the most plausible explanation. Nothing is known of Cimabue's artistic activity before the undated Aretine cross, usually considered his earliest known work. In any case, the close connection of these Crucifi xions strongly indicates that Cimabue's Assisi mural belongs well before the papacy of Nicholas IV.
For further evidence regarding the chronology of Cimabue's paintings we turn to his Madonnas. We fi nd that their detailed scrutiny reveals certain characteristics lending themselves to chronological sequencing not yet suffi ciently explored. As indicated, they include the comparative treatment of pictorial depth, as well as others, such as the realistic consideration of pliant cloth.

Cimabue's Santa Trinità Madonna (Figure 1)
The restoration of Cimabue's Santa Trinità Madonna by Alfi o del Sera in the years 1992-1993 has transformed its appearance. Belying its age, it looks virtually new. One keeps in mind that the frame is not original, and was initially considerably wider than it is now. 35  On stylistic grounds the attribution of this large painting to Cimabue, reverting to Renaissance sources, is generally accepted. 36 Recently it has been placed late: into his "fi nal period", some time after his assumed activity at Assisi under the pontifi cate of Nicholas IV. 37 As has been discussed, the latter makes little sense, and, accordingly, the chronology of the Santa Trinità Madonna must be reconsidered.

The basic composition
The monumental throne dominates the total image in frontal view. It consists of the throne proper set on a separate structure resting on arches. No such throne appears elsewhere in late medieval panel painting. Of its supports only the four in front of the base structure are depicted, consisting of composite piers with engaged columns at the front and pilasters at the sides. The side piers connect with the adjacent panel borders so that this lower structure covers the entire width of the pictorial fi eld. In its impressive scale and the articulation of the piers the lower structure gives the impression of architecture, although some of its surface decoration, such as the leaves on the lower panels, suggests relief carved in wood.
The throne proper recedes into distance as it rises from the platform on which it rests. The latter also supports the Virgin's concave foot stool (suppedaneum) consisting of two steps (Figure 13). This foot stool is not connected to the throne since the luxuriously embroidered cloth covering the latter descends between them. Its concave shape is implicitly transmitted to the throne, assumed to curve around the back of the monumental Virgin, thus underscoring her full volume.
The unique treatment of pictorial depth in this painting excludes any indication of a receding ground on which the base structure is set, since only its front supports, shaped as composite piers, are depicted, rising from the base line. These connect with three upward rising curved architectural shapes, which frame the upper bodies of four Old Testament fi gures. These curved shapes consist of two side arches, and a wider central concave form extending into distance. All this is spotted against a fl at golden background. Depth recession begins only with the upper surface of the base structure.
The latter is spatially ambivalent. On the one hand the side arches follow a vertical direction, while the larger central curved shape connects with the receding upper level surface on which the Virgin's throne and foot stool are set.
Here the absence of a receding foreground in the lower part of the image serves symbolic ends: the idea that the New Testament is founded on the Old. Cimabue's Old Testament fi gures set close to the observer prophecy the incarnation of the Divine Child in the Virgin's womb and her intercession for mankind. Beginning with Jeremiah at the left side, the text on his scroll reads: "creavit Dominus novum super terram: femina circumdavit virum" (The Lord has created a new thing on earth: woman who protects man) (Jeremiah 31, 22). Abraham's scroll states: "In semine tuo benedicentur omnes gentes" (In your seed will all peoples be blessed) (Genesis 22, 18), words spoken by the angel after God kept Abraham from sacrifi cing his son as he had been commanded. On King David's scroll is written: "De fructu ventris tui ponam super sedem tuam" (The fruit of your womb I shall set on your throne) (Psalm 131, 11); and, fi nally, on Isaiah's scroll one reads: "Ecce Virgo concipiet et pariet" (Behold a virgin con-ceives and gives birth) (Isaiah 7, 14). It can be assumed that the choice of these Old Testament fi gures and their texts was made, not by the painter, but by the commissioning body. 38 The basic composition of Cimabue's Old Testament half fi gures set within a seeming arcade brings to mind the shape of contemporary dossals offering half-fi gures of saints fl anking the Madonna or Christ, set below an arcade, as in Guido da Siena's Dossal 7 in the Siena Pinacoteca, dated by inscription 1272. In the latter, however, the arches form part of the wooden frame, while Cimabue's "arcade" is painted.
The imitation chrysography on the throne supporting platform consists of a pattern of diagonally disposed pointed lines converging upward in distance. It appears as well, similarly disposed, on all of the other level architectural parts. Accordingly, here reverse perspective is not used, plausibly a "logical" consequence of the adoption of the frontal throne.
As stated, the base structure and the foot stool are separate entities since the respective curves defi ning their concave shapes do not quite agree. At the same time Cimabue integrates the two by making the concave sections of both units darker than the rest.
Obviously, here the entire elaborate architectural formation of the throne and its supporting structure signifi es the Church. One is reminded that the Church emerged from Here the basic theme of Christian history, that the New Testament emerged from the Old, is expressed in architectural form. This scheme is hardly new, since it appears effectively on northern cathedral portals. There one often fi nds Old Testament fi gures: patriarchs, kings, queens and prophets, applied on the jambs located below the New Testament subject matter appearing above in the tympana, as on the royal portal of Chartres Cathedral. Projecting the same idea, the thirteenth century also witnessed the blossoming of the Throne of Solomon theme as the Virgin and Child enthroned, representing the Wisdom of the Church, are identifi ed with King Solomon's Old Testament temple. Close in time to Cimabue's painting, the Throne of Solomon was represented rising high on the sharp gable above the west portal of Strasbourg Cathedral, King Solomon appearing beneath the Virgin and Child enthroned, while far below Old Testament fi gures lined the jambs of the central entrance. It can be concluded that the unique spatial structure of Cimabue's monumental throne depended on its symbolic role.
Cimabue's Virgin is clearly of the Hodegetria type, referring the beholder to the Christ Child who is the teacher holding a small scroll which represents the Word of God. Both Virgin and Child dominate the composition in their extreme hierarchic scale. Four superposed angels symmetrically fl ank each side of the throne, all holding onto the latter. They are tightly fi tted between it and the panel border, their wings partly extending beyond. The rhythmic variations in the symmetry of the angels' poses, tilting their heads one way and the other, serve to enliven the composition. With the exception of the two who are second from the bottom, the angels join the Virgin and the Christ Child in looking directly at the beholder. This contributes strongly to the idea of the image's forward projection toward the latter, here strongly felt. Although the two closest angels, like the others, are placed in back of the throne, their visible leg and foot extend sharply forward so that the toes protrude into space in front of the platforms supporting them. Accordingly, the space occupied by both the angels and throne is compressed. The angels are evenly superposed holding onto the throne's side posts, thus spatially wedded to the latter. The two uppermost angels are placed beyond the back of the throne, as if space were partly reaching into distance. Altogether, pictorial recession, insofar as it applies, involves the throne and the angels occupying the upper portion of the painting. Typical of medieval spatial convention, here the pictorial recession of the throne and the adjoining angels does not exclude the very opposite: the strong impression of forward projection. Beside her huge scale, which contributes to this effect, this is indicated by the position of the Virgin's pointed feet, like those of the front angels extending forward beyond the steps on which they rest.
The description of the painting must include reference to the arbitrary physical connection of the Virgin and her Divine Son. He is plainly suspended in space. Given the low position of the Virgin's left foot, neither her left knee nor thigh could have supported him! This denial of the force of gravity was deliberate. Conversely, in Cimabue's other Madonnas, to be discussed, including the earlier in the Louvre, Mary's left knee is clearly placed directly beneath the Child's body! The suspended position of the Divine Child in the Uffi zi panel simply informs that the incarnate Christ transcends natural law. Used for similar religious ends, the arbitrary elevation of the Christ Child above the Virgin's lap returns, signifi cantly, in Giotto's later Ognissanti Madonna, as well as on Arnolfo's Madonna statue originally serving the façade of the Cathedral of Florence. Accordingly, this particular arbitrary gravity denying convention, extending into the trecento, has no bearing on the establishment of Cimabue's painting's closer chronology. When

Cimabue's other Madonnas
The Louvre Madonna is generally placed earlier than the Santa Trinità Madonna. Their thrones differ in substance and orientation. The Louvre throne is made of wood and presented in oblique view, as is often the case in dugento Italian painting. It differs from the frontally depicted masonry structure in the other. Typically shown in reverse perspective, here, as in other dugento Madonnas with oblique thrones, Cimabue connected the centered position of the back of the throne with the asymmetrical oblique view of throne seat, so that the Virgin's central hierarchic presence could be maintained. 40 A The frame of the Louvre Madonna is original.
Here the connection of the throne to the ground on which it rests deserves close attention. Beneath the throne Cimabue introduced two solid oblong bars, one at each side. They change directions, from lateral to oblique. The bar at the left side is the more prominent, receding further into depth than the other. Signifi cantly, the throne's four feet rest on these bars at different levels, the front feet set duly below those at the rear. In turn, these bars are set arbitrarily on two other bars, the latter equally changing directions. Cut off by the base line, they approach the beholder. From their side terminations it is evident that the lower bars are seen from the right side, while the upper bars are seen from the left, these conforming to the oblique position of the throne. Beneath the Virgin's foot stool's curved front face the space between them consists of a neutral gold ground. These ground bars, in spite of their arbitrary spatial presentation, introduce an element of pictorial depth and material substance closely connected to the picture plane.
To my knowledge, these segmented ground bars used to support the feet of the Virgin's throne and foot stool are only found in Cimabue's paintings. They return in his Servite Madonna in Bologna and in the small Madonna panel recently discovered in England, now in the National Gallery, London.
Rhythmically superposed, three angels fl ank both sides of the Louvre throne. Signifi cantly, similar to the Uffi zi panel, here too the placement of their two visible feet projects the closest angels toward the beholder. Touching the base line in their slanted positions, they seem to fl oat in space. As a result, here throne and angels remain closely tied to the picture plane.
Interestingly, while in the Uffi zi panel chrysography covers the entire attire of the Virgin and Christ Child, in the Louvre painting it is restricted to the red dress of the latter. And, as in the Bologna Madonna, to be considered, here the gold background is decorated with a fi ne grid made with dotted lines, less complicated than the one in the Uffi zi painting.
The Madonna in the Servite church in Bologna is much restored. 41 Its throne is now lyre shaped and there are only two angels present, placed in back of the throne. Unfortunately, close to the base line the painting is largely obliterated. Enough remains, however, to indicate a similar use of ground bars serving as supports for the throne feet. Here the back of the throne, covering nearly the entire available image width, is still centered, 42 while the throne seat is again seen in oblique view from the left. As a consequence, as in the Louvre painting, on her seat the body of the Virgin is shifted somewhat to the left of center. The Virgin's head is also set along the vertical axis, and reverse perspective still applies. However, from what remains it is evident that now the throne is less tied to the picture plane. The oblique ground bar at the left side is aligned with the receding side of the throne, its length determined by the locations of the throne's respective rear and front feet resting on it.
Signifi cantly, now the Virgin's foot stool is clearly placed in front of the throne. It is a separate structure, and has its own supporting platform. Also, its rear and front feet, the latter partly obliterated, were originally duly set at different levels. Further, now ground bars and platform conform to the same oblique view as both the throne and foot stool. Clearly, compared to the Louvre Madonna here foreground spatial recession is more coherently presented and reaches further into depth! The small Madonna panel recently acquired by the National Gallery, London, was recently discovered in a private collection in England. It constitutes an important contribution to Cimabue's autograph oeuvre. 43 Here only two angels fl ank the throne. They are as tall as the Virgin in her seated position, so that her hierarchic presence is reduced. And the throne demands less proportionate space than in the previous paintings. Signifi cantly, its feet still rest on solid ground bars. The ground bar at the left side, again changing direction from lateral to oblique, offers support to the throne's respective rear and front feet at appropriate levels, the projecting portion of the ground bar cut off by the base line. Signifi cantly, now the angels stand on the rear portions of these ground bars, that is, beside the throne's rear side posts. Accordingly, they have been moved into depth! Their spatial location shows that here Cimabue has become more aware of coherent pictorial recession.
The mural of the Madonna Enthroned with Saint Francis in the lower crossing of the church of San Francesco at Assisi cannot be excluded from this enquiry. Considering the mural's poor condition, the basic composition is still le gible. It has been reduced at the left side. 44 Now fl anked by the murals of a Giottesque shop active sometime in the second decade of the trecento, it is the earliest extant mural in the crossing of the lower church. Considering its reduction at the left side, the assumed original central location of the throne and the Virgin and Child cannot be confi rmed.
In essence, its composition follows that of Cimabue's two previous Madonnas considered. However, there are some signifi cant changes. The carpenter's throne now rests on a platform, which replaces the ground bars found in the former. This platform is presented in oblique view and in traditional reverse perspective. It duly supports the throne's rear and front feet at appropriate different levels. However, the forward projecting foot stool, here connected to the throne, is still spatially compressed, since its front feet are laterally closely aligned with those of the throne near the platform's front edge.
Signifi cantly, of the four angels here included, according to the placement of their feet the closest two are clearly placed behind the throne, near the rear edge of the platform supporting them. Their projection into depth even exceeds that of the angels in the London Madonna! Similarly, Saint Francis, standing to the right of the throne, is also set in a middle ground according to the location of his feet. Smaller than the Virgin and the angels, he submits to a clear hierarchic order. In its treatment of pictorial depth, among Cimabue's extant Madonnas the Assisi mural is clearly among the most advanced! One wonders if the appearance of the Christ Child would corroborate this conclusion. He looks to the side, fi rmly seated on the Virgin's left knee. The speaking-bles sing gesture of his right hand follows the direction of his face. Discarding the scroll, he also lacks the formal position and dress of the Uffi zi and Louvre Christ Children. Instead, he holds onto part of his cloak with his left hand. His position substantially conforms to that of Duccio's Child in the Rucellai Madonna dated close after 1285! The tendency toward presenting the Child in a more intimate and child like manner is even more pronounced in both the Servite Madonna in Bologna and the small Madonna in London. In the former the Child, properly supported by the Virgin, strides affectionately toward his mother holding onto her shoulder. He wears a light purple dress slit at the bottom so that his forward striding foot is visible. This type of Christ Child appears in other Tuscan paintings from around 1300. 45 And in the London Madonna, wearing a plain white shirt and dress, the Child is seated on his mother's lap playfully holding her right wrist and fi ngers. 46 Fig. 18 The change in treatment of the luxurious throne cloth offers further information bearing on their chronology. In the Louvre Madonna it covers the back of the throne in rigid vertical folds. And, as has been noted, in its treatment the throne cloth of the Santa Trinità Madonna is not far removed. Consider the similar rigid edges along their left (our right) sides.
Signifi cantly, the throne cloth of the Servite Madonna categorically differs from the former in its pliant aspect. Covering the back of the throne, it responds to the pressure of Virgin's body as it extends over her seat cushions, before descending in back of her legs.
In the recently discovered Madonna in the National Gallery, London, the throne cloth consists of two parts. A red cloth covers the back of the throne, while another, pliant, transparent and white, reaches over the Virgin's seat cushion and then descends beyond her feet. The transparency of the latter corresponds to that of the loin cloth of the crucifi ed Christ on Of all of Cimabue's known thrones, the monumental throne in his mural of the Virgin and Christ Enthroned in Heaven in the choir of the upper church of San Francesco at Assisi (Figure 18) is closest to the one in the Santa Trinità Madonna. Frontally disposed, it also nearly covers the full height of the available space. And it also consists of two parts: a substantial lower concave structure resting directly on the mural's base line, and the actual throne rising above, set two steps back. As a consequence, as in the Uffi zi Madonna, the Virgin is seated high in the image, this time beside her Divine Son.
Here the stool supporting the feet of the Virgin and Christ is set on two projecting steps. The spatial connection of the throne to this foot stool, also consisting of two steps and viewed obliquely in reverse perspective from the right side, remains unclear. Seen from above, the placement of the foot stool's three visible feet suggests that it fl oats somewhat before the throne. It is clearly presented as a separate structure since the lower portion of the throne cloth descends between the two. The gold lines of the imitation chrysography located on all plane parts of the throne and the lower structure reach upward and inward in distance, as in the Santa Trinità Madonna, while those on the foot stool do the opposite.
The material of which the throne is made is uncertain. On the one hand, the sheer monumental size of the throne suggests masonry. On the other, the visible parts of the throne back, the knobs lining its upper edge, the intricate shape of the rear side posts, and the leaves covering much of its visible surface, suggest rather that it is made of wood. Interestingly, the upper side posts, seemingly turned on a lathe, resemble those in his Santa Trinità Madonna in the detail! The leaves on the thrones' vertical surfaces are also of similar shape.
Further, the disposition of the throne cloth is also similar. In both paintings it decends sharply in a straight line at the right side, its descent continuing behind the foot stool below the seat.
Finally, the two thrones share the same symbolic role of representing the Church and its salutory Divine mission. Arranged in overlapping registers, here Old Testa- ment fi gures, located at the right side, are present in greater abundance. Joined by saints and angels, they all turn toward the heavenly pair in gestures of awe and adoration. Similarly composed, at the left side of the throne appear kneeling friars and clerics. While all fi gures at the right side have haloes, the crowded clerics and friars at the left side do not. Accordingly, they would represent the living church. The scale of the side fi gures is arbitrary, since those in front, that is, closest to the beholder, are smaller than the fi gures located beyond, following common medieval practice reverting to late antiquity. Exclusive of the two angels putting their hands on the Virgin's and Christ's cushion, here only the praying hands of a friar, and those of a cleric below him, reach over the throne in their direction. In turn the heavenly pair gestures toward this friar. The fi gure above him repeats the Virgin's gesture as he extends his open hand diagonally downward. Signifi cantly, his halo differs from the others in the mural in that it is not radially grooved. His head is also unusually worn, suggestive of having originally been painted in secco, which did not last. Interestingly, neither the praying friar whom the Virgin introduces to Christ has a halo. This would exclude his being Saint Francis, although one would expect him to be the person introduced to God in his funerary church. Whoever he may have been, the mural surely represented, sanctioned by Christ and the Virgin, the renewal of the earthly church by the order of Saint Francis.

Cimabue's Evangelists in the Upper Crossing Vault
Cimabue's awareness of Byzantine art is clearly evident in his Evangelist murals in the upper crossing vault of the church of San Francesco at Assisi (Figure 19). This includes their spatial rendering. There he covered the surface of their cells with gold leaf, imitating, using a different medium, the gold mosaic ground found in many Byzantine church interiors and on the apses of many Italian Early Christian and medieval churches. On these spatially neutral golden surfaces he then applied the Evangelists, their chairs and pulpits. Signifi cantly, he set each of these on a separate platform, which he then stacked one on top of the other, successively supporting their pulpit, the Evangelists' chair, and the lowest their feet. As has been indicated, this use of platforms, appearing single or sometimes doubled, for supporting particular objects or fi gures, appears widely in medieval Byzantine art: on ivory reliefs, in panel paintings and book illuminations. However, to my knowledge, Cimabue's three-fold superposition of such platforms, contributing, somewhat incoherently, to the Evangelists' spatial presence, is unprecedented. Here Cimabue takes a Byzantine convention and elaborates on it! This solution seems connected to the ground bars in his Madonna paintings used to support the Virgin's throne.

Concerning Chrysography
Chrysography, the application of gold striation patterns on Christ's garments, probably originally transferred and adapted to painting and mosaic from cloisonné enamel technique, is widely found in Byzantine art and its sphere of infl uence. It is traditionally selectively used as a divine attribute. As here interpreted, Cimabue's "chrysography" would include both: typical striation patterns using actual gold; and also its imitation using related colours. 50 On the whole, his use of chrysography does not seem to follow a straightforward evolutionary pattern. In the Louvre Madonna chrysography is used sparingly. It is absent from the Virgin's blue cloak, and in the Christ Child's attire it is restricted to his tailored orange dress. It is also applied on the orange upper portions of the angels' attire. And imitation chrysography appears on much of the throne.
Signifi cantly, in the Santa Trinità Madonna its presence is more prominent. It appears on the entire attire of the Virgin and the Christ Child, as well as portions of those of the angels and Jeremiah. And, as in the Louvre painting, imitation chrysography is applied on all plane sections of the elaborate throne. This particular emphasis on chrysography corresponds to its extended presence in a number of Tuscan paintings dating from the sixties and seventies of the Dugento, where it is often used to excess! 51 On the whole, insofar as one can tell from their poor condition, the chrysography in Cimabue's murals in the upper church of San Francesco conforms to its appearance in the Santa Trinità Madonna. Signifi cantly, in the scene of Christ and the Virgin enthroned in Heaven, the attire of both is covered with it, with the exception of the decorated cloth cast over the Virgin's feet. It appears as well on portions of the attire of the angels and Old Testament fi gures. And chrysography also appears on all plane sections of the throne and foot stool. Its extended use is also found in other murals of the same cycle, including the Angel inspiring Saint John on the Island of Patmos. All this does not mean that Cimabue used chrysography consistently in the course of his artistic career. As time progressed his interest in realistic optical and physical qualities of cloth proved incompatible with its obvious arbitrary aspect. Consider his painted crosses in Arezzo and S. Croce, Florence. In the former chrysography is prominent, covering all garments. However, it had no place in the latter, where the loin cloth is largely transparent, nor, for the same reason, in the small Flagellation panel in the Frick Collection, which has been connected by Joanna Cannon, as indicated, to Cimabue's recently discovered London Madonna. 52 Nor does it appear on the attire of the Virgin and Child in the Servite Madonna in Bologna, except for the thin bands at the edges of her dress at the neck and wrists.
Unquestionably, in later dugento Italian painting the use of chrysography, given its obvious connection to Byzantium, was widely debated. In the mural decoration of the upper church of San Francesco at Assisi the presence of chrysography diminishes after Cimabue's presence, duly restricted to Christ's or God the Father's attire. This can be seen in Torriti's scene of the Creation of Man on the upper north nave wall close to the crossing. It also covers the entire red dress of Christ in the mural of the Road to Calvary in the second bay of the south nave wall. Its restricted use applies as well to the Life of Saint Francis cycle lining the lower nave walls as well. Interestingly, chrysography is also absent from Giotto's representations of Christ in the mural decoration of the Arena Chapel. Exceptionally, however, there it does appear on the attire of the Old Testament fi gures fl anking Christ ascending to Heaven. Here its application has clearly a symbolic value, connecting the Old Testament fi gures with an antiquated stylistic practice obviously known to Giotto.
In overview, chrysography was a Byzantine import on central Italian soil. Its use escalated in the later sixties and seventies of the dugento, often applied to excess according to its original Byzantine purpose. 53 Judging by the murals of the Sancta Sanctorum Chapel, in Rome it met with resistance, since it was excluded from the red dress of Christ enthroned, modeled by the partial superposition of sharply contrasting light blue colour ( Figure 6). In the mid-twelfth century apse mosaic of the Virgin and Christ Enthroned in Heaven in the church of Santa Maria in Trastevere one already observes, excluding chrysography's purely decorative aspect, the use of golden tessere on divine or saintly garments for obtaining luminous modeling effects. This tradition carried into the later dugento, as can be seen in Torriti's apse mosaic of

Concerning Cimabue's artistic connection to the Saint Francis Cycle
In his Assisi murals the sharp contrast of Cimabue's concept of spatial depth and narrative structure with the later murals of the Life of Saint Francis in the same upper church has already been underscored. In the view of its masters Cimabue's style had become thoroughly outdated, this including his Byzantine legacy. A vital aspect of this contrast, often examined, involves the perspective conventions used in the respective mural borders. Here the plain console band extending above Cimabue's Life of the Virgin murals in the upper choir differs fundamentally in its perspective rendering from the framing system of the Life of Saint Francis cycle, which presents a covered passage extending before the murals between the engaged nave piers.
Signifi cantly, the latter offers, within the space of each bay, console friezes located below and above the narrative scenes, as well as a coffered ceiling serving the covered passage. Here the orthogonals get duly closer in distance, approximating correct perspective structure ( Figure 21). Differently, in the case of Cimabue's console series the principle of reverse perspective still applies, since the direction of the orthogonals is reversed, since in distance they get farther apart. Accordingly, both sides of the central console, located directly above the papal throne, are visible, seen simultaneously from opposite oblique directions (Figure 22). One must keep in mind that here we are dealing with architectural illusion important for its own sake, rather than serving narrative or iconic ends. Nevertheless, this radical change from reverse toward real perspective representation shows that this issue was discussed around the time that the framing scheme of the Saint Francis cycle was invented.
The closer reading of the initiation of the Saint Francis cycle still largely defi es resolution. In the fi rst place, we do not know precisely when the cycle was begun. It is generally accepted that it was painted sequentially following the course of Saint Francis' life, offering a stepped increase in realism and compositional complexity. This assumption is confi rmed, on the whole, by the sequence of the giornate. 54 The narrative, moving clockwise around the lower nave wall, begins next to the northern transept. However, the very fi rst narrative scene there located, of Saint Francis honoured by a simpleton, was actually painted last. This is confi rmed by both the sequence of the giornate and its unusual style, connecting the latter to the three murals by the same hand located in the same bay on the opposite nave wall. Their painter has been identifi ed as the Santa Cecilia Master, after his altarpiece in the Uffi zi Gallery. This is the only painter partaking of the Saint Francis cycle on whose identity there is closer agreement. The fi rst scenes of the cycle painted are the two following the simpleton honouring the saint. They offer Saint Francis giving his cloak to the poor knight ( Figure 23) and his dream vision of the palace fi lled with arms bearing the sign of the cross. Interestingly, the investigation of the sequence of the giornate indicates that the main part of the latter scene was painted before the gift of the cloak. 55 Be this is it may, the mural of the gift of the cloak is surely the most conservative in the whole Saint Francis cycle. Its composition is among the least complicated. There the fi gures are the largest. They are simply laterally rowed close to the picture plane. The frontally posed Saint Francis stands along the vertical axis. The very center of the pictorial fi eld is located on his halo.
And the slanting sides of the hills are directed toward his head. Last but not least, his diagonally downward tilted feet suggest that rather than standing on it he fl oats before the ground. Differently, the feet of the poor knight standing at the side are more believably laterally oriented. And all the feet, including the horse's hooves, are sharply spotted against a light ground, the chiaroscuro contrast contributing to the impression of their collective forward projection. Moreover, the rocky ledge following closely the base line is a traditional Byzantine feature in outdoor settings, already found in the Good Shepherd mosaic in the mid-fi fth century mausoleum of Galla Placidia in Ravenna. Altogether, compared to the later murals of the cycle, here a reduced conservative visual narrative parlance prevails. Certain of its components, including the frontal position of Saint Francis and his diagonally tilted feet, conform to that of the soldiers in Cimabue's Crucifi xion mural in the southern transept. Wholly dismissing the idea of a supporting ground, there the fi gures fl anking the cross stand on each others' slanting crossing feet (Figure 10)! In the light of the conservatism of Cimabue's Assisi murals, one wonders what went on in the minds of the painters responsible for the Life of Saint Francis cycle as it progressed. Where in all this brew the young Giotto belonged, struggling to fi nd his own way, and the Isaac Master, whoever he may have been, 56 remains for me as elusive as ever, although it can be assumed that they were close.

Cimabue's Architectural Forms
Obviously, compared to the most progressive architectural representations in the Life of Saint Francis cycle, those in Cimabue's Assisi murals are yet downright primitive and archaic. The contrast is so extreme that it alone excludes a closely compacted chronology of the mural decoration in the upper church as has been recently proposed. 57 Consider, on the one hand, the elaborate Gothic interior hall in the mural of Saint Francis preaching before Pope Honorius ( Figure  24). Occupying the entire mural, the hall, duly receding in distance, effectively contains all the fi gures in their respective spatial locations, separated from the beholder by a triple arcade crossing the mural close to the picture plane. And in the Miracle of the Christmas Crib at Greccio ( Figure 25) the painter places the event within a church: in the choir in front of the high altar, a crowd pressing through the central door of the iconostasis. Here the realistic representation of the parts in their respective locations enables the beholder to place the event within a greater familiar architectural whole. Interestingly, in another mural of the Saint Francis cycle, the funeral of the Saint, the orientation of the same interior ambient is reversed, with the scene set in the nave. Now the iconostasis is viewed from the opposite side since the icons rising from  it are no longer seen from the back but from the front. The painter even includes a portion of the coffered apse vault beyond the icon of Saint Michael. Here, moving from one scene to another, the painter shifts the viewer's position within one and the same greater structure. Differently, in Cimabue's choir and transept murals nothing even approaching this kind of sophisticated architectural rendering can be found. Individual buildings are yet reduced in shape and scale. In the scene of Saint Peter healing a Cripple ( Figure 26) the reduced buildings at the two sides simply converge, according to their roof lines, toward the lower center where the principal action is set. Also, following medieval tradition, here the symbolic role of reduced architecture often takes precedence. Consider the central location of the plain domed church set directly in back of Saint Peter healing the lame, or the plain baldacchino similarly placed in back of Saint Peter exorcising demons from the bodies of men. As previously considered, in his Evangelist portraits in the upper crossing vault ( Figure  19), the platforms stacked one on top of another, supporting separate objects and fi gures and set against a neutral ground, yet clearly indicate a disinterest in rendering a unifi ed architectural setting. By contrast the latter is substantially realised in the vault murals of the Doctors of the Church in the fi rst bay of the upper nave. There each cell offers one extended platform which supports all fi gures, objects and structures there located, the platform consisting of the fl at roof of a building barely rising beyond the lower cell borders ( Figure  27)! All this information clearly separates Cimabue's upper church murals from the latter, as well as the cycle of the Life of Saint Francis.

The Role of Legend in our understanding of Cimabue's connection to Giotto
Virtually nothing is known of Giotto's origins. Vasari writes that Giotto was born in 1276, 58 that his father was a farmer in the region of Vespignano. He relates that when Giotto was ten years old his father sent him to pasture sheep, and while doing so he drew one on a rock. It just so happened that Cimabue was passing by. Struck by the realism of Giotto's drawing Cimabue, with the father's permission, took the young artist into his shop, and that is how Giotto's career got started. 59 Obviously, all this is the stuff of legend. It conveniently connects the lives of Florence's two greatest painters of the proto-Renaissance. Vasari did not invent it, but borrowed it from Ghiberti's Commentaries. 60 Where Ghiberti got it is not known. In essence, Vasari was fully aware of their distinct historical roles. This is how he comments on the art of the young Giotto: "… in poco tempo, aiutato della natura e ammaestrato da Cimabue, non solo pareggiò il fanciullo la maniera del maestro suo, ma divenne cosi buono imitatore della natura che sbandi affatto quella goffa maniera greca, e risuscitó la moderna e buona arte della pittura…" (in a short time, assisted by nature and taught by Cimabue, the boy not only equalled the manner of his master, but became such a good imitator of nature that he effectively rid himself of that awkward Greek manner, and revived the good and modern art of painting). 61 Considering the fundamental differences of their styles, and how little we really know of Giotto's formative years, does it make sense to assume that Cimabue was Giotto's teacher? If so, the apprentice would have thoroughly rejected the elder master's way of painting in the search for his own artistic identity. As indicated, the extant visual evidence raises severe doubts. For what this may be worth, Vasari is known to have inserted legend in certain of his artists' biographies. He did so in his life of Buffalmacco, of whose work in his time virtually nothing was known, referring to Sacchetti's and Boccaccio's literary accounts of the master's practical jokes. 62 The contemporary Dante was certainly aware that Cimabue's and Giotto's artistic personalities differed, implied in the assertion that the fame of one eclipsed that of the other. According to the present state of knowledge the connection of the young Giotto to Cimabue, if not discarded, is best left open.

Appendix. A comment on Giotto's symbolic use of architecture in his Arena Chapel murals
According to the narrative sequence of Giotto's murals of the Life of the Virgin and Christ in the Arena Chapel, the earliest scene is the Refusal of Joachim's Sacrifi ce in the Temple located on the upper right side of the nave (Figure 28). Taken from the apocrypha of pseudo-Matthew, it initiates the account of the Virgin's miraculous birth, contributing to the absolute purity of the future mother of the incarnate God. Giotto devoted the uppermost nave tier to her early life up to the time of her marriage to Joseph, which legitimized her forthcoming miraculous maternity according to man's terrestrial law.
The upper portion of the triumphal arch offers the second scene marking a crucial moment in the mural cycle's narrative progress (Figure 29). According to its shared border it includes two successive events. The fi rst fi lls the wall space above the arch. There one observes the heavenly court in session as God the Father, reconciling the contrary claims of Truth and Justice, decides on Christ's salutory earthly mission. 63 Below appears the Annunciation, with Gabriel located at the viewer's left side of the arch and the Virgin at the other. The total scene introduces the narrative of Christ's life, which occupies the lower nave walls, beginning with his birth represented in the adjacent mural on the right nave side. Interestingly, it is precisely in these two scenes that Giotto deliberately reverted to obsolete spatial practices at the time long since discarded. He does so, however, with exceptional sophistication. If my reading is correct, he did so deliberately for symbolic reasons, using them as chronological indicators.
The decoration of the triumphal arch offers obvious confl icting perspective views. On the one hand, viewed by the observer duly standing toward the middle of the nave, the empty vaulted rooms located at the lower sides conform to coherent perspective construction, since the orthogonals get closer in distance. However, their spatial treatment contradicts that of the two fl anking structures appearing above, opened at the front, revealing Gabriel and the Annunciate Virgin kneeling inside (Figure 33). At the sides of these structures appear two projecting balconies supporting a fl at roof. Seen from below, both balconies and roofs project obliquely upward and forward into the interior of the chapel, their oblique direction leading one's eye toward God the Father. According to real perspective they are viewed respectively impossibly from beyond the nave walls. Accordingly, considered as a unit, their orthogonals separate as they extend into distance. Here the principle of reverse perspective applies!
The mural of Joachim's Sacrifi ce Refused also reveals an obsolete convention, but of a different kind. Here Giotto locates all his fi gures and structures, with the plausible ex-ception of the rear pulpit, on a solid platform which recedes sharply into distance. It is seen from a vantage point roughly at level with the painted fi gures. This platform is set obliquely on a lateral brown band suggestive of the ground. Above the latter extends a fl at blue ground. Awareness of spatial depth is restricted here to the platform and the fi gures and structures it supports. It is roughly rectangular in shape, projecting slightly at the rear sides. Its front corners are cut off at an angle, so that the closest part connects with the base line and the picture plane, thus keeping the platform from projecting into the viewer's space. On this platform the fi gures and structures are closely compacted. This eminent use of a platform for supporting objects and fi gures is reminiscent of their traditional use in Byzantine painting, and especially Cimabue's Evangelists in the upper crossing vault of the church of San Francesco at Assisi. These, it can be assumed, Giotto certainety knew.