The five canvases at Mells – the present Meeting of Anna and Joachim at the Golden Gate (MM7), The Visitation (MM8), The Presentation of Christ in the Temple (MM9), Christ disputing with the Doctors (MM10) and The Wedding Feast at Cana (MM11) – are en suite with four others: two in the Galleria Sabauda, Turin, representing the Birth of the Virgin and the Annunciation; and two until recently in the Stanley Moss Collection, New York, and now, reportedly, in the Museum of Art in Mexico City, which illustrate the Marriage of the Virgin and the Adoration of the Magi. The nine canvases, which are more or less identical in size, are generally attributed to Jacopo Bellini and his studio and widely thought to be the remains of a more extensive cycle of narratives of the lives of the Virgin and of Christ once in the Sala Grande or Sala Capitolare on the piano nobile of the Scuola di San Giovanni Evangelista, Venice.

The context is laid out most clearly by Patricia Fortini Brown. As early as 1414, the governors of the Scuola commissioned a series of narrative paintings celebrating the miracles of the Reliquary of the True Cross (a reliquary that they owned) for the Scuola’s Albergo.¹ In 1421 they decided to commission a cycle of scenes from the Old and New Testaments for the Scuola’s Sala Capitolare.² This project is not documented further but there is no evidence that it did not go ahead and scenes from the Old and New Testaments are mentioned in the account of Francesco Sansovino, which will be discussed below. The style of paintings executed in Venice in the 1410s and 1420s would have been predominantly ‘International Gothic’, with crowded compositions, close attention to local detail, strongly patterned drapery, minimal concern with pictorial space and much surface decoration.

In 1453, some rebuilding work was undertaken in the Scuola and this date has been taken as an approximate terminus post quem for the Mells-Mexico-Turin group; however, no documentary evidence has been published to support the proposal that a cycle of narrative paintings was commissioned at this time. Had such a cycle been commissioned at that date then it would presumably have been either a continuation of, or a revival of, the project discussed in 1421, or a replacement for all or part of it, although it would be unusual to replace a cycle begun only thirty years before. Jacopo, as the leading painter in Venice in this period, would have been an obvious candidate to execute such a cycle, had one been commissioned, especially if he had, as Sansovino believed, been involved in the earlier Old Testament/New Testament cycle. It has been noted that Jacopo received a payment of 8 ducats for unspecified work on 1 January 1465, a plausible terminus ante quem for the completion of the putative new cycle. But the payment may well have been a final account for Jacopo’s now lost altarpiece, commissioned in 1457, of the Dead Christ supported by Angels. The 1414 cycle of the Miracles of the Reliquary in the Albergo was replaced between the early 1490s and 1512 by an imposing new cycle executed by a team of Venice’s most prominent painters, including Jacopo’s elder son, Gentile.³ Whether or not any of the 1414 paintings (assuming that they were not frescoes) were salvaged at this time is unknown.

In the first edition of his Vite, published in 1550, Vasari mentioned a cycle in the Sala Grande but his account is vague, cursory and evidently from hearsay: ‘Furono le prime cose che diedono fama a Iacopo per gli aiuti de’figliuoli una storia che alcuni dicono che è nella Scuola di San Giovanni Evangelista’ dove sono le storie della Croce. Le quali furono dipinte da loro in tela, per avere del continuo costumato quella città di far lavorare in quella maniera’.⁴ The ‘dicono’ is significant. Vasari had not seen the Scuola and was conflating Jacopo’s cycle with the True Cross cycle in the Albergo, an error to which Ridolfi drew attention. In stating that the canvases were early Vasari did not realise that, if that were true, Jacopo could not have collaborated on them with his sons. Vasari either muddled what he had been told or his informant was ignorant of Jacopo’s chronology, or both. In 1568 this account was modified a little but not improved. Vasari writes:

Le prime cose che diedono fama a Jacopo, furono il ritratto di Giorgio Cornaro e di Caterina reina di Cipri, una tavola che egli mandò a Verona, dentrovi la passione di Christo con molte figure, tra le quali ritrasse se stesso di naturale et una storia della croce, la quale si dice essere nella Scuola di S. Giovanni Evangelista, le quali tutti e molte altre furono dipinti da Jacopo con l’aiuto de figliuoli; e questa ultima storia fu fatta in tela, sì come si è quasi sempre in quella città costumato a fare . . .⁵

This adds to the confusion by attributing to Jacopo portraits certainly executed by Gentile and placing them among Jacopo’s earliest paintings. Vasari obviously did not enter the Scuola when he revisited Venice in 1566. But while Vasari’s reports in both editions are effectively worthless, it is clear that he – or, rather, his informant – believed the cycle to have been executed by Jacopo Bellini in collaboration with his sons.⁶

The next known account, published in 1581, was by Francesco Sansovino: ‘Vi sono medisimamente pitture diverse, della historia del testamento vecchio & nuovo, con la passione di Christo, non punto volgari, & la seconda parte di questa opera fu di mano di Jacomo [sic] Bellini, che fece anco la seconda parte della Natività’.⁷ Sansovino’s text is not wholly clear but it seems that he knew, from direct observation, a cycle in the Scuola that was divided into two parts: one comprising Old Testament subjects, the other comprising New Testament subjects, perhaps displayed in typological pairings, side by side or facing each other. If so, this would correspond plausibly to the project outlined in 1421. But it is uncertain whether what Sansovino calls the Nativity constituted the whole or simply a part of the New Testament cycle; or whether it was an entirely separate series of the Life of the Virgin. However, since no other information suggests the presence of a third cycle in the Sala Capitolare, it should probably be accepted that Sansovino was writing loosely and that the ‘testamento . . . nuovo’ and the ‘Natività’ (or Life of the Virgin) were one and the same. Sansovino believed Jacopo to have had a part in both cycles but that he had executed neither in its entirety.

Shortly after Sansovino wrote, the cycles seem to have been dismantled and removed from the Sala Capitolare. They were replaced, between 1595 and 1626, by a series of canvases by Domenico Tintoretto, Sante Peranda and Andrea Vicentino and, perhaps, others.⁸ But some of the older paintings may not have been discarded; any that were retained would probably have been placed in storage, perhaps on the Scuola’s ground floor.

 In 1648 the paintings in the Scuola were described by Ridolfi in his life of Jacopo Bellini. He said nothing of an Old Testament cycle, of which there is no trace after Sansovino, nor of a New Testament cycle but he described in detail the Life of the Virgin which he attributed to Jacopo Bellini and in which he noted Jacopo ‘servendogli i figliuoli d’alcuna aiuto.’ Ridolfi stated that the cycle was in poor condition and said that he had relied on reports by older painters but it seems that he knew some of the compositions directly. Following an introductory reference to Jacopo’s altarpiece with the Body of Christ supported by Two Angels, he lists seventeen stories. His text is reprinted in extenso in the note below, for it is the single most important source for knowledge of the cycle.⁹ With the exception of Fortini Brown, it has not invariably been considered or interpreted with sufficient attention.¹⁰ Reduced to a simple list, the narratives were (with Mells-Mexico-Turin canvases conjecturally connected):

1. The Birth of the Virgin (Turin but St Joachim is not present)
2. The Virgin at work sewing in the Temple (lost)
3. The Marriage of the Virgin (Mexico)
4. The Annunciation (Turin but without a choir of angels)
5. The Visitation (Mells, MM8?)
6. The Adoration of the Child (lost? The Mexico canvas shows the Adoration of the Magi, not the Adoration of the Child)
7. The Presentation of the Child (Mells MM9? but the doves are absent)
8. The Flight into Egypt (lost)
9. The Holy Family in Egypt (lost)
10. The Return of the Holy Family from Egypt (lost)
11. Christ disputing with the Doctors (Mells MM10?)
12. Christ bidding farewell to the Virgin (lost)
13. St John informing the Virgin of the arrest of Christ (lost)
14. The Procession to Calvary (lost)
15. The Crucifixion (lost)
16. The Risen Christ appearing to the Virgin (lost)
17. The Coronation of the Virgin (lost)

The nine canvases in the Mells-Mexico-Turin group include three subjects not described by Ridolfi: the Meeting at the Golden Gate, the Wedding Feast at Cana (MM7 and MM11) and the Adoration of the Magi (what Ridolfi actually describes is a Nativity that recalls Nativities painted by Filippo Lippi in his International Gothic revival phase) in Mexico. In addition, some of Ridolfi’s detailed descriptions of subjects that are to be found in the Mells-Mexico-Turin group do not match them: thus, he says the Annunciation contained ‘a numerosa schiera d’Angeletti, festiggiante’ and that in the Presentation was an offering of ‘due candide colombe’.

In general, Ridolfi’s descriptions evoke the staging and figural interests of the flowery phase of International Gothic, the predominant style when Jacopo began his career and, as remarked earlier, appropriate to the 1420s and 30s, rather than that of the 1450s and 60s. Critics have tended to overlook the details of what Ridolfi says but, if his descriptions, many of which are unusually circumstantial, are taken at face value, they cannot, as Brown observed, be reconciled with the known canvases, either in their subjects or in what is implied of their styles. The historian is therefore compelled to make a choice. If Ridolfi’s account – which does not conflict with that of Sansovino – is ignored, then the Mells-Mexico-Turin group could in theory have originated from a cycle once in the Sala Capitolare of San Giovanni Evangelista. If the accuracy of Ridolfi’s descriptions is accepted, then the Mells-Mexico-Turin group cannot have formed part of that cycle. Supplementary evidence pointing to the latter conclusion is Ridolfi’s reference to Jacopo’s cycle as ‘divorati dal tempo’; the surviving canvases, while not in good condition, are not in as bad a state as this would suggest, nor do they seem to have been massively re- or over-painted – but a technical examination would be required to establish this with any security.

When, during the Napoleonic occupation of Venice many religious institutions were dissolved, the Mells-Mexico-Turin group came on to the market, apparently in 1808. We do not know if the nine canvases constituted all that then became available or only a part. They were acquired by the painters and dealers Natale Schiavoni and Francesco Cavella, who reported that they had come from the Scuola of San Giovanni Evangelista. It should be underlined that their statement is the single piece of evidence that supports the provenance of the canvases from the Scuola. The veracity of Schiavoni and Cavella has been accepted by most critics and it would be otiose to question it without solid evidence of deceit on their part. But to accept what they said entails disbelieving what Ridolfi said. The only hypothesis that this writer can envisage which preserves the accuracy of Ridolfi and the honesty of Schiavoni and Cavella is that Jacopo’s cycle, much deteriorated, was discarded and lost after 1648 and that the nine canvases in the Mells-Mexico-Turin group originated from some other scheme and were at an unknown date acquired by the Scuola di San Giovanni Evangelista or, perhaps, retrieved by the Scuola from some subsidiary site, conceivably on the terra firma. An alternative to the second possibility is that the Mells-Mexico-Turin canvases were merely deposited in the Scuola at the time of the suppressions and did not originate there. After Schiavoni’s death in 1858, the Mells-Mexico-Turin group was inherited by Elisa, his elder daughter, who was also Canella’s widow. Following her death they passed to her son Francesco, who in 1873 sold the Birth of the Virgin and the Annunciation, presumably regarded as the best of the group, to the Galleria Sabauda. The two now in Mexico were acquired in 1902 by Roger Fry from the collection of Ferdinando Ongania, who had presumably acquired them from Francesco Canella. Francesco, presumably, was the source of those now at Mells, acquired by William Graham in Venice, probably during the 1870s.¹¹

Ridolfi’s description evokes a remarkably rich Marian cycle within which several episodes from the life of Christ were intercalated. The cycle does not give Christ and the Virgin equal weight. None of the episodes in which Christ appears illustrate His mission. Mary appeared in all the canvases: it is her Nativity (or the Mells Meeting at the Golden Gate, if that is included) that initiates the scheme and her Coronation that concludes it. Among the episodes described by Ridolfi, the Education of the Virgin, the Holy Family in Egypt and their Return from Egypt are rarely represented and, as far as this writer is aware, that of St John informing the Virgin of Christ’s Arrest is unique. This scene would have underlined St John’s privileged relation both to Christ and the Virgin, proclaimed by Christ in the Crucifixion, and would be entirely appropriate in a commission for a Scuola dedicated to him. The nine canvases of the Mells-Mexico-Turin group include no uncommon subjects and nothing that would indicate any attachment to a particular commissioning body.

Reconstruction of the San Giovanni Evangelista Cycle

It must be repeated that no document records any commission by the Scuola of a cycle of paintings in the 1450s or 60s. But, whether or not the Mells-Mexico-Turin canvases are in question, the siting of the Scuola’s cycle or cycles is of interest. Collins divided the 17 (or 18 or 19 or 20 if one adds the Meeting at the Golden Gate or the Adoration of the Magi or the Miracle at Cana) episodes from the life of the Virgin into two roughly equal groups no doubt arranged in narrative order, placed on the facing long walls of the relatively narrow Sala. All nine surviving canvases are lit from the left but this would not invalidate a reconstruction that placed them on opposite walls for, if Collins’s reconstruction is to be trusted, the Sala was illuminated by a row of circular windows set high in the long walls, which would have provided even but non-directional illumination. Without a single dominant light-source, there would have been no need to light episodes from opposite directions even had they been placed on facing walls. This reconstruction of Jacopo’s cycle is followed by Eisler. However, Brown’s objections are hard to discount. She pointed out that Collins made no attempt to include the Old Testament/New Testament cycle and that, given that the length of the Sala Capitolare was 27.5 metres, all the Marian canvases listed by Ridolfi could have been placed on the same wall, facing the Old Testament cycle opposite. In addition, Professor Humfrey has pointed out to this writer (private communication, 2020) that the Sala Capitolare was some seven metres high and that he doubts whether such a large ‘room would have been decorated with canvases only 112 cm high. In scale they [would have been] too small.’¹²

Style, Attribution and Dating 

Scholars are agreed that all nine canvases show signs of work by different hands. This is unsurprising: collaboration, or sub-division of responsibility, as in the later Albergo cycle, would be understandable, even normal, in an extensive and probably ill-remunerated scheme, perhaps executed under pressure. There are also, no doubt, problems of condition; but the present writer cannot descry the massive repainting that would have been required to convert the subjects as Ridolfi described them into those that we now see. 

Some scholars have felt able to identify specific hands in the canvases. Roberto Longhi, for example, in his famous Viatico, detected the presence of both Gentile and Giovanni in the Birth of the Virgin and more substantially, of Gentile in the two Mexico canvases.¹³ His views are, broadly, followed by Lucco, who discusses only the Turin canvases, and by Mazzotta who, however, differs from Lucco by dating the canvases to the 1450s rather than the 1460s. However, any involvement on Gentile’s part in the Turin canvases is dismissed by Meyer zur Capellen. But, leaving aside whether or not Longhi’s judgements are correct, it should be pointed out that the two canvases in the Galleria Sabauda have been available for study for a century-and-a-half while the two now in Mexico have been little seen and the five at Mells were not even reproduced before 2016. It would require prolonged acquaintance with all nine canvases for the present writer – who has not seen those in Mexico, has no memory of those in Turin and has studied the Mells canvases only once and from a ladder – to feel at all confident about making distinctions of hand. So what follows is inevitably impressionistic.

In his view, the nine canvases present major deficiencies of design and conception as well as of execution. While all the episodes are clearly composed, their stories clearly narrated, their pictorial fields effectively filled, they have a diagrammatic rather than a pictorial clarity and efficacy. Throughout, the compositions resemble greatly enlarged miniatures or illustrative woodcuts. In design several of them – for example the Marriage of the Virgin and the Adoration of the Magi – seem to resemble the stolid simplicity of the Florentine painter Paolo Schiavo (1397–1478) who in the 1430s was active in North Italy and whose ideas can be seen in a more or less contemporary cycle of designs for Florentine ecclesiastical vestments.¹⁴ However, the writer freely admits that this relation may be no more than coincidence.

All the groupings are more or less naïf in arrangement, the figures are poorly structured, flat in modelling, their interrelations are clumsy and their gestures often stiff and without emotive force. Draperies rarely register the forms beneath; their colours are without any energy and some contours, for example those of the Virgin in the Visitation, are of remarkable dullness.

As narratives the Mells-Mexico-Turin canvases are simplistic and unimaginative in comparison with painted narratives by Venetian contemporaries. And compared with those by Mantegna (whose influence in the group has sometimes been detected by scholars but not by this writer) differences are massive and manifold. Mantegna, Jacopo Bellini’s son-in-law, was a predominant inspiration on both Gentile and Giovanni in the 1460s and the severity, formal precision, perspectival control and solidity of forms in paintings such as Giovanni’s Carità Triptychs and Gentile’s Organ Shutters, both of the early to mid-1460s, evoke a world entirely different from that of the Mells-Mexico-Turin group.

In sum, the present writer finds these nine canvases hard to integrate into the work of Jacopo or of his sons, either in conception or execution. Jacopo treated the Nativity of the Virgin and the Visitation in the predella of his Annunciation in Brescia of c.1440 and both these compositions are much more inventive in design and much richer in execution, despite their small scale, than those of the related canvases. The predella panels of the Gattamelata Altarpiece of 1460, thus putatively contemporary with the Mells-Mexico-Turin group, are incomparably more sophisticated in conception and unify figures, lighting and settings with landscapes with considerable fluency. It should be acknowledged that the Adoration of the Magi (now in Ferrara) has some compositional links with the Adoration of the Magi in Mexico but the latter is clumsy in arrangement, the main figures are poorly spaced and the actions of the attendants are muddled. And when we turn to Jacopo’s figure types, those in his later paintings, the St Jerome polyptych, of which three panels are at Vassar College, Poughkeepsie and the surviving section of the main panel of the Gattamelata Altarpiece, they are obviously very different, elongated and severe, quite unlike those of the figures in the present canvases in proportions or characterisations.

The forms, poses and modelling of figures designed and painted by Giovanni and Gentile Bellini between c.1460 and c.1470 bear little relation to the poses and structures of the figures in these canvases. A large number of narratives of various kinds were devised by Jacopo Bellini in his two drawing-books (London and Paris) and none of those in either book bears any close relation either in conception or in detail with these canvases. Hammond’s linking of Christ in the Temple with Jacopo’s drawing of the same subject in the British Museum’s volume seems to this writer only to underline their lack of similarity, as does his reference to Jacopo Bellini’s panel St Jerome in the Wilderness in Verona to support the attribution to him of the canvases.

This writer would not deny that there are a few links between the canvases and works by Jacopo and his sons but these are peripheral. Certain decorative features, the patterns on the robe of the Virgin in, for example, the Visitation and the Presentation, distantly (and coarsely) recall work by Jacopo’s master Gentile da Fabriano of the 1420s – but not the decorative manner of Jacopo as we know from the Brescia Annunciation, in which the draperies of Gabriel and the Virgin are infinitely richer and more subtle both in conception and execution. And, as far as one can tell from his surviving work, Jacopo ceased to employ this kind of decoration after that time. But a constant feature of Jacopo’s work, in different styles, is a great variety of surface refinement, obvious, for example, in his various treatments of the Virgin and Child and his bust of San Bernardino, whereas the surfaces of the present canvases are, throughout, dull and, even where best preserved, devoid of textural variety – or felicity.

Only occasional figures and details have any stylistic relation to works by Jacopo Bellini or either of his sons at the period in question. The head-types and wimples of the attendant female figures in some of the canvases, the Birth of the Virgin for example, resemble those seen in some of the early Madonnas by Giovanni Bellini. There is some similarity between the landscape in the Visitation (MM8) and that in Gentile’s St Francis receiving the Stigmata in his organ shutters, Joseph in the Adoration of the Magi and Christ disputing with the Doctors (MM10) is posed similarly to Jacopo’s St Peter in Berlin, a panel from a dismembered polyptych, probably of the 1430s, whose surviving parts are reproduced by Eisler (p. 68), but otherwise there is little of substance.

Throughout the nine canvases there is no interest in effects of light, so apparent in the Gattamelata predella and in the early works of Giovanni. Perspectival depth – one of Jacopo’s obsessions, as can be seen from his drawing books, plays a part only in the Annunciation, perhaps the liveliest of the nine compositions, but even there only as a backdrop. The Wedding Feast at Cana, which might seem to have called for a staging in depth, is composed as though it were a Feast in the House of Simon. Expressive drawing is absent and movements lack all fluency.

In short, the present writer is unable to fit the paintings in the Mells-Mexico-Turin series into any plausible account of Jacopo’s art or that of his sons. His provisional conclusions are that the canvases were not painted for the Scuola di San Giovanni Evangelista but for some other (perhaps poorer) community or church, conceivably – but not necessarily – a subsidiary possession of the Scuola; and that they were not painted by Jacopo or his sons but were produced by a satellite atelier, directed by a master who had had some association with the Bellini workshop and made use of a few Bellinesque ideas.

It follows that if the nine canvases are detached from the Scuola di San Giovanni Evangelista, there is no obligation to date them to 1455–65; and if Jacopo’s direct involvement with any of them is excluded, then his death in 1470 or 1471 would be no obstacle to dating them later: Berenson, who did not accept the attribution of the canvases to Jacopo, believed them to be of the 1470s and painted by some follower of Giovanni Bellini.

This conclusion inevitably invites us to reconsider Jacopo’s work in San Giovanni Evangelista as described by Ridolfi, which, regrettably, we must regard as entirely lost. We do not know how many scenes were commissioned or planned to be commissioned in 1421 but if we accept that those described by Ridolfi comprised the Nuovo Testamento series, then it is reasonable to assume that the Vecchio Testamento series was comparable in extent. Such a project would have been a very large one and would have taken some, perhaps many, years to complete. The length of time taken to execute the replacement cycle in the Albergo suggests that the scheme in the Sala Capitolare might have been underway for two or three decades. If so, then Jacopo’s work could well have been protracted into the 1450s and, consequently, it is highly likely that he would have involved his sons in the paintings’ execution. Gentile, born c.1429, could well have been active in his father’s shop by the mid-1440s; Giovanni’s birth date is much disputed and, indeed, is sometimes placed before Gentile’s, but even if, as the present writer believes, he was born nearly a decade later than his brother, he too would have been available to assist his father by the early to mid-1450s; he co-signed the Gattamelata Altarpiece with his father and brother in 1460 and seems also to have become active independently around this time.