9) Visual Narratology as an Index of Artistic Identity

[230] This chapter is in part another excursus and in part the first section of my conclusions. In particular it addresses a very obvious question; what, if anything does visual narratology add to the existing plethora of approaches, meta-languages and pseudo-methodologies available to the medieval art historian? I will attempt to answer that question by working through a specific example to show how, whatever other benefit it may or may not bring to the study of medieval art, visual narratology can sometimes serve as a useful index of artistic identity while also providing clues about the relationship between the artists and patrons of early thirteenth century stained glass.

The glazing programme in the ambulatory at Bourges originally consisted of twenty-five large single-lancet windows, fifteen of them divided equally among the five radiating chapels, the other ten occupying the ambulatory walls in between (see Fig. 9.01). The sills of these windows are unusually low (less than three metres above the pavement of the ambulatory) making the imagery particularly legible. Apart from the axial chapel and the lower registers of most bays in the other radiating chapels, the windows are largely intact and the major restoration campaign of the 1850’s was precociously sympathetic to the principle of documenting all changes.1 Louis Grodecki, one of the few twentieth century scholars to pay close attention to the Bourges glass, identified the work of three different workshops. To one of these he gave the Notname of ‘the Good Samaritan Workshop’, who, as well as the eponymous window in bay 13, were responsible for five other windows at Bourges and some at Poitiers.2 The designs of this particular workshop have repeatedly attracted my attention through their highly imaginative treatment of frames - something I have already had cause to mention in chapters 6.b and 8.b. The originality of this workshop and/or its master does not stop there however.

In 1941, Francois Quievreux published a study in which he identified two distinct workshops, involved in the Bourges glass (which he rather unimaginatively named ‘A’ and ‘B’) and tried to relate them to what he believed was the earlier glass at Chartres. 3 Seven years later Louis Grodecki dismissed most of Quievreux’s findings and, based on a diligent comparison of figure [231] styles, drapery patterns, colour palette etc, identified three workshops or masters, each of whom he named after their most distinctive window.4 His ‘Good Samaritan Workshop’ corresponded to Quievreux’s ‘Workshop B’, while the latter’s ‘Workshop A’ was split by Grodecki into two workshops named after the ‘New Alliance’ and ‘Relics of St Stephen’ windows (see Fig. 9.02). Extracting all the key details from Grodecki’s somewhat convoluted essay, the basic characteristics of the three workshops are summarised in Table 16.

Table 16 - Characteristics of the Bourges workshops identified by Grodecki (Quievreux’s names in brackets)

[232] Using these characteristics one can assign the surviving windows to workshops thus:

Table 17 - Allocation of individual windows to the three workshops identified by Grodecki

Unsurprisingly (given how little interest art historians have traditionally shown in such matters) Quiévreux and Grodecki both largely ignored issues of framing. The nearest either of them came to mentioning frame violations was when Grodecki described the different workshops’ styles of narrative composition:

Unlike the work of the other Bourges masters, the compositions of the Good Samaritan artist are often ill-balanced, overcrowded and confused. In a great many cases, he seems to have aimed at filling his picture-space as full as possible; this is particularly so in the Passion scenes, where whole groups are represented without any attempt to reduce the scale of the figures or to distribute them in groups. The figures, indeed, are hardly contained within their medallions, they tread on, and sometimes even beyond, the edgings. [...] Here is none of the noble and majestic harmony of the Alliance master, none of the sense of ease and balance of the Relics master. 5

[233] To say that that the figures ‘...tread on, and sometimes even beyond’ their frames is something of an understatement - though to attribute this characteristic to lack of ability would be a mistake.

I have already discussed in some detail the example of metalepsis in the lower register of the Good Samaritan window, a detail which in the past has been written off as clumsiness (see chapter 6.b). A closer inspection reveals that such ‘abuse’ of {picture} frame boundaries is a deliberate and recurring feature of this artist’s work. In his treatment of {picture} frames, one can identify two highly distinctive practices - firstly the deliberate elongation of pictorial elements beyond the normal border and secondly, the systematic use of ‘explicit continuous’ groupings of panels (see chapter 3.e). Of the latter, some of the most impressive examples are in the windows of St Nicholas (bay 19) and Mary Magdalen (bay 17). In both of these narrow lancets, the metal armature consists of a simple three-by-ten rectangular grid (with the exception of bay 21, the other workshops always use a two-by-ten grid for the narrower windows in the chapels - see Fig. 9.02), over which the artist has superimposed stacked medallions - large, two-register-high circles for St Nicholas, flattened single-register ovals for the Magdalen. He then allocates one, two or three individual {chronotopic} frames to the three panels per register, sometimes using the vertical bars of the armature as borders and sometimes deliberately ignoring them. Take for example panels 13-15 of the St Nicholas window in bay 19 (Fig. 9.03). Here we have two episodes from the segment concerning the three murdered clerics - a precursor to the Sweeny Todd story. Three young clerics (in some versions they are students or children but in this window they are tonsured adults) come to stay at an inn during a famine - in the night, the innkeeper and his wife murder them, chop up their bodies and salt them in a barrel, intending to sell the meat as bacon. St Nicholas then comes to the inn, opens the barrel and brings out the three young men unharmed. In panels 13-15, this segment has been reduced to its two key {chronotopic} frames - the moment of the murder shown in an explicit continuous scene split across panels 13 and 14, and Nicholas retrieving the men from their barrel in panel 15. Closer inspection however reveals that some elements of panel 14 spill over into the other scene in panel 15 - the left elbow, bag and skirt hem of the innkeeper’s wife. In the topmost register of that same window, in panels 28-30, we find the narrative segment from the [234] Life of St Nicholas that was mentioned briefly in chapter 8.3 - the miracle of the boy with the gold cup.6 Here again the artist has split two {chronotopic} frames across three panels (see Fig. 9.04). On the left, in panel 28, is the incident at sea - the boy, trying to fetch some water in the gold cup falls overboard (here helped by a little blue-grey devil), much to the obvious dismay of his parents. The prow and sail of the ship and the waves beneath it spill over into panel 29. On the far right of the register, in panel 30, is the Shrine of St Nicholas, where the grief-stricken parents have come to complete their pilgrimage. The replacement gold cup stands on the altar, where the boy’s father kneels in prayer. His wife turns in amazement towards panel 29 - where their son appears, guided by the saint, safe and well and still holding the original cup. Just visible in the bottom right hand corner of panel 29 is the tip of the right foot of the kneeling father, providing a physical link to panel 30. As a piece of visual story-telling this is virtuoso stuff. The artist has captured the two decisive moments very precisely - the moment of loss and the moment of rediscovery. Standing in the centre of the composition is St Nicholas himself; the hinge around which the story unfolds.7

Some of the most spectacular of the Good Samaritan master’s explicit-continuous framings are found in bay 17 - the Magdalen window. Panels 22-24 for example show the funeral of Lazarus (Fig. 9.05). One can compare the almost baroque drama of this image with its rather sedate equivalent at Chartres (Fig. 2.01). Whereas at Chartres the spatial relationship between the two panels is ambiguous, here at Bourges, all three panels in the register have been fused into a single image, with Mary and Martha at either end, framing a symmetrical composition. Their neighbours help them place their brother’s shrouded body in the tomb, aided by a white cloaked priest. Again the narrative fusion of the panels is emphasised by visual elements that traverse the vertical armature, particularly noticeable in the two men holding the corpse. Here the artist has excelled himself in the over-extended toes department - note how the toe of the man on the right just reaches into the middle panel, while the trailing foot of the man holding Lazarus’ shoulders has actually spilt into the border at the top of panel 19, in the register below.

In chapter 6.b I argued that the Good Samaritan master used a more extreme form of spillage into the border in the lowest register of his eponymous window as a metalepsis in order to re-engage the viewer at the most important point in the narrative. A similar, though more subtle, trick appears in one of his other large windows - the Passion window in bay 06 (see Fig. 9.06). This window is perhaps the busiest of all his designs, based around two columns of six roundels each linked by red and white vertical bands - very similar to the Job window in bay 112a at Poitiers but with the addition of small trefoil or quatrefoil panels containing subsidiary scenes [235] squeezed in between. Here the metalepsis is at the very top, in the Harrowing of Hell scene stretching between panels 27 and 28 (Fig. 9.07A). As per the standard iconography, Christ carries a triumphal cross - but here it is massively elongated, one end forcing open Hellmouth, the other extending out of the top left of panel 27 and across the decorative border (note His wrist and hand in the border) into panel 28, where an angel flies down carrying a golden crown.

The obvious question when confronted by such deliberate and systematic frame spillage is ‘why?’ Splitting an element across two stained glass panels like this is not a trivial task - it is certainly not something an artist could do by accident. Making a window, one starts with the design, or vidimus, which is drawn life sized on a white-washed table or board. The individual pieces of glass are cut, painted, re-fired as required, repositioned on the vidimus, fitted into their ‘cames’ (strips of lead with an ‘H’-shaped cross section) and mounted within edging leads (a thicker came with a glazing channel on one side only). Once the joints in the cames are soldered, and the edges sealed with putty, the completed panels can finally be fitted into the ‘T’-shaped saddle-bars of the iron armature. The key thing here is that the basic ‘constructional unit’ for these early thirteenth century windows is the panel (in the St Nicholas window the panels are around 40x60cm each - i.e. slightly smaller than four sheets of A4 paper) and each panel could be constructed on an individual vidimus. This is a far more practical solution than trying to lay out a whole six-metre high window at once, however it also tends to encourage a ‘frame-by-frame’ design mentality. The vast majority of medieval stained glass windows do indeed give the impression of having been designed one panel at a time (which is also how they tend to be published and studied by art historians). Of course the original reduced-scale design would have encompassed the whole window, at least in outline - but for the windows at Chartres, or for the New Alliance and Relics workshops at Bourges, the glaziers could have worked on individual panels without ever having to worry much about the transitions between them, except for the task of marking up the positions of the decorative borders where they crossed from one panel to another. If one looks closely at a range of windows, one finds that these borders rarely match up perfectly - panels are normally free to move slightly in the armature - and in some cases the glaziers just seem to have got their measurements slightly ‘off’. Because of the nature of the decorative patterns, these misalignments are easy to overlook, but when part of a human body spans two panels, any inaccuracies become painfully obvious. This kind of misalignment is clearly visible in Fig. 9.07A, where Christ’s arm and staff in panel 27 don’t line up properly with their continuation in the border. For some reason, the Good Samaritan master deliberately chose to make his life difficult, not only spanning panels with decorative borders but also doing so with various limbs and other visual elements in his narrative scenes.

This use of explicit continuous panels really is quite unusual. Of all the narrative windows at Bourges there are only three places where explicit continuous panels appear in windows made [236] by the other two workshops. One of these is actually a red herring; the boat in which Mary of Egypt travels from Alexandria spans panels 05 and 06 of bay 21 in a manner worthy of the Good Samaritan master himself but these two panels are purely nineteenth century inventions, made by Coffetier to fill gaps left by seventeenth century alterations. The only original explicit-continuous panel groups made by the Relics workshop appear in the top sections of two windows in the Chapel of All Saints - bays 08 and 12. The first of these is a rather half-hearted attempt (see Fig. 9.08B). St Laurence’s execution in panel 18 is overseen by Decius in panel 17. Executioners in panels 19 and 20 assist in the operation in what is clearly supposed to be a continuous scene. Yet the only details that actually span the panels are a few small tongues of flame from the fire in panel 18, which just spill into the bottom of panel 20. Other elements, such as the patch of grass at the bottom of panel 17 stop awkwardly at the armature. A more convincing explicit-continuous panel pairing appears at the top of the St Vincent window (bay 12), between panels 19 and 20 (see Fig. 9.08C). Here the scene in panel 20 where two men are trying to dispose of the saint’s body spills across to panel 19, where we see Dacian giving orders. Although the extending of the oarsman’s arms across two panels has worked well, the overall design still lacks the finesse - and the narrative richness - of the panels at the top of the St Nicholas window (compare Fig. 9.08C and Fig. 9.04). Apart from a few minor details in this same St Vincent window, there are no other attempts at explicit-continuous scenes anywhere in the Bourges ambulatory (none at all by the New Alliance workshop), except in the windows of the Good Samaritan workshop. At Chartres there are a great many implicit-continuous scenes but relatively few where the continuity is made explicit by elements spanning the armature; the fishing net used by the Apostles in bay 00, panels 14-15 (although here the decorative border overlaying the net in panel 14 reduces the effect - see Fig. 3.11A), the titulus spanning panels 03-04 of bay 01 (Fig. 3.11B) and the body of Thomas Becket laid out across the topmost panels of bay 18 (Fig. 9.08A).8 Compared with this handful of examples from other workshops (and there are none at all in the surviving thirteenth century narrative windows at Sens, Le Mans or Tours), explicit-continuous frames feature in every one of the six windows at Bourges attributed to the Good Samaritan workshop.

One advantage of his unusual approach is that the Good Samaritan master had far more flexibility over framing than other stained glass artists. He could work perfectly well with the normal allocation of one {chronotopic} frame to one physical panel - as he did for example in panels 16-18 of the Magdalen window (see Fig. 9.09) but he could always merge two or three panels into a single image as required. The available space could therefore be adapted to the needs of the story telling, rather than vice versa.

[237] Another consequence of this habit of thinking beyond the individual panel is that the artist could exploit more subtle interactions between adjacent frames for dramatic effect. At the dead centre of the Passion window (Fig. 9.06), panel 14 shows the Agony in the Garden. While His disciples slumber on the left, Christ kneels in prayer to the right, his eyes raised towards the heavens. Yet follow his gaze across the border, past the armature and into panel 17 and it becomes clear what he is staring at (see Fig. 9.07B); Christ’s gaze leads straight up the ladder that He will later ascend to be nailed to the Cross. A related characteristic of the Good Samaritan master’s work is that even when treating panels as individual scenes, he still seems to design his formal compositions across the register as a whole. This is particularly noticeable in the Magdalen window where several registers show a strong sense of bilateral symmetry. In the ‘funeral of Lazarus’ scene discussed earlier (Fig. 9.05), Mary on the left and Martha on the right are positioned at opposite ends of the frame and although their gestures differ, as if demonstrating their chosen roles (the vita contemplativa and the vita activa respectively), their inclined heads and identical costumes serve to emphasise the symmetry. Similarly with the exception of the white robed priest overseeing events from the left, their friends and helpers are also paired-off around the central axis, even in the way that the two gentlemen lowering Lazarus into the tomb both step neatly out of the red frame at the bottom. Even when a register is treated as separate frames, the design retains its symmetry. For example panels 16-18 are treated as three separate {chronotopic} frames (see Fig. 9.09). The first shows Mary listening at Christ’s feet, in the middle is Martha working and then, at the end, Martha complaining about her workload. Yet if one views the register as a whole, the two seated figures of Christ at either end, each turning inwards to face a female interlocutor, act as compositional parentheses framing the central panel, where Martha, slaving away in her kitchen, is shown in an almost frontal pose, her bucket and cooking pot again emphasising the symmetry. Yet there is one place in the Magdalen window where this symmetry takes on a strange and wonderful twist. In the scene of the meal at the house of Simon the Pharisee (Fig. 9.10), we find a long table stretching all the way across the three panels, its continuity emphasised by the tablecloth and by a fish extending through the armature that separates the central and right-hand panels. In the central panel, seated behind this long table, are three figures, Simon himself on the left, wearing his conical Pharisee’s hat, plus two otherwise anonymous disciples, identified by their nimbed heads and bare feet. At the far left of the table is Christ, who appears to be holding up a morsel of food and addressing a comment to Simon. Beneath the table, stretched out over two panels, is Mary Magdalen. Having entered Simon’s house during the meal and washed Christ’s feet with her penitent tears, she now dries them with her hair. At the other end of the table, we find Christ once again - and again the Magdalen kneels at his feet, this time anointing them with perfume from a jar she holds in her left hand. The potential for confusion in this panel is greatly compounded by the artist’s limited repertoire of male facial types - it is only Christ’s cruciform nimbus that makes His identity at either end of the table unquestionable. Of the central group of [238] figures, Simon and one of the disciples clearly engage with the Christ at the left of the table. Their companion turns to glance at the other Christ, yet with his right-hand he gestures towards the Magdalen at the other end of the table. The resulting network of glances and gestures guides the viewer’s own eyes in a confused dance around the scene as we try to disentangle the presence of two Christs and two Magdalens at the same table. As a simple explanation, one could argue that is no more than the use of that ‘primitive’ narrative mode, classified by Wickhoff as ‘Continuous’, whereby multiple stages in a story are shown within a single physical space with key characters repeated. However, in none of the examples studied by Wickhoff did characters in the different phases of the story interact with each other as they do here. Far from showing his naivety, the window’s designer has deliberately constructed a clever visual paradox. The chiastic composition draws the viewer in, rather like the examples of metalepsis discussed earlier.9 One’s eyes bounce back and forth, guided by the glances and gestures within the image, in search of a solution to this paradox. This trick serves to reactivate the recipient - a recipient whose attention could so easily be overwhelmed by the richness of the ambulatory windows with their four hundred or so narrative scenes.

One question which is often asked about complex artistic programmes such as stained glass windows is how much was decided by the patrons and how much by the artists? Presumably the overall programme and the choice of subject matter for each window must have been decided by the Cathedral chapter. It is also generally assumed that the choice of episodes within each narrative would have been set out by the clergy. But what about those aspects of design that fall within the remit of visual narratology, such as the physical layout of panels, the arrangement of scenes within them and the use of meta-communicative devices like metalepsis? Was this also determined by the commissioning clergy or was it left to the artists?

One clue that might help resolve these particular questions is Grodecki’s observation that the Relics and New Alliance workshop’s windows all use armature and medallion patterns similar or identical to ones found at Chartres, whilst the Good Samaritan workshop uses patterns not found there.10 This would imply that each workshop carried its own set of designs for armature and medallion patterns. Whilst the choice of patterns may have been made in conjunction with the client and in consideration of the surrounding bays, the geometric design of windows [239] (i.e. the ‘arrangement’ of the polyscenic narrative - see section 2.f.3) can at least be seen in part as a basic index of workshop identity. But what of the interactions between the panels in a window?

As has already been noted, the Good Samaritan workshop, or possibly just its master, also worked at Poitiers. Grodecki’s perfectly reasonable hypothesis was that:

...a master glass worker from Poitiers was summoned to work at Bourges, and there came under the influence of the Chartres school, while still retaining sufficient traces of the Poitiers decorative repertory and the Poitiers traditions of composition for his origin to be recognisable.11

Grodecki reached these conclusions on the ground of traditional connoisseurial considerations of drapery, figure style and so on - but they are also supported by a consideration of the artist’s treatment of narrative framing. Sadly, apart from the justly famous mid-twelfth century Crucifixion window in the middle of the apse, the Poitiers windows are in a sorry state, victims of Huguenot iconoclasm, inconsiderate restoration and general neglect. Several windows have been botched together out of miscellaneous fragments sourced from windows made over the course of two or three hundred years. Nevertheless, enough has survived of the Poitiers glass to recognise the Good Samaritan Master’s distinctive use of explicit continuous panels. The best preserved of the thirteenth century windows - the Joseph window (111a) - has already been discussed in chapter 8.b, where I mentioned the horizontal pole of the gibbet in panel 15 which stretches back into the dream scene in panel 16 (see Fig. 8.12A). Two other windows also display this same narratorial habitus. Three panels now in bay 113b, show the events of Exodus 7:10-13; Aaron casts his rod to the floor and it turns into a snake - Pharaoh’s magicians do the same and their rods also turn into snakes, all of which are gobbled up by Aaron’s (see Fig. 9.11). Amidst the fragments of later windows that have been used as stop-gaps, it is difficult to make firm connoisseurial assessments, although a few surviving fragments of the drapery of Pharaoh’s magician are certainly reminiscent of the Good Samaritan master’s style. The absolute give-away however is the relationship between the panels as the two protagonists poke their rods past the clumsy modern armatures and into the central panel. The design of this dramatic register with three explicit-continuous panels working together in a symmetrical composition is unmistakeable. The one other example that survives is in bay 117a, the Joshua window (see Fig. 9.12). Sadly this window has suffered the insult of unsympathetic restoration on top of the injuries of Huguenot vandalism, but it is still possible to detect the agency of the Good Samaritan master in the way battle scenes spill from one panel into the next. The only intact pairing is in panels 17-18 (Fig. 9.13) though at the very least it looks as if the current panels 04, 05 and 12 also once participated in explicit-continuous groups. Even in its current state, it is possible to see in this window a shadow of what must once have been an exciting and dynamic composition that perfectly matched the dramatic story of Joshua’s battles.

[240] The obvious question with all these examples is whether this use of explicit continuous framing (and associated trickery such as metalepsis and paradox) is the work of one artist or whether it is a characteristic of Poitevin glaziers generally. This latter point will be difficult to prove. More work needs to be done on the different workshops at Poitiers - though the fact that explicit continuous frames appear in some windows but not in others may point to it being associated only with one workshop or master. Either way, it seems likely that this narratological ‘style’ was transported to Bourges, as Grodecki suggested, by the temporary loan of one workshop, or perhaps even just one master, who took his stock of window patterns, his unconventional attitude to framing and his indefinable esprit ludique with him to take over an existing workshop at Bourges (which could explain why not all of the stylistic features of the Bourges workshop are also found at Poitiers).

In some respects, drawing a design that spans the panels of a stained glass window may seem rather mechanical - trivial almost. In reality it is anything but. Designing visual narratives that work across a window rather than panel-by-panel is really something special. It demands a story teller who can, both literally and metaphorically, think outside the box. Clearly there was something extra that was distinctive about the designer of the Good Samaritan window at Bourges - something that was ignored in traditional stylistic and iconographic analyses of those windows. This, I would suggest, is one example of how visual narratology can add an additional and useful level to the more traditional investigations of medieval art.