[73]

3) Ontological and narratological implications of frames

‘That's a great deal to make one word mean,’ Alice said in a thoughtful tone.
‘When I make a word do a lot of work like that,’ said Humpty Dumpty, ‘I always pay it extra’. 1

If everyone working in the humanities followed Humpty Dumpty’s lexical remuneration policy, the word ‘frame’ might be one of the wealthiest in the dictionary. Since the mid twentieth century, this word has suffered from creeping polysemy to such an extent that its utility as signifier is on the verge of being undermined. Just to complicate matters further, throughout this thesis, I will need to use the word ‘frame’ in a number of its different but very specific senses. The following chapter will explore some of these in more depth but to minimise the risk of confusion it makes sense to summarise in advance the various roles that ‘frame’ performs.

3.a) {framing} frames

Throughout this chapter and beyond, wherever the particular meaning of ‘frame’ is not obvious, I will disambiguate it with a suitable prefix enclosed in curly brackets {thus} - for example as ‘{picture} frame’ or ‘{cognitive} frame’. Observant readers will note that this is the same bracketing used in chapter 1 to denote the label assigned to some particular {cognitive} frame or other. As should soon become clear, this is not mere coincidence, since these clarifying terms are themselves cognitive meta-frames, i.e. they communicate to the reader the particular sense in which the word ‘frame’ should be interpreted. Before embarking on the fragmenting of the frame it is worth reflecting that most of its apparently divergent meanings are fundamentally related. As Werner Wolf put it,

...all of the different approaches to ‘frames’ converge in one frame function, namely to guide and even to enable interpretation - be it with reference to everyday experience and communication or to medial performances, artefacts etc.2

With this caveats in mind, one may suggest the following set of {frame} frames which the reader is likely to encounter either in this thesis or in the existing discourses of narratology:

• {Picture} frames; at the level most familiar to art history, a {picture} frame is a physical boundary - a liminal mark that surround and defines an image. It may be a [74] more or less decorative physical support of a different material composition from the image itself, a decorative border made from the same material as the image (vegetal or geometric borders around the main panels of a stained glass window, sculpted cusped quatrefoils around a bas-relief scene, fictive architecture or even a simple black line around a manuscript illumination, etc) or a structural entity in its own right, such as an iconostasis or a tabernacle.
• {Cognitive} frames; individual schemata in the fundamental model of human cognition (deriving mainly from studies in artificial intelligence) described above in Chapter 1.a.
• {Interpretive} frames; a somewhat looser construction than the {cognitive} frame, the {interpretive} frame owes much to the normal sense of the word ‘context’. The New Shorter Oxford English Dictionary helpfully defines ‘context’ as:

The part or parts immediately preceding or following a passage or word as determining or helping to reveal its meaning; the surrounding structure as determining the behaviour of a grammatical item, speech sound, etc.

If one extends this definition to encompass all communicative media, this is broadly the sense in which Erving Goffman considered ‘frames’ in his hugely influential sociological studies of the 1970’s.3 Goffman argued that a correct interpretation of any observed behaviour (whether it be the behaviour of men in job interviews, otters at play in a zoo, or even organisations and governments on the international stage) depended on an understanding of which {interpretive} frame was operative at the time. Such frames are not rigid rules but rather heuristic guidelines which aid the interpretation of the enframed. Goffman’s work has permeated much of the subsequent theorising regarding frames and has, appropriately enough, been interpreted slightly differently by different authors. For the purposes of this thesis however, I will use the term ‘{interpretive} frames’ in particular to describe those situations in which the physical setting or context of an image or sub-image helps guide the viewer towards the correct interpretation of its meaning.

• {Chronotopic} frames; the individual scenes or episodes in a narrative. The word chronotope was coined by Mikhail Bakhtin to describe the characteristics of the temporal and spatial continuum within which the action of a novel is played out.4 Although Bakhtin’s own use of the term was altogether denser (owing more to its application as an analytical tool in what was effectively an early form of genre-studies), I feel justified in adopting it here, not least because etymologically it encapsulates the unity of time and space required to delineate unique ‘happenings’ in a visual narrative. [75] It neatly expresses the innate ‘single-threadedness’ of narrative discourse; if two events occur at different moments in fabula time - or at the same moment but in different locations, then normally they will have to be treated as distinct in any visual or textual discourse. Typically each {chronotopic} frame within a polyscenic narrative will be bounded/delineated by its own {picture} frame, except in the simultaneous/complementary mode, which by definition shows a one-to-many relationship between {picture}frame and {chronotopic} frame. Even in these cases however (for example the Hellenistic Medea krater shown in Fig. 3.01A), the separate chronotopic frames can usually be distinguished by ‘preference rules’ based on grouping, alignment, architectural caesurae or other considerations.
• {Paratextual} frames; the term paratext was invented by Gerard Genette to describe those textual elements outside the core text block of a book which in some way comment upon the text or might otherwise precondition the reader/viewer’s response to it. 5 Some of these elements physically surround the core text and are controlled by the producers of the book; cover illustrations, dust-jacket copy, publisher’s notices, author biog, reviewer quotes, chapter headings, etc - collectively known as the peritext. Other elements only come into being after publication and may be beyond the control of creators of the work (reviews, commentary, unauthorised biographies of the author, etc) - these elements Genette christened the epitext. All paratextual elements may have a significant effect on a reader’s reception of the book but prior to Genette’s study had generally been ignored by literary theorists. 6 Whilst specifically verbal paratexts may not seem very relevant to medieval narrative art, there are situations where they become important, for example when the visual reception of some object takes place during what one might call the ‘performative paratext’ of the liturgy or other ceremonial acts. The idea of the epitext also becomes interesting when considered in relation to the common medieval (and later) practice of re-setting older narrative artworks into new physical contexts.
• {Conversational} frames; an approach to linguistic analysis/interactional studies pursued by Erving Goffman and Paul Ricoeur (amongst others), which treats conversational discourse as a sequence of narrative segments framed by verbal ‘ontological parentheses’ which function to prepare the interlocutor, evoke approval, signal the end of a narrative segment etc.7 [76]
• {Frame narrative} frames: These are the containers for a ‘story-within-a-story’, such as the narrator’s account of leaving Southwark at the start of the Canterbury tales (see also the discussion of diegetic levels and embedded narratives in chapter 4).

In recent art-historical discourse, the status of the {picture} frame has become quite a popular subject. Towards the end of the eighteenth century, in his Critique of Judgement, Kant wrote:

Even what we call ornaments (parerga), i.e. what does not belong to the whole presentation of the object as an intrinsic constituent, but is only an extrinsic addition, does indeed increase our taste’s liking, and yet it does so only by its form, as in the case of picture frames [...]. On the other hand, if the ornament itself does not consist in beautiful form but is merely attached, as a gold frame to a painting so that its charm may commend the painting for our approval, then it impairs genuine beauty and is called finery.8

Kant’s views on frames (and on equivalent parergonal phenomena such as drapery on figures or colonnades around buildings) do seem to have been rather confused, sometimes at odds with his view of the aesthetic object complete ‘in and of itself’. However since they were themselves only peripheral to his main arguments, the flaws in his theory largely went unnoticed, at least until the end of the 1970’s, when Jacques Derrida took a few pot-shots at them in his book La Vérité en Peinture. 9 Bennington and McLeod’s translation of that work into English in 1987 seemed to re-stimulate interest in frames as a worthy topic of discourse.10 Although it has attracted far less comment, an equally significant development in the revival of interest in frames was the rehabilitation of the German sociologist Georg Simmel, whose insightful 1902 essay on the picture frame was finally translated and published in 1994. 11 Fed by contributions from Victor Stoichita and Jean-Claude Lebensztejn, the discourse surrounding the {picture} frame ended the twentieth century in a very healthy state. 12 This renaissance continued into the new millenium, with an impressively interdisciplinary volume on framing, edited by Werner Wolf and Walter Bernhart, which included several essays on {picture} frames.13

Unfortunately however, almost everything that has emerged from this discourse on the {picture} frame, from Kant, through Simmel, to Derrida, Lebensztejn and beyond, is largely irrelevant to the current discussion. The one thing that unites all of those discourses on the frame is that they [77] deal only with the enframing of the kind of ‘Art’ that aspires either to naturalism or else to some more modern goal - and not to objects which operate within the kind of sign systems discussed earlier in relation to the semiotics of medieval images (chapter 1.e above). Moreover they focus on the kind of artwork that is meant to be complete in and of itself - not the kind of objects which were created to participate in the Gesamtkunstwerk of the medieval cathedral. As such the emphasis tends to be on the simplistic finestra aperta model of Alberti (i.e. the {picture} frame as {window} frame), or else on the illusionistic frame; the trompe l’oeil ceiling of Signorelli at Orvieto or the trick canvases of Gijsbrechts for example.14 Even in the most recent collection of essays on framing in various media, the earliest artwork treated in any depth was the 15^th century Hours of Mary of Burgundy; a late medieval work perhaps but one which is famous mainly because of its trompe-l’oeil borders, whilst another essay in the same volume, although it did discuss the famous Genoa Mandylion and its 14^th century re-framing, did so purely in terms of post-Reformation conceptions of the image.15

In truth, the medieval {picture} frame was a far more complex, flexible and fascinating beast than any of the gilded parerga that exercised Kant, Derrida, or their respective followers. Within the medieval sign-system, the nature of {picture} frames was closer to that of a semantic enclave (to use Mieczysław Wallis’s terminology), exercising their own modes of semiosis, rather than simply acting as ornaments for the ergon. Instead, medieval {picture} frames were active participants in the creation of meaning and they performed a range of narrative and meta-communicative functions - as the remainder of this chapter will attempt to demonstrate.

3.b) ontological implications of the {picture} frame

Common to all of the media considered in this thesis is the convention of using some form of boundary marker to define the space occupied by an image or by separate narrative episodes. The formal nature of these boundaries varies considerably, including elaborate fictive architecture, colourful poly-lobed borders, simple black lines and functional support elements (such as the armature of a stained-glass window or the bare-metal edges through which enamel plaques were nailed to a wooden core). These {picture} frames help to construct narrative worlds that are spatially and/or temporally distinct from their surroundings. Some of the consequences of this ontological function, and what happens when it is deliberately violated, will be discussed later, in the sections on diegetic levels and metalepsis (chapters 5 and 6 respectively), [78] but first it is worth briefly considering the origins of such frames, and some of their characteristics in medieval art.

Perhaps the only major theorist to consider the semiotics of the frame across a broad chronological range was Meyer Schapiro. As he observed:

...we take for granted today as indispensable the rectangular form of the sheet of paper and its clearly defined smooth surface on which one draws and writes. But such a field corresponds to nothing in nature or mental imagery where the phantoms of visual memory come up in a vague unbounded void.16

Like Plato’s shadows, the early paintings on the walls of the cave occupied just such a ‘vague unbounded void’. The artists of Lascaux for example, who made their famous images some 15,000 years before Christ, felt no need to delineate the boundaries of their image field. Indeed, much prehistoric art shows a deliberate antipathy towards image boundaries, with later interventions overlapping earlier ones; perhaps as acts of spatial appropriation but also emphasising that there was no privileging of particular areas of the rock surface (an artistic practice preserved in the activities of modern-day ‘taggers’). At the opposite extreme, the early-modern and modern art-worlds came to see framing as a significant action in the creative process - both literally and metaphorically an act of closure. Framing marks the completion of the artwork and asserts its status as commodity. Even artists’ sketches and exploratory doodles can be transformed by dealers into independent artworks by the simple act of framing them. In between these two extremes of prehistoric and modern image making are a range of different approaches to framing, some of which give the frame a more ‘active’ role in the construction of meaning than it typically performs today.

Classical Greek art in various media embraced frames, generally employing the same delineating markers (geometric forms like the ‘Greek fret’ or vegetal forms such as palmettes) regardless of medium or scale, thus marking frames as a property of images in general rather than a characteristic of some particular techne. Often such frames served only to delineate the image field as a whole; that field might contain multiple different narrative and extra-narrative elements, jumbled up in what was characterised by Franz Wickhoff as the Complementary method or by Kurt Weitzmann as the Simultaneous (see chapter 2.f.2), as in the early Hellenistic ‘Medea Vase’ shown in Fig 3.1A. This large unitary image-field contains a main image of Medea killing her children, within an architectural setting appropriate to the action, surrounded by subsidiary scenes, some of which depict earlier or later moments in the story, and some of which depict the Gods on Olympus (here not as part of the narrative but as the omnipresent audience for human folly). Throughout much of Greek art, one tends to find a [79] similarly straightforward use of the frame, as in the horizontal bounding lines of friezes and the box-like rigidity of the metope. By contrast, Roman art, both narrative and otherwise, revelled in the illusory power of the frame.

The art of the mid-first century BC (August Mau’s ‘second style’) favoured frescoes showing deeply recessing architectural capriccios, such as the famous examples from Boscoreale, now at the Metropolitan Museum in New York (see Fig. 3.01B). Here the viewer is challenged to disentangle the myriad layers of fictive architecture and re-assemble them into a plausible three dimensional world - a task which is facilitated by the layers of architectural framing devices, through which one sees deeper and deeper into the imaginary cityscape. Despite of all their perspectival complexity however, in terms of what is being represented, these Boscoreale frescoes employ frames in a simple, one-dimensional fashion; all of the planes of imaginary architecture belong to the same represented world - to the same level of reality. Despite its illusionistic depth, the image is ontologically flat. A very different kind of trompe l’oeil was constructed by the artist who decorated what may once have been the bedroom of Augustus in the imperial villa at Boscotrecase, in Mau’s third style (Fig. 3.01C). Here the artist depicted fictive furniture and illusionistic elements like birds and miniature landscapes, which float ambiguously on the mural surface (like Schapiro’s ‘phantoms of visual memory’). Yet he also depicted fictive depictions - including fictive portrait medallions and a fictive papyrus purporting to contain a painted Egyptianising scene. The crucial characteristic of these pictures-within-the-picture, the elements that invite us to interpret them as artworks which are in some way separate from the wall itself, are their fictive frames. What we ‘see’ on the wall at Boscotrecase is not an Egyptian scene but a piece of papyrus upon which an Egyptian scene has been painted. The fictively framed elements within the Boscotrecase murals work on the same level as the Veronica prayer-sheet pinned to the wall in Petrus Christus’ Portrait of a Young Man in the National Gallery, London (Fig. 3.02A). In each case the presence of the fictive {picture} frame suggests a separate substrate - and that separate substrate implies a distinct and independent image within the main image (one which pretends to be the result of some other artist’s agency). Thus the fictive {picture} frame marks out what it purports to contain as belonging to a different reality from the rest of the image. It fulfils what Gregory Bateson classed as the ‘meta-communicative’ function of frames, since it ‘...gives the receiver instructions or aids in his attempt to understand the messages included within the frame’.17 Even if we temporarily imagine that the room occupied by Petrus Christus’ young nobleman is the real world and not a painting, we would still regard Christ’s face, pinned to the wall, as being a painting and not part of that real world - all the more so for this particular subsidiary [80] image since iconographically every Veronica implies a separate substrate. 18 Thus, the fictive {picture} frame marks the enframed as being ontologically distinct, both from the frame itself and from the world that contains it. As Victor Stoichita put it:

The frame separates the image from anything that is non-image. It defines what is framed as a meaningful world as opposed to the outside-the-frame which is simply the world experienced.19

It is this function of the frame as an ontological marker - as a delineator of ‘meaningful worlds’, that most interests me, since it is key to understanding the role of frames in the reception of narrative images. Frames are the meta-communicative function through which the viewer is able to distinguish narrative environments that are spatially and/or temporally distinct, not just from the ‘outside-the-frame’, but also from adjacent {chronotopic} frames.

By way of an example (Fig. 3.02B), consider an early polyscenic manuscript illumination; folio 125r of the so-called Augustine Gospels (Cambridge, Corpus Christi MS286) - an Italian manuscript of the sixth century. Here, an outer frame, rather crudely painted to suggest marble, contains a grid of multiple sub-images, each with its own frame. We are so familiar with such multi-frame images that we tend not to think about their inherent complexity. Yet it is nothing more than artistic convention that leads us to assume that these particular frames proceed from left to right and from top to bottom, and moreover that each sub-frame within this page represents a chronologically distinct phase taken from a single story - unlike, for example, an iconostasis, which shares the same basic grid format but which constructs an entirely different relationship between the individual images. In the one other full-page illumination that survives in the Augustine Gospels, the author-portrait of St Luke on folio 129v (Fig 3.03A), we face an even greater potential challenge in disentangling the various narrative and non-narrative worlds constructed upon the fictive architectonic frame. Indeed, it is a measure of the effectiveness of frames and of the extent to which they are ingrained in our perception of images, that most viewers simply accept without question or confusion the various modes of semiosis that are built into the different parts of St Luke’s honorific archway (itself a throw-back to the frames of late-classical author-portraits). For example, despite our experience of historiated triumphal arches, few people would interpret the narrative scenes flanking this aedicule as representing polychromed relief carvings occupying the same physical reality as the evangelist seated within it. Instead most viewers will correctly decode the various ontologies represented within this complex image, using the various nested {picture} frames as guides to the shifting levels of reality it contains (Fig. 3.03B).

[81]

3.c) frame systems - {interpretive} macro-frames

In the twelfth and thirteenth centuries, manuscript illuminators often used structured groupings of {picture} frames as templates or schemata, by means of which the viewer could short-circuit the task of decoding what realities the various parts of the page belonged to and how they might interrelate. As an example of this practice, I will turn to one of the last great manuscripts in the Ottonian tradition, the Gospels of Henry the Lion (made around 1185), in which several pages employ broadly similar formal arrangements of framing elements. 20 Folio 20r is typical (see Fig. 3.04). Here, the main part of the page is divided into two large rectangular frames containing sequential narrative scenes - the Magi before Herod in the lower frame and the Adoration of the Magi in the upper one. In addition to these two narrative frames, we also have four non-narrative medallions in the corners of the image area (details Fig. 3.05). These sub-frames contain what might be considered as ‘hypotactic clauses’ - subsidiary elements which in this case observe or comment on the main action. Thus in frame c (lower left) we have King David who, in Psalm 72 prophesied the coming of the Magi, and in frame d (lower right), Balaam, whose prophecy of the Messiah (Numbers 24:17) gave Herod good reason to be nervous about his oriental visitors and the star that guided them. 21 Thus in terms of medieval exegetical practice, these two lower medallions perform a standard historical/typological role, providing straightforward Old Testament prolepses of the New Testament narrative episodes in the two main frames above them.

The upper medallions work rather differently however. Medallion a (upper left) shows a crowned male figure, labelled in gold as sponsus, who holds a scroll on which one can make out the words of Hosea 2:19-20; ‘I will espouse thee to me forever in faith and justice’, while upper right is a crowned female figure labelled sponsa whose scroll carries the words of the Canticles; ‘Let him kiss me with the kiss of his mouth’. 22 Clearly this is not a simple historical typology.

[82]

The Canticles had a special status for the Church Fathers and for the medieval theologians who continued their work. Honorius of Autun (c.1080-1157), in his Exposition on the Song of Songs, explained its rich betrothal/marriage symbolism thus:

With respect to allegory, ... marriage comes about in two ways: one, whereby the Word of God joined flesh with himself, i.e., whereby God assumed the human nature, which, exalted at the right hand of the Father, he has set on the throne of glory; the other, whereby Christ the God-man has associated the universal Church, i.e., the whole multitude of the faithful, with himself through commingling his own body with it...23

Thus unlike the straightforward typology of the lower medallions, these frames represent a ‘higher level’ exegesis of the scene below them; an exegesis operating in the allegorical sense, in which the Incarnation (...whereby God assumed the human nature...), shown in the upper half of f.20r, is seen as the marriage of Christ and His Church. This pairing of sponsus and sponsa appears in the corner medallions on several other folia throughout this manuscript in association with other narrative scenes which also relate to this allegorical tradition, including f.21 (Transfiguration / Entry into Jerusalem), f.75r (Ascension - Fig. 3.06C), f.111r (Annunciation to the Shepherds / Presentation in the Temple) and, most tellingly, f.171v (the crowning of the book’s original owners, Duke Henry and his wife Mathilda, in the presence of the Church Triumphant - Fig. 3.06B).

Returning to the frame structure of f.20r, within each of the four small medallions we also find a set of subsidiary frames, each maintaining a strict semantic relation to its contents (Fig. 3.05). Always we have three concentric frames, the outer one giving the name of its occupant in gold letters on a red ground, then an inner ring for the scroll that bears their speech (in red letters on a white ground) and in the central frame an iconic image of the speaker themselves, their gestures indicating the act of speech recorded in the surrounding ring.

Elsewhere in the manuscript the mise-en-page varies slightly. Some pages retain the same basic layout described above but reverse the sequence of the narrative frames (e.g. f.170v - Fig. 3.06A), while some adopt square medallions instead of round ones (e.g. f.171r, f.75r and f.112r - see Fig. 3.06B, C and D). Other pages use the upper and lower rectangular frames to distinguish between the celestial and earthly realms instead of presenting sequential narrative episodes (e.g. the crowning of the Duke and Duchess on f.171v), while a few pages abandon the large rectangles entirely (e.g. the Majestas Domini on f.172r). Generally however, despite slight formal variations, we find the same basic functional schema repeated throughout, with different frames, identified by their shape, size and/or position within the overall page design, representing the three different levels of reality - narrative scenes (sensus literalis), typological [83] exegesis (sensus historicus) and allegorical exegesis (sensus allegoricus). 24 As a result, with each new page encountered, the viewer quickly learns to identify the formal {picture} frame elements corresponding to the distinct {ontological} frames. Not only does this make interpretation easier, the schema also encourages the viewer actively to seek out the different levels of interpretation, thereby helping to inculcate the exegetical mindset.

This then is the beauty of medieval framing systems. Once the relationships between the various parts of a particular mise-en-page has been recognised, the viewer can re-apply this knowledge heuristically in the deciphering of subsequent pages. In effect the network of subsidiary frames combining at the whole-page level has itself become a meta-communicative device; an {interpretive} macro-frame that simplifies recognition of the subject matter while also encouraging the recipient into a ‘higher’ mode of viewing. It has become an ontological and epistemological map, not just preconditioning the viewer’s reception of each distinct part of the image but also shaping their broader horizon of expectation about the reception of similar objects.

Such {interpretive} macro-frames do not just apply to manuscript illuminations of course (although I shall present several further examples of these later, particularly when discussing the so-called ‘royal Psalters’ in chapter 6.c). Examples in other media are less obvious but remain important, guiding the viewer to the correct interpretation of otherwise ambiguous scenes within polyscenic narratives and aiding their navigation through them. One commonly encountered example of an {interpretive} macro-frame is the practice of locating ‘signature panels’ in the lowest register of stained glass windows - a practice followed in a high proportion of French glazing programmes, particularly those from the first half of the thirteenth century. It is only the possession of the correct {interpretive} macro-frame that instantly allows the viewer to recognise that, for example, the fishmongers in the first three panels of the St Julien window at Rouen (Fig 3.07), or the carpenters plying their trade in that position within the same saint’s window at Chartres, represent the contemporary tradesmen who were in some way associated with that window, rather than active participants within the narrative proper. 25 More importantly, it is the {interpretive} macro-frame that contributes our default expectations of how we should read such polyscenic narratives. Thus when reading a French thirteenth century window like the one in Fig 3.07, the viewer knows to begin at the bottom (probably discounting the first register as signature panels), then read upwards and finish at the top, allowing for the possibility of an extra-narrative image of God in Majesty in the final panel, which may or may [84] not be directly related to the story beneath. By contrast, our imaginary viewer’s English cousin of the same period would probably default to reading from the top downwards and not expect to find signature panels, since that was the local ‘preference rule’ in England at the time.26

3.d) border violations and the porosity of frames

As a container, the medieval {picture} frame often seems rather leaky. Time and again, particularly in manuscript illuminations, one encounters arms, feet, weapons, horses and many other details which overlap or in extreme cases extend well beyond the supposedly bounding lines of the frame. Whilst in a very few cases this may simply have been carelessness on the part of the artist (and in others it resulted from borders being drawn retrospectively around images that originally lacked any) the vast majority of these ‘border violations’ must have been deliberate.

In manuscripts (perhaps the medium with the longest and most nearly continuous record), the birth of the {picture} frame broadly coincides with the shift from roll to codex. Given the fundamental importance of that transition for manuscript scholars, it is understandable that the behaviour of frames has not been top of their list of research interests yet this is a good area to begin looking for leaky frames. 27 An admittedly superficial survey seems to suggest that the practice of drawing over the frame correlates with the movement away from the naturalism of classical art. It may be that just as the fictive frame helped in the quest for illusionism in Roman art, so the gradual move away from illusionism created a new role for the frame as an active participant in the narrative image, rather than simply its boundary marker (the frame of its finestra aperta). A comparison of the surviving few late-classical illuminated manuscripts is revealing (the Vienna Genesis can be excluded from this list since its images are unframed):28

• The Quedlinburg Itala (Berlin, Staatsbibliothek Preussischer Kulturbesitz, Cod. theol. lat. fol. 485) - possibly made in Rome in the second quarter of the fifth century. Six leaves survive, containing four illuminations from what appears to have been an extensive series illustrating the Books of Kings. The surviving page (which shows 1 Samuel 15:13-33), features four regular square frames with a solid red outline which is never broken. Although of course a single page is not enough to prove there were no [85] border violations in this manuscript, the classicising, illusionistic style of the illustrations mitigates against them.
• The Ilias Ambrosiana (Milan, Bibl. Amb. Cod. F.205 inf) - a copy of the Iliad from around the end of the fifth century with fifty eight surviving miniatures in a Hellenistic style. Compositions range from single figures to busy battle scenes, all neatly contained within a double border - which is never overlaid or otherwise violated.
• The Vergilius Vaticanus (Vatican, Bibl. Apostolica, Cod. Vat. lat. 3225) Early fifth century copy of the works of Virgil - seventy-six surviving leaves containing fifty illuminations, all of which have solid borders consisting of a thin black band surrounded by a thicker red band. In every single case, the images are neatly contained within the frames - sometimes touching, or overlapped by the inner black frame but never overlapping it.
• The Vergilius Romanus (Vatican, Bibl. Apostolica, Cod. Vat. lat. 3867) - one of the most complete early texts of Virgil and also arguably one of the first extant manuscripts to escape from an obsessively classical pictorial style. All but one of the nineteen surviving illuminations are contained within thick rectangular frames made of multiple bands of colour or gilding. In four of the eighteen framed images the artists have deliberately included elements that spill beyond these frames (Fig. 3.08A); the shepherd’s hovels in folios 11r and 16v extend markedly into the side margins, while on f.108r (Dido and Aeneas sheltering from the storm in a cave), the head of the spear held by a seated Trojan pierces the top of the frame and protrudes into the upper margin. Also, in the rather clumsy depiction of the fight between Aeneas and Turnus on f.188v, the shaft of the latter’s spear sticks out behind him into the border (where it has been partly effaced by a later binder). 29 Note that in each of these cases the artist could easily have avoided such spillage, indicating that the extension beyond the frame was the result of a conscious decision.
• The Ashburnham Pentateuch (Paris, Bibl. Nat. MS nouv. acq. lat. 2334) - a late sixth or early seventh century manuscript containing the first five books of the Old Testament.30 Eighteen full page miniatures survive out of an estimated sixty nine, painted in a post-classical ‘conventionalising’ style. On most of the illuminated pages an outer frame consisting of a thin red line with curlicues at the corners, encloses multiple scenes distinguished either by simple subsidiary frames or else by different colour blocks (see Fig. 3.08B). Both inner and outer boundaries are often transgressed, the outer frame generally less so, though some of these transgressions are quite [86] spectacular and visually arresting. For example, the sequence of events depicted on f.25r (Fig. 3.08B) ends in the bottom right hand corner of the grid, with Jacob lying asleep in Haran, having his dream about the ladder (Gen.28:11-15), which is depicted above and beside the sleeping figure. Having woken from his dream, Jacob neatly steps right out of the picture frame and into the border, ready to walk off in the way of the Lord. Even more impressive is f.76r (Moses receiving the Law), where the towering peaks of Mt Sinai smash right through frame at the top of the page and into the border beyond (Fig 3.08B).

Ignoring carelessness, one can suggest five main reasons why an artist would deliberately extend his images over the frame boundary:

1. To suggest movement either into or out of the frame, as with the example mentioned above of Jacob walking out of the Ashburnham Pentateuch. As Meyer Schapiro pointed out; ‘...a figure represented as moving appears more active in crossing the frame, as if unbounded in his motion’.31 This later became a standard fixture of annunciation scenes, even in the days when artistic style precluded other means of showing the urgency of Gabriel’s message (see the example from the twelfth century Eichstadt Psalter in Fig. 3.09A).
2. To emphasise the general scale of some element or feature - as with the Ashburnham Mount Sinai bursting through the top of its frame.
3. To contribute to a rudimentary sense of depth, according to whether elements cover, or are covered by the frame edges. This practice seems to have emerged in late classical ivory consular diptychs, where one frequently finds elements overlapping the edge of the decorative frame, quite strongly in some cases; for example, the diptychs of Probianus (c.400, Berlin Staatsbibliothek) and of Boethius (c.487, Brescia).32 Where it occurs, this overlapping usually gives an impression of the figure emerging forward out of the plane of the ivory plaque. The desire to create an illusion of depth in this way may well be the origin of frame violation as a wider artistic practice Once accepted in iconic imagery, it would have been but a short step to realise the narrative potential of the device.
4. creating spatial distinctions between inside and outside the frame, either in terms of simple physical space or with overtones of moral alterity (e.g. inside good, outside bad). A particularly nice example of simple spatial distinction is the star which led the Magi, shown hovering above the decorative border of the Epiphany scene in the Gospels of [87] Henry the Lion (visible at the top of Fig. 3.04). This trick of using the frame and border to extend the narrative stage beyond its proscenium was not limited to northern Europe, as the example from an eleventh century Byzantine manuscript shown in Fig 3.09B nicely demonstrates.33
5. creating ontological distinctions between inside and outside the frame. As discussed later, frames are a convenient way of distinguishing different ‘mode of being’. A well known example of this use of the frame is in those illustrations to the Book of Revelation, where John is shown standing outside a {picture} frame (with or without an angel to guide him), from where he can safely experience the visions shown within it.
6. creating a sense of general chaos, particularly in battle scenes. The Morgan Picture Bible contains some of the most extreme demonstrations of this practice (see Fig 3.09C for just one example out of dozens), though there are plenty of other examples - such as the marvellous scenes of the Trojan wars spilling out of their roundels in the English genealogical rolls shown in Fig 3.09D.34

3.e) narratological implications of the frame

Some image fields, such as tympana, are born framed. Some, like manuscript illuminations, achieve frames. Others have frames thrust upon them. For stained glass, the frame was a structural necessity. The great twelfth and early thirteenth century single-lancet windows, such as those in the choir at Bourges and in the aisles and choir at Chartres are, on average, 2.3m wide, with two or three panels across at each register. Even with the aid of horizontal glazing bars, such wide expanses of heavy glass and lead cames could never be self-supporting. Instead the window had to be broken up into smaller structural units of less than one square metre or so, which could then be slotted into a rigid (and intrusive) iron armature. Without these armatures, there could be no great windows. The glass artist was therefore obliged to construct his image sequences in ways which worked with the frames. Of course they were not restricted to using only the patterns created by the armature. Although art historians often talk about ‘armature patterns’ in these windows, the majority of twelfth and thirteenth century windows use a combination of decorative grounds and borders to create internal framing arrangements that modify the basic form of their metal supports and which often dominate the overall visual appearance of the window. For example, the justly famous ‘Incarnation’ window in the centre of the west façade at Chartres (bay 50), made around 1155, has an iron armature which simply [88] forms a regular square grid. However when one views that window in most lighting conditions, the dominant geometric pattern is determined by the roundels that fill alternating squares, together with the vesica pisca form housing the Sedes sapientia at the top (see Fig. 3.10). It is the medallion of the glazier, not the blacksmith’s armature, that frames the narratives of stained glass.

Architectural sculpture is necessarily constrained by architecture. It is a commonplace in the historiography of medieval art that the transition from Romanesque to Gothic was marked by the ‘liberation’ of sculpture from its architectural setting; that figures no longer needed to be distorted and twisted to fit within the tympana of churches like Vezelay or Autun. Whilst there is a grain of truth in this, the liberation of sculpture from its architectural bonds was rather a Pyrrhic victory. With the notable exception of Rheims cathedral, the new frames for narrative sculpture tended to be as inflexible as their predecessors. Though the quantity of sculpture burgeoned, the content remained strictly controlled by the syntagma of the Gothic portal, with its clearly demarcated narrative and hieratic zones.35

Different media and contexts of display developed their own ways of using frames. The simple rectangular grid seen earlier in the Augustine Gospels (Fig. 3.02B) faded somewhat as more adventurous patterns were explored, though it re-asserted itself strongly towards the end of the thirteenth century, becoming the norm for wall painting, stained glass and polyptych altarpieces alike as visual narratives generally became more ‘readerly’.

As discussed earlier, the default assumption with {picture} frames is that each one represents a distinct {chronotopic} frame - a self-contained world either ontologically or temporally distinct from its neighbours. In practice however this need not be the case and in the case of polyscenic narratives the opposite is often true. In such systems, adjacent {picture} frames can relate to each other in any one of six distinct modes (some of these are demonstrated in Fig. 3.12, which also introduces a suggested notation for ‘narrative flow’ diagrams):

• Explicit continuous scene; the image stretches across two physically distinct panels as if the armature/frame-boundary was not there. One or more visual elements (persons, props, scenery, tituli etc) span both frames, as if the frame boundaries had been retrospectively superposed onto a single panel. The extent to which the spanning element is obscured by the frame boundary varies. Take for example the miraculous draught of fishes as shown in the Chartres Apostles window (bay 00, panels 14 and 15 - see Fig. 3.11A). Here the net stretched between the two boats disappears ‘behind’ the armature and also behind the decorative border of panel 14. By contrast, in the scene of [89] Simon and Jude arguing with the Persian sorcerers in the adjoining window (bay 01, panels 03 and 04), the spanning element, in this case a titulus giving their names, overlays the decorative border and is obstructed only by the metal armature (Fig. 3.11B). Generally, artists arrange things so that nothing important is actually obscured by the frame border - the aim is to indicate the contiguity of space between the panels without drawing attention to it. There is a magnificent and deliberate exception to this subtlety in the Genesis cycle on the dado frieze at Auxerre, where Noah’s Ark barges its way between two cusped quatrefoils as if seeking to emphasise the artificiality of the frame boundaries (Fig 6.45). As we will see later (chapter 6.e), this is just one of several examples of how the sculptors at Auxerre ‘played’ with frames.
  
  
• Implicit continuous scene; no overlapping elements but gestures, postures etc employed to denote that the two panels constitute a single episode. For example in the Incarnation window at Chartres (see Fig. 3.10), the meeting between the Magi and Herod is split between panels 05 and 06, the Adoration of the Magi is split between panels 07 and 08 and Christ’s entry into Jerusalem, accompanied by His twelve disciples, is split between panels 22-24. In each case nothing overlaps the border between the frames and it is only poses and a knowledge of the story that makes the continuity of the panels obvious.
  Depending on the needs of the narrative, the two frames, whilst representing the same chronological moment within the narrative, can be used to suggest a continuous space (as in the previous example) or conversely to suggest two distinct but adjoining spaces, such as in the dado at Auxerre, where David, in his palace spies on Bathsheba in a neighbouring house, (see chapter 6.e and Fig. 6.44). A common variant of the implicit-continuous mode is the use of hypotactic frames for censing angels, who generally occupy a frame off to one side of the main image, swinging their censors towards an image of Christ in Majesty or some other extra-narrative image but without transgressing the frame. See for example panels 25 and 25 in Fig. 3.10 (contrast this more typical form with the censing angels in the windows of the Bourges Good Samaritan Master, whose censers swing behind frame boundaries and armatures without hindrance. For more on this master of the explicit-continuous scene, see chapter 9).
  
  
• Discontinuous sequential scenes; the two panels show distinct {chronotopic} frames from the same narrative segment. The second frame, while belonging to the same segment, might not follow immediately from the first in terms of events in the underlying fabula, in which case the boundary between the physical frames constitutes a narrative ellipsis.
  
  
• Discontinuous but explicitly related scenes; the two panels show distinct {chronotopic} frames from the same narrative segment - but through gestures, [90] posture or conventions of composition, the figures in one scene indicate that the other scene is directly related to it. This relationship normally takes one of two forms:
• Showing causality. For example in the window of saints Simon & Jude at Chartres (bay 01), in panel 11, Baradach, advised by his magician, instructs a guard to arrest the saints - the guard gestures right to panel 12, where the two saints are shown being imprisoned (Fig. 3.11C). The combined composition clearly implies that the events shown in panel 12 are a direct consequence of the events in panel 11.
  
• Indicating ‘diegetic embedding’ (see chapter 5), such as reported speech, the topic of conversation or the content of dreams or visions.
  
  
• Narrative breaks; the transition between consecutive frames marks the end of one narrative segment and the start of a new one. The material depicted in some types of polyscenic narratives (particularly hagiographies based on gestae and libelli) was assembled from short discontinuous episodes. Typically, saintly image cycles depict several such sub-stories, usually with no formal marker to denote the jumps. With more continuous narratives, such as the Gospels, the narrative breaks often correspond to ellipses and are marked in the text with conventional formulae such as ‘and then it came to pass’ or ‘some years later’. Generally there is no visual equivalent for this - so for example in the Chartres Incarnation window (see Fig. 3.12), there is nothing to indicate the thirty years or so that are elided between Christ’s return from Egypt in panels 17-18 and His Baptism in panel 20.
  
  
• Reiterative groups; consecutive frames are distinct/non-continuous but contain images which are either identical (sometimes re-using the same cartoon) or at least functionally equivalent in narrative terms. Frames repeating the same elements or actions are sometimes used to give greater emphasis to some key narrative event in what is effectively a visual form of the rhetorical device of amplificatio. For example the Noah window at Chartres (bay 47) includes several reiterative groups showing the antediluvian populace (panels 04-08), the loading of the Ark (13-16), the men and beasts drowning in the rising waters (18-21 - see Fig. 3.13B), the corpses exposed by the receding waters (24-25) and the disembarkation (27-30). Interestingly, the later sculptors of the west façade of Bourges cathedral did much the same thing with the dado spandrels - in a Genesis cycle that does not include any other repeated scenes and elides certain important episodes, there are no less than four distinct frames showing the victims of the Flood (right embrasure of the inner-right portal - spandrels 24, 26, 28 and 30b - see Fig. 3.13A).

Although the preceding modes cover the majority of situations, there was always scope for subtle twists as artists experimented with ways of enriching polyscenic narratives. One such [91] trick is the ‘hinge-function’ - a triangular relationship in which one panel shares an implicit continuous relationship with two different {chronotopic} frames, even though those two frames are quite distinct. A particularly elegant example of this occupies the whole of the seventh register of bay 28b at Chartres - the Life of the Virgin window (see Fig. 3.14). On the left is Herod with his astrologers. On the right is the Adoration of the Magi, with the eldest Magus kneeling before the Virgin and Child to present his gift. The hinge panel in the middle shows the remaining two Magi together with the star (upper right) - both of them turn towards the left panel to converse with Herod but the younger one points right towards their companion in the Adoration scene. Thus the two Magi in the central panel are simultaneously participants in two separate {chronotopic} frames - the Magi before Herod and the Adoration.

Another clever story-telling device - one which adds an extra {chronotopic} frame without increasing the number of images - is what I call the ‘bounce-back’. A good illustration of the device can be found in the window at Chartres dedicated to St Thomas the Apostle’s trip to India (bay 23, situated in the first chapel of the northern ambulatory). Having been sent by Christ to evangelise the Indies, Thomas is attending a feast in honour of the king’s daughter’s nuptials - the Golden Legend describes what happens thus:

The wine steward meanwhile, noticing that the apostle was not eating or drinking but sat with his eyes turned towards heaven, struck him a blow on the cheek. The apostle addressed him: ‘it is better for you to receive here and now a punishment of brief duration, and to be granted forgiveness in the life to come. Know that I shall not leave this table before the hand that struck me is brought here by dogs.’ The servant went out to draw water, a lion killed him and drank his blood, dogs tore his body to pieces, and a black one carried his right hand into the midst of the feast. This greatly disturbed the whole company.36

This sequence, which derives from chapter 3 of the apocryphal Acts of Thomas (based on a third century Syriac Gnostic text) has five basic {chronotopic} frames which could potentially be illustrated:

1. The wine steward strikes Thomas on the cheek
2. Thomas addresses the Steward (the prophecy issued)
3. The servant, having gone outside to fetch water is attacked by a lion and some dogs
4. A black dog carries the steward’s hand into the feast (the prophecy fulfilled)
5. The whole company is disturbed.

The artist responsible for bay 23 has shown all but one of these {chronotopic} frames in just two panels. The sequence starts at the left of panel 05, where the steward strikes Thomas with his right hand. No attempt has been made to show Thomas speaking in response. The action [92] then cuts to panel 07, which sticks close to the Golden Legend account - the setting includes trees, vegetation and a spring (lower right) besides which the steward has dropped his cup. The steward himself lies on his back being mauled by a lion (note the use of red-streaked glass for his blooded face), while a yellow dog bites at his right wrist. The action then bounces back to the interior feasting scene in panel 05, where we see the yellow dog in the foreground entering from the right and carrying the steward’s right hand. At the right of the table one of the guests points to the dog and converses with the king seated alongside him, both of them looking suitably disturbed. Apart from the colour of the dog (black dogs being particularly unsuited to depiction in stained glass) the artist has followed the text very closely.

The necessary ingredients of bounce-back then are a panel showing events at two distinct moments (A and C) but at the same location, with a second panel showing an intervening event (B) at some other location, which provides the missing link in the causal chain A-B-C. It is a form of Weitzmann’s ‘complementary’ image - but a particularly clever one. An even more subtle bounce-back of this type will be explored in the next chapter when looking at the relationship between panels 18 and 19 of the Charlemagne window at Chartres.