Journal of Early Christian Studies

SCHOLARLY ANTECEDENTS OF THE CANON TABLES

As a tool to provide a comprehensive indexing feature, the Eusebian apparatus represented a major bibliographical advance.⁷ But the full consequence of the Eusebian apparatus can only be appreciated in the context of the larger scholarly environment from within which they were derived and the theological, exegetical, and apologetic needs of early Christian communities, especially those in and around Caesarea where Eusebius was centered, that they were intended to address. The canon tables emerged from a tradition of Christian scholarship extending back at least to Origen, and this wider intellectual-historical context is important because it helps to elucidate how the tables would have been received.

Although Eusebius tells us little about the legwork that went into crafting his tables, he drew upon two works that were clearly critical to his own efforts.⁸ The earliest prototype of the system of cross-correlation that Eusebius developed was Origen’s famous Hexapla, the text-critical, columned version of the Old Testament that included the Septuagint with other Greek translations next to the original text in both Hebraic and Greek transliterative versions. As Barnes and others have noted, the Hexapla was revised and disseminated by Pamphilus and Eusebius from their perch in Caesarea.⁹ The significance of the Hexapla to Eusebius’s own scholarship is thus in part as a bibliographic prototype, evidenced by the canon apparatus and some of his other efforts, like the Chronicles, which also used a tabular format to collate and present information.¹⁰ But additionally, it spoke to the need for bibliographic tools that could serve the needs of early Christian [End Page 406] scholars. As Ruth Clement has persuasively argued, the text-critical apparatus of the Hexapla made available a scholarly tool both to intercede in intra-Christian theological controversies as well as to counter critics outside of the faith, especially Jewish polemicists.¹¹ The columned form allowed Origen to create a recension of the LXX that could thereby serve, not only the apologetic, but also the polemical and exegetical needs of the third-century Christian community in which Origen lived and worked.¹² The novelty of the Hexapla was that it showed how form could create function; it provided a model for assembling such bibliographical tools, and demonstrated their importance in the context of vibrant internal and external debates by sharpening the impact of Christian apologetics. The downside of Origen’s work was its enormous size and expense; since the correlated texts were collected together, the full version would have run to many volumes, constituting “a veritable library in itself.”¹³ Thus, the Hexapla also demonstrated the need for tools that could provide robust text-critical approaches but remained lithe enough to allow the texts to retain their utility and accessibility.

The second scholarly inspiration for the canon tables was a concordance drawn up by Ammonius of Alexandria, a work known as a result of Eusebius’s acknowledgment of this effort as a forebear of his own. In the Letter to Carpianus, he writes:

Ammonius the Alexandrian, having employed much industry and effort (as was proper), has left us the fourfold Gospel, placing the corresponding passages of the other evangelists beside the Gospel of Matthew so that the continuous thread of the other three is necessarily broken, preventing a consecutive reading.¹⁴ [End Page 407]

This passage draws attention to two important features regarding the crafting of the tables. First, it is clear that the canon tables were a deliberate expansion upon earlier efforts to build a text-critical approach to the Gospels. Matthew Crawford has recently suggested that the production of the tables was a “joint scholarly enterprise” between Eusebius and Ammonius, whom he argues (following Zahn) is likely the same person referenced by Eusebius in the Ecclesiastical History and would thus be tied to the Alexandrian sphere of Origen.¹⁵ Crawford advances a plausible argument for an Ammonian “diatesseron” that was not a harmony like Tatian’s, but rather hexapla in form as suggested by Eusebius’s description.¹⁶ Whatever the case, the tables were clearly designed to address extant concerns surrounding the problem of how to read the four texts of the Gospels as a single text, a matter of some urgency at least for the Christian scholarly community centered around Caesarea. Origen himself had laid out the issue directly, noting that what critics “fail to understand is that just as he is the one whom the many preach, so the gospel written by the many is in its effect [τῇ δυνάμει] one. There is truly only one Gospel, though written by four.”¹⁷ The exegetical issue engages, at a basic level, the question of reading, which is the technical limitation of the Ammonian harmony that Eusebius identifies in the Letter to Carpianus.

Thus, the second element of interest in the Carpianus text is that it draws attention to the specific problem of reading the text, suggesting that the Eusebian innovation was deliberately crafted to address this problem. It seems clear that the Ammonian harmony did not exploit the possibilities of the codex in the way that the Eusebian apparatus did. In devising his tables, Eusebius made use of (to adapt a phrase from Peter Head) a bibliographical software to capitalize on the opportunities of a codicological hardware, since the tables required the use of the developing codex format for the Gospels.¹⁸ Unlike the traditional roll, this format easily permitted, [End Page 408] indeed invited, non-linear reading. Moreover, the use of the codex connects to a more general shift in the culture of reading that was propelled by such examples of bibliographical innovation to exploit new modes of approaching texts made possible by the codex.¹⁹ The tables thus offered a dramatically new way of approaching the Gospels, and the innovation, in contributing substantially to the success of the codex, gave not only theologians and exegetes, but also scribes and illuminators, a new, preponderantly Christian, medium for their sacred writings.²⁰

Eusebius’s achievement with the tables was thus twofold: first, the adaptation of and innovation upon existing organizational models to address the synoptic problem that replaced the diatesseric form. This allowed for the integration of all the Gospel accounts into a single text, thereby fulfilling the mandate laid out by Origen to capture textually the concept of single evangelical witness across the Gospels.²¹ The innovation, drawing upon the non-linear possibilities of the codex, went beyond the organizational to encompass exegetical and theological dimensions that were implicit within the so-called Ammonian sections.²² As Thomas O’Loughlin has observed,

the Apparatus was a tool that could be used not only in apologetics but also in answering the problems, internal to Christian theology, of the single gospel and the four texts. . . . [It] met the needs of the Christian reader, [End Page 409] while declaring that differences, where they occurred, are not discrepancies but additions needed for the whole truth.²³

While the degree to which the apparatus inspired future exegetical enquiry is unclear, it was elaborated at a time and place where this was a vital need and thus constituted a major advance in the overall scholarship surrounding the redaction and production of the canonical texts.

Once devised, the canon tables, for practical purposes, had to be placed at either extremity of the text. That they were placed at the beginning and not at the end is worth considering since this suggests something about how their function was understood. Unlike the Hexapla or what can be gathered of the Ammonian diatesseron, the tables contained no actual text. Instead, they were a profane instrument that made accessing the text easier. That they were nonetheless deemed fit to be placed at the opening of the Gospel likely reflects contemporary recognition of their importance, both for the scholarship that had gone into crafting them, and for the profound way in which the apparatus changed how the Gospel could be read. Here, for the first time, was the opportunity for a Christian reader to have access to the full witness of Christ so that, as Eusebius put it in his Letter to Carpianus, “you may know the specific passages in each Evangelist in which they were compelled by love of truth to say the same things . . . . You shall immediately discover from those [words] prefixed at the head of the canon, both how many and which ones (of the Evangelists) spoke concerning what you are seeking (ζητεῖς).”²⁴ That idea of ζητέω I think explains why the tables were given an exordial rather than a terminal role. The tables offer a way to gain access to the text that the text itself cannot provide, and as such they are something the reader traverses to acquire the truth of the Gospel. No mere bibliographical exordium, they are, in this way, a propyleum into the text itself.

That they served this role from the very beginning is attested to by the surviving evidence from the first Gospel that we know of to incorporate the Eusebian apparatus: the Codex Sinaiticus, the earliest of the extant uncial codices.²⁵ While there is general agreement that the Codex Sinaiticus dates to the mid-fourth century, Skeat artfully revived the conjecture that [End Page 410] the Codex Sinaiticus was made as part of Constantine’s famous request to Eusebius in 331 for fifty copies of Scripture for distribution in Constantinople, before being ultimately rejected in favor of the smaller format of the Codex Vaticanus.²⁶ Pertinent here is Skeat’s careful reconstruction of the material requirements needed for their production.²⁷ Outside of their dimensions, one of the most marked differences between them is that the Sinaiticus includes the Ammonian sections while the Vaticanus does not.²⁸ The actual canon tables, however, are missing. Skeat notes that

the Old Testament concludes with Job, ending with the last leaf of quire 72. . . . The first quire of the New Testament was numbered οδ (= 74). . . . The quires on either side of this point are in perfect condition, and it is thus impossible that quire 73 could have simply dropped out of the manuscript. What, then, has happened?²⁹

Skeat suggests the omission was deliberate to save parchment (although he admits that the difference would be marginal) based on the fact that the sectioning does not extend to the entire Gospel, but instead cuts out in Luke, which he links to the decision not to include the tables.³⁰ It seems equally plausible, however, that the tables were drawn up and included [End Page 411] in the original execution of the manuscript, but that since Sinaiticus was, for whatever reason, abandoned and not actually bound until much later (probably in the sixth century, when it was collated with a text of Pamphilus), the quire containing the tables was simply removed and put to some other use.³¹

Setting aside whether or not the Eusebian tables were actually made for the Codex Sinaiticus, the point is that, from the very beginning, they were granted the privileged position as the gateway to the Gospel. If they were, in fact, drawn up and then eventually separated from the manuscript, then the conjecture that they furnished the protocol for placement and a design archetype is not unreasonable.³²

ANTECEDENTS OF THE CANON TABLE DESIGN

What was this archetype? And why did the tables require any such decorative embellishment at all? (Indeed, in most Coptic examples, there is none.³³) It is true that the shift to the parchment codex meant that the possibilities for illumination were significantly greater than for either the traditional scroll or the papyrus codex. And the derivation of a new bibliographical instrument to introduce the Gospel texts likely raised a basic graphic design question.³⁴ Beckwith suggested that “the purely decorative illuminated ornament derived originally from the desire to have a monumental framework” to house the tables.³⁵ But was it, in fact, “purely decorative”—in other words an aesthetic issue—or was the mise-en-page of the tables connected to the way in which they changed how readers of the Gospels could approach the text? [End Page 412]

Further, why was the frame architectonic in form and why did that form become standard? To borrow a phrase from Finney, the canon tables share a common visual syntax, namely the use of arches and arcades within a broadly aedicular motif, which points to a common source since this is the only form by which the tables are decoratively framed in early illuminated Gospel manuscripts; thus, it is clear that this syntax was developed for an archetype from which later versions derive.³⁶ Nordenfalk adduced a number of possible antecedents that, based on stylistic grounds, could have served as models for the decorative scheme of the canon tables. There are many examples of architectonic representation of arches and arcades, including sculpture, carving, frescoes, and mosaics, of which the canon tables might reasonably be said to be a simulacrum. But attribution through stylistic similarity alone cannot explain why such a form was adopted in the first place since there was no convention that linked such designs with manuscript illumination.³⁷ Other texts that featured tabular material, such as Origen’s Hexapla or Eusebius’s own Chronicles, did not have any embellishment at all based on the surviving examples we have.³⁸ The only example of a similar design that predates the earliest known table illuminations is the so-called Filocalus Calendar of 354 c.e., but this is a [End Page 413] very unlikely candidate to serve as a precedent.³⁹ For one thing, it is probable that the design protocol of the canons had already been worked out by this point. Even if it had not, the Calendar of 354 had no Christian significance, raising a basic question of influence and transmission. And while it is possible that the Calendar’s design represents the sole surviving example of a more widely employed architectonic trope for textual embellishment, it seems improbable. In short, the design of the canon tables appears to be sui generis whether placed narrowly within a Caesarean textual tradition or more broadly within a wider tradition of manuscript illumination in antiquity.

That said, if the design appears sui generis within a textual tradition, the archetypal syntax did not spring from nothing. Instead, it was drawing upon the widely employed use of aediculae as visual frames, through an adaptation that Weitzmann characterized (rather obliquely) as the “transformation from monumental [fresco] art . . . into miniature painting.”⁴⁰ But we are left with the question of why the Greco-Roman aedicula was repurposed for the specific needs of a Christian text. And what accounts for the considerable stylistic latitude we find emerging in the various families of the canon tables? Certainly, the earliest Armenian or Syriac table designs look nothing like those of the Western or Greek illuminated traditions. And the Filocalus calendar aside, the migration of the aedicular motif from wall frescoes, mosaic pavements, and monumental or sarcophagal carving to the codicological page is known exclusively from the table designs in the Gospels (although other instances may, of course, have been lost over time). This suggests that there was a specifically Christian interest in the form.

One possible explanation may start with the emergence of a specifically aedicular iconography in Jewish synagogues after the destruction of the Second Temple in 70 c.e.⁴¹ The aedicular design of the fresco that encloses the Torah niche at Dura-Europos is the best-known example.⁴² But similar [End Page 414] exemplars permeate synagogue architecture, iconography, and decoration during this period (third through sixth centuries), including, notably, the façade of the Kfar Bar ʿ am synagogue, and mosaic pavement motifs found in the synagogues of Beth She ʿ Arim, Beth ʿ Alpha, and Hammat Tiberias, among others.⁴³ It is likely that such floor mosaics were part of a larger iconographic program, which included wall frescoes, as at Dura-Europos, and other liturgical and scriptural ornamentation.⁴⁴ Based on the layout, it seems clear that the intent of the decorative program was not simply to underline the sacredness of the synagogue as a whole, but further was intended as a way of foregrounding the liturgical moment, which was focused around the Torah shrine, typically enclosed by an aedicula.⁴⁵ As Meyers and others have suggested, the spatial solemnity of the physical layout to thus concentrate focus upon the text itself (i.e. the reading of the Torah scrolls) appears often to have been emphasized by the presence of a raised lectionary bema.⁴⁶ While there is still a lively scholarly debate surrounding the construction and the evolution of design in post-70 c.e. synagogues, it is clear that the evolution of their iconographical program [End Page 415] reinforced the centrality of the Torah shrine to highlight the importance of Hebrew Scripture to both religious and community identity.⁴⁷

For present purposes, the main interest is that the archaeological record attests to the evolution in Jewish synagogues after the destruction of the Second Temple of an aedicular space—the Torah shrine—which enclosed the Torah niche.⁴⁸ This created a relationship between the architecture of a ritual space and the text as a sacred object. The aedicular architecture of the Torah shrine is also represented both above the shrine itself, as with Dura-Europos, and before it, as with the mosaic pavement at Hammat Tiberias (with additional ritual objects, such as the menorah, the four species, and the incense shovel).⁴⁹ Such depictions intensified the liturgical meaning of the shrine, whether meant as a representation of the Torah ark itself, the First Temple, or, possibly, both.⁵⁰

As a number of scholars have argued, the ornamental, narrative, and architectonic features of synagogue design, starting in the second or early third century, were being worked out in an environment where Christian and Jewish scriptural and liturgical differences were being ever more sharply defined.⁵¹ Herbert Kessler, for example, concluded his study of Dura-Europos by noting that Jews

discovered that narrative art—accessible to diverse audiences—was particularly forceful in asserting the historical claims before their antagonists. Christians—some of them converted Jews—then responded by adopting the same system and even specific images, but turning them to their own purposes.⁵²

Thus, the Christian adoption of a narrative approach to space drawn from Jewish antecedents repurposed specific elements of the narrative scheme. And while it may be that aedicular decoration of the Torah shrine was meant as symbolic, the point still holds insofar as the symbolic center is the culmination of its narrative surroundings.⁵³ Other instances of the aedicular form, such as those found in mosaic floors, also embedded the [End Page 416] symbolic as part of a larger narrative sequence, since the aedicular motif was typically placed outside of the apse or room which housed the Torah niche.⁵⁴ The sacred meaning of the niche is thus reinforced by an accompanying symbolic representation of its liturgical significance that stylizes the features of the niche itself. The value of such ornamental aediculae was critical then to explicating the symbolic meaning of a ritual space.

The iconographic features of ancient synagogues point to a narrative scheme that was embedded within a spatial practice that redounded to the function of the building itself, even while some of those features, such as mosaic depictions of the zodiac circumscribing the Sol Invictus, seem jarringly out of place in this context.⁵⁵ Such incongruities notwithstanding, there is general agreement that the programmatic intent was a narrative foregrounding and symbolic intensification of the meaning of the Torah shrine and ark, since it is the physical presence of the text itself that imbues the synagogue with its monumental significance. Thus, the effect is derived from the hierarchical relationship that links physical space, symbol, and text to create a figurative passage. As I shall discuss below, this program foreshadows the propylaic intent found in the illuminated canon tables tradition.

The particular adaptation of the Greco-Roman aedicular motif found in the post-Second Temple synagogue emerges sporadically over the next centuries at various points of Judeo-Christian interculturation.⁵⁶ Perhaps the clearest iteration is when the motif made the leap onto the page as [End Page 417] the frontispiece to the famous Ashburnham Pentateuch. As scholars have long-observed, the mise-en-page draws from the architectonic and ornamental program of the Torah shrines of post-70 c.e. synagogal iconography, including, notably, the Dura-Europos Torah ark.⁵⁷ The similarity led Gutmann to suggest an almost literal parallel, namely that the Ashburnham illuminator “had the unveiling ceremony of the sacred scriptures in mind.”⁵⁸ It does seem clear that the monumental design is specifically linked to the act of accessing a holy text, such that the architectonic elements (including, importantly, the curtain as depicted in the Ashburnham Pentateuch) serve as a symbolic frame for the ceremonial act, limning a frontier that separates the profane from the sacred. While the degree to which the illustrations of Ashburnham reflect a Jewish influence or provenance is perhaps more open to doubt than previously thought, the similarity of the opening folio with synagogal antecedents remains unperturbed.⁵⁹ The fact that, as Dorothy Verkek has recently argued persuasively, the Ashburnham Pentateuch was of Italian provenance, probably from Rome, and drew upon Christian and not Jewish antecedents adds further to its significance, because it means we have a specific example of the aedicular motif associated with the Torah niche being refashioned to serve the needs of a Christian codex for “a clerical audience.”⁶⁰

Thus, some forms of post-70 c.e. Jewish synagogal decoration, such as that preserved at Dura-Europos, provide a stylistic prototype that shares not just the design features, but the symbolic functionality of the canon tables. Given the extent to which early Christian art drew upon existing Judaic models, the possibility that there is a link between the symbolic intent expressed in Dura-Europos, recreated later in the Ashburnham Pentateuch miniature, and that which informed the design of the canon tables is not inconsiderable.⁶¹ Given the striking parallels between the Ashburnham [End Page 418] Pentateuch miniature and those of several of the early canon tables, which I shall consider below, such as the exordial placement, the use of columns to enclose text, the depiction of curtains, and so forth, it is possible that they sprang from the same tradition and with much the same intent as that of the decoration of synagogal Torah aediculae.⁶²

THE GOSPELS AS ARCHITECTURAL TROPE

We are still left, however, with the problem that the mechanism connecting an extant visual trope, wherever it sprang from, with the actual business of laying down the design for the tables once they had been worked out remains obscure. Helpfully, a remarkable text from Eusebius himself offers an insight into how the propylaic function of the table design may have emerged. The text is his Panegyric on the Cathedral at Tyre, which was delivered at the consecration of the newly rebuilt church, the original structure having been destroyed as a result of the Diocletian Edict of 303.⁶³ This offered one of the first opportunities to celebrate a new form of Christian expression made possible in the climate after the Edict of Milan in 313.⁶⁴ Since there was no existing template for this kind of Christian consecration panegyric, Eusebius drew upon various classical rhetorical models to compose his remarks, which offered his audience not only a dazzling display of biblical erudition to describe the achievement represented by the reconstruction, but also a remarkable ekphrasis of the physical structure of the church itself.⁶⁵ Christine Smith has noted that Eusebius’s Panegyric employs the rhetorical tactic of the laus urbis, or praise of the city, as [End Page 419] well as the periegesis, associated with classical writers like Pausanius or Strabo, in which the meaning, beauty, or aesthetic value of physical space is described and captured through a kind of textual circumambulation.⁶⁶ And Jeremy Schott has more recently argued for reading Eusebius’s oration as a form of “spatial practice,” in order to “elucidate the complex relationship between texts and architectural spaces.” For Schott, Eusebius uses the physical structure of the church as a way “to consider the nexus of text and event, of the Biblical past and the Christian present.”⁶⁷ And in the ekphrastic tradition, Eusebius turns the physical towards the metaphorical and metaphysical in order to imbue both the physical space of the church, as well as its symbolic value as a triumph of the Christian faith with a meaning that lays stress on the spiritual and the soteriological. The periegetic description thus mirrors the journey of the faithful, the material substance of the church being transformed into the spiritual substance of faith, salvation, and the truth of the Word.

The way Eusebius achieves this is by describing the physical structure of the church, laid out on a horizontal plane, as also representing a vertical ascendance towards salvation. This starts from the very threshold of the church itself, where those who see it from the outside “might be attracted and be induced to enter by [its] very sight.”⁶⁸ Moving across the horizontal strata of the church is thus a complex passage, one that mirrors the process of “dissolving the flesh from corruption to incorruption.”⁶⁹ It entails crossing a ritualized, codified hierarchy that rises from entry through ablution, catechumen, and baptism, to the Word itself:

Building therefore in righteousness, he divided the whole people according to their strength. With some he fortified only the outer enclosure, walling it up with unfeigned faith; such were the great mass of the people who were incapable of bearing a greater structure. Others he permitted to enter the building, commanding them to stand at the door and act as guides for those [End Page 420] who should come in; these may be not unfitly compared to the vestibules of the temple. Others he supported by the first pillars which are placed without about the quadrangular hall, initiating them into the first elements of the letter of the four Gospels. Still others he joined together about the basilica on both sides; these are the catechumens who are still advancing and progressing, and are not far separated from the inmost view of divine things granted to the faithful. Taking from among these the pure souls that have been cleansed like gold by divine washing, he then supports them by pillars, much better than those without, made from the inner and mystic teachings of the Scripture, and illumines them by windows.⁷⁰

The architectonic features of the church thus serve as a series of thresholds that characterize the journey from pagan (i.e. “those outside”) to Christian, through all the liminal steps toward salvation.⁷¹ And the culmination of this soteriological journey is the “support” of those who are baptized by pillars which are constituted, in Eusebius’s formulation, by the Gospels themselves.

Thus, in the Panegyric, Eusebius offers a specifically Christian ekphrasis of the church, in which the physical layout is mapped onto a soteriological path, such that the physical elements of the structure are hierarchized to serve metaphorical ends that entail “progressing” (προκοπῇ) toward salvation (baptism) and ultimately the “illumination” of the true faith.⁷² This means that not all pillars of the church are created equally. As we pass across the space of the building, the physical elements—literally the pillars that hold up the structure—are necessarily constituted by ever more “uplifted” components requisite to that journey, including the “first element” of the gospel, the catechumen, and finally the “mystic teachings” of the Holy Writ.⁷³ The Eusebian ekphrasis of the Cathedral at Tyre furnishes [End Page 421] a compelling instance of the way in which, at roughly the same time the canon tables were being formulated, the Gospels were understood as constituting in metaphorical terms the innermost structural support of the church. This image implicates the kind of “spatial practice” that links together the text with the church and suggests that the Gospel-metaphor of the Eusebian Panegyric as the inner pillars of the church could connect to the archetype model for illuminating the Eusebian apparatus.

THE EARLIEST CANON TABLE ILLUMINATIONS

I suggest that the propylaic function of the Eusebian canon tables emerges from the nexus of the three developments I have traced above. First, the derivation of a powerful bibliographical tool that changed how the Gospels could be read as a single, interconnected text helps situate the meaning of the apparatus to the text itself for early Christian readers, explaining both why it appears as a frontispiece and why it demanded some kind of monumental frame. Second, the liturgical, symbolic, and/or narrative use in synagogal settings of an aedicular form to enshrine the Torah offers a clear antecedent for how such forms could serve a semiotic function to delineate and solemnize the space of sacred texts within a larger architectural setting. Third, an increasingly elaborate understanding of how the physical church could act as a metaphor for a Christian soteriology reinforced the role of the Gospel itself, considered both as textual object and metaphysical witness, as the liturgical focal point within a hierarchical architectural plane. Given this, I wish to consider how we might approach the propylaic meaning of the canon tables in the earliest surviving Gospel texts (i.e., pre-seventh century) from the Syriac, Armenian, and Greek traditions, such as the Diyarbakir, Rabbula, Garima, Etschmiadzin, and Beneventanus Gospels.

Over time, this initial propylaic function undoubtedly expanded. By the early eighth century, for instance, it was possible for the Armenian Bishop Step‘anos Siwnec‘i (d. 735) to envision the tables as an encompassing and multi-faceted symbol of the mysteries of Christianity itself.⁷⁴ It is tempting to think that his text, only partially preserved, was expanding upon an earlier hermeneutics of the tables, given how it builds on the core idea [End Page 422] that the tables represent a progression through the mysteries of the faith, from the witness of the Evangelists through to the principle of salvation and rebirth. Such a tradition would align with a reading of the canon tables that links their propylaic function for the Gospel text with the mystery of the divine Logos, as I shall explore below.

By drawing attention to specific features of a Christian space, the architectonic design of the canon tables reflects the development of a distinctly Christian architecture, since the diffusion of the tables coincides with the rise of an ambitious building program after the Peace of Constantine in 324.⁷⁵ As Eusebius’s panegyric makes clear, this also entailed an evolving reading of the meaning of space within the church. From being a place for informal (and often clandestine) gathering, the physical building assumed from the fourth century a codified significance that reflected the hierarchy of the church itself, reinforcing in spatial terms access to the rituals of worship and highlighting the importance of the liturgy. Just as the canon tables offered an innovative and new way to access Scripture, so too did an evolving architecture and spatial layout of the physical buildings in which those texts were read.⁷⁶ This shift from repurposing existing structures for congregation to purpose-built structures that embodied the sanctification of worship that they housed meant that the holiness of the word was extended to the holiness of the site. Text and space fuse together to recall and reveal the divine Logos through Christ’s incarnation and resurrection.

Laudatory works like Paul the Silentiary’s ekphrasis for Justinian’s rebuilding of the Hagia Sophia or the Sogitha for the Cathedral at Edessa reflect this evolution by embracing the specifically architectonic within the mystery of the divine:⁷⁷

[The temple’s] lofty dome!Behold, it resembles the highest heaven!And like a helmet the upper [part] is firmlyplaced on its lower [part].The splendor of its broad arches!They portray the four ends of the earth!⁷⁸ [End Page 423]

Hymns like the Sogitha codified the link between the physical architecture of the church and its spiritual symbolism. It is reasonable to think that the elaboration of a visual protocol for the canon tables during this period would share this interest and that the design was intended to make specific allusion to a Christian space characteristic to those communities from within which they were produced. The architectural motif of the early tables thus serves to link the word (logos) of the gospel with the divine Logos animated by the ritual and liturgical practices within the physical space of the church.

One of the most intriguing characteristics of the earliest tables is that, while they share a common design syntax, they demonstrate quite radically different iterations of that syntax.⁷⁹ Nordenfalk elaborated a primary distinction in variants of illumination of the canon tables based on the formal disposition of the superposed arcading and suspected that the canon tables of the tenth-century Etchmiadzin (Ejmiacin) Gospel came the closest to a Eusebian archetype as a faithful copy of a much earlier exemplar that closely followed a Caesarean urtext.⁸⁰ But quite aside from what the original design may have looked like, of greater significance is why, from the earliest period of their transmission, they came to take on such radically different styles, even while maintaining a shared architectural form. If the London, Garima, and Rabbula (and possibly the proto-Etchmiadzin) tables are even roughly coeval, it is clear that the differences between them were not the result of some evolutionary process in which an archetypal design was gradually transformed through scribal copying and recopying.⁸¹ Rather, their different styles reflect radically different ideas of what the monumental decorative frame should look like, even if the overall architectonic form was broadly shared.

The simplest explanation is that they refer in some way to different styles employed in early Christian churches, which had a diverse architecture and [End Page 424] ornamentation; that diversity then shows up in the significantly different variants within the broad architectonic syntax of the tables. If they represent, even approximately, the internal aspects (including ornamentation) of the Christian churches found in the communities from which these texts sprang, then it is clear that these varied widely depending on where you were. This is consistent with archaeological evidence that shows considerable diversity of layout and design in the building program of churches in late antiquity. Thus, when we look into the pages of the Rabbula, Garima, London, or Etschmiadzin (or any other later Gospels that were faithful copies of much earlier exemplars), what we might be seeing are the different architectonic and ornamental forms of Latin, Syriac, Byzantine, or Armenian churches. The representations may have been stylized, to be certain, when miniaturized on the page. But stylization alone hardly seems sufficient to account for the extensive variation found in the canon table designs from texts that were produced during the same period.

I think we can go further than this, however. Of the illuminations found in the early Gospels noted above, perhaps the most striking are those found in the Rabbula Gospel. Much scholarly attention has been paid to the illuminations in this Gospel, largely because of the highly unusual depictions of biblical scenes after the end of the tables (fols. 13a–14v).⁸² Their most salient feature has largely gone unremarked beyond merely a descriptive acknowledgment, namely the pronounced horseshoe-shape arches that comprise the frame. This architectural feature of the canon frames is found in a number of early Gospels both from within the Syriac tradition, such as Armenian and Cilician examples, as well as outside of it, such as the late sixth-century Harley Gospel.⁸³ Nordenfalk concluded that the horseshoe form generally reflected a Eusebian archetype and suggested therefore that the Rabbula tables were following this original.⁸⁴ But beyond a vague acknowledgement of the horseshoe arch as a feature of “Syro-Palestinian art,” Nordenfalk offered no further explanation of why this particular form should have been employed for the canon tables, nor why the form should have been substituted with the more standard [End Page 425] Roman (i.e. semicircular) arch shape in other traditions.⁸⁵ Since then, no one to my knowledge has addressed further the potential significance of this prominent, if curious feature found in many of the earliest extant examples of the table design, albeit in significantly different ways. Yet it is noteworthy because this particular form may tell us something about their representational intent.

The first point suggested by the use of the horseshoe arch as a prominent feature in the framing of the canon tables is that it was not intended as a general architectural representation of the church. This is because churches in Syria and Palestine did not incorporate the horseshoe arch in their exterior architecture.⁸⁶ Thus, the arches of the tables do not recall a narthex or other type of entrance into the church, nor are they likely then to be metonymic for the building as a whole. However, if the horseshoe shape was not associated with a general construction program in Syro- Palestinian churches, they do nonetheless appear. In horizontal form, of course, the horseshoe shape was commonplace in the layout of both apses and bemas in churches across the region.⁸⁷ That this sympathy of design between the bema and the apse was linked is suggested by the fact that in the slightly later rupestrian churches of Cappadocia, the apse, almost always horseshoe shaped, was raised and served as the bema.⁸⁸ Thus, the horseshoe shape in horizontal form was an important part of the interior layout of churches across the region, signaling the innermost sanctuary where the liturgy was read and the Eucharistic act performed.

Now, surviving examples of the horseshoe arch in Syrian churches are exceedingly rare: Butler only identified two extant occurrences in the course of his surveys.⁸⁹ But he did find evidence that the horseshoe arch had been much more widely employed in Syrian church interiors “usually appear[ing] in relieving arches over doorways, . . . [and] occasionally employed in the arcades of naves.” While what evidence remains for this feature in Syrian [End Page 426] ecclesiastical design has now likely been further degraded, this element survives elsewhere in Armenian, Iberian, Cilician, and Cappadocian examples. The sixth-century Kızıl Church in Anatolia, for instance, employs both arch types—horseshoe and semi-circular—consistent with the Syrian scheme of using the two forms to demarcate external and internal space.⁹⁰ The horseshoe arch thus appears not to have been a form that was associated with entering (or exiting) the church, but rather with passages (e.g. naves) that were traversed once one was already inside the church, with the focal point being the bema. Thus, the horseshoe arches of the Rabbula illuminations seem to be drawing attention to a specifically internal space, one that, as we have seen, was engaged by the Eusebian idea of progression (προκόπτω) found in his Panegyric of Tyre. As such, one way of understanding why the framing of the tables uses this unusual and specific arch form is that it is making a spatial reference. It captures the idea of a hierarchical progression across a horizontal space, where the apex is an identification between the entirety of witness of the Gospels with the divine Logos, that culmination where, as Eusebius puts it, “the pure souls that have been cleansed like gold by divine washing” are supported “by pillars, much better than those without, made from the inner and mystic teachings of the Scripture.”⁹¹

The idea that the architectonic motif of the canon tables, at least in these early iterations, denoted an internal church space that fused the logos as both word and spirit is further reinforced by two features of the early table illuminations.⁹² The first is the prominent placement of curtains in both the Garima and Etschmiadzin Gospels, a very unusual feature of early canon table iconography (although featured prominently in the Ashburnham frontispiece), but found in early Georgian, as well as many later Armenian and Ethiopic examples.⁹³ Curtains were, of course, a feature of some early Syro-Palestinian churches.⁹⁴ But the placement of a curtain here almost [End Page 427] certainly recalls the traditional veil in front of the Torah niche; the drawn curtain is likely a reference to the metaphor of the torn (i.e. parted) curtain (Matt 27.51) signifying the ability of the faithful to behold the inner mysteries of the holy. The curtain may also link to the idea of an inner veil as allegorized by Origen and developed in later Christian thought.⁹⁵ Either way, the motif references the christological fulfillment of the law (Matt 5.17) to which the Gospels give witness and to which similar thematic reference can be found in later illuminated Gospels.⁹⁶ The correlation between the motif in the Garima and the Etschmiadzin Gospels with that of the Ashburnham Pentateuch suggests a common symbolic elaboration of the function of the text, not just as the revelation of truth though the word, but further with reference to the specific spatial context of the inner sanctuary of the church itself.

The idea that the framing of the tables is intended to signify a passage, not into, but through the church towards its inner sanctuary also elucidates the meaning of the tholoi or tempietti that appear as part of the canon tables (usually in terminal position), starting with several early Gospels from across different traditions, including the Rabbula and possibly the proto-Etschmiadzin.⁹⁷ There is disagreement as to the symbolic significance of this architectonic structure, but conjecture that it had specifically baptistery connotations (Underwood) or was an aedicular form specific to the Gospels themselves (Klauser) seems more compelling than their serving simply as a generic representation of ecclesia (Bandmann).⁹⁸ Such readings of the tholoi are made more compelling if the representational function of the tables’ architectonic design was figuratively to connect the text of the Gospel with the internal sanctuary of the physical church. Since the lectionary point (whether bema or ambon) was often the spatial nexus within the church that linked to both Gospel shrine and baptismal space, either Underwood’s or Klauser’s reading is plausible.⁹⁹ What does seem [End Page 428] clear, as Underwood notes, is that the temple motif links to the idea of “the concordance of the Gospels in revealing the ‘good’ news of the plan of salvation.”¹⁰⁰ The placement of tholoi at the conclusion of the larger iconographic program of the canon tables thus reinforces the idea that the visual design was intended to serve as more than a broad representation of the church itself, to instead capture a progression through a spatial hierarchy—as elaborated by Eusebius’s remarks in his panegyric—that culminates with salvation.¹⁰¹

To summarize, efforts to trace some kind of an Urbild to explain the canon table iconography are difficult owing to the attenuation of possible archetypes, whether from a broader visual tradition of antiquity that features architectonic miniature, such as sarcophagi, or even from those rare instances where similar tabular content does appear in broadly sympathetic visual frames. In this regard, what helps move the argument forward is to note, first, that the function of the tables should not be divorced from the form, which observation leads to a potential link between synagogal iconography echoed elsewhere in contemporary Judaic and later Christian iconography, with the canon tables’ visual syntax. The aedicular motif in synagogues established a specific precedent for the framing of sacred texts (the Torah scrolls), whether figuratively, such as in mosaic design, or literally, such as the niche frescoes at Dura-Europos, thus creating a liturgical focal point that emphasized the hierarchy of the physical space of the temple around the text itself. And unlike the other principal signifier of this function, the raised bema, this motif was adaptable and reproducible in a variety of other media, including on the page as illumination.

In looking for a suitable design for framing the tables in the first Gospels to feature them, the crafters may thus have drawn upon what linguists call a “symbolic competence,” the ability to derive meaning abstracted from an existing set of indexical or iconic references.¹⁰² Within the physical setting of the synagogue, the aedicular motif had both iconic and indexical significance, signifying the physical location of the Torah as well as the role of the ark or the Temple in Jewish scriptural narrative. Outside of this setting, the aedicular or similar architectural motif needed to draw upon a symbolic competence that could allow it to be translated from an iconic [End Page 429] anchor within a defined space to a symbolic referent that retained its meaning as a signifier of both the text and its function of spatial demarcation.

The diverse palette of designs found in the earliest Gospels with illuminated tables thus points to the different ways in which the underlying iconicity and/or indexicality is being referenced. Details like horseshoe arches, or curtains, or tempietti (among others) are thus not simply reflections of local idiosyncrasy or available “off-the-shelf” designs, but instead are vital to the meaning of the framing of the tables. This suggests that the design of the tables was making a specific allusion to the progression through the church that comes about as a function of the concord of witness from the Gospels themselves. The frame thereby conveys a referential meaning drawn from the very textual cross-references that are being framed. This explains the inclusion of the horseshoe arch in Gospels like Rabbula as not simply a novelty of design drawn from some vague Syro- Palestinian affinity for the form, but instead as a specific allusion of the meaning of the Gospel text to the hierarchy of a physical interior space. Similarly, this explains the use of drawn curtains and the terminal placement of tempietti, as important markers of that progression. More broadly, it offers an explanation for the dramatic differences in even the earliest examples of the tables, since, depending on where they were being produced, the design syntax had to be adapted to the symbolic competence of specific communities in order for this meaning to resonate (or possibly be rejected altogether).¹⁰³

The ekphrastic moment found in the Eusebian panegyric at the very beginning of post-Constantinian Christian expression, and broadly coincident with the elaboration of the apparatus, draws an explicitly soteriological metaphor from the architecture of the Christian church, and links specifically the innermost pillars that support the physical structure, with Scripture itself. Thus, framing the canon tables within an elaborate [End Page 430] propylaic design that designates passage through the interior space of the church seems consistent with the need both to honor the significance of the achievement of the apparatus for the text, as well as providing a visually intelligible way of capturing their meaning within or as part of the text itself. Given that the elaboration of the canon table archetype coincides with the ambitious building regime that followed the Constantinian peace, this interest in trying to devise a formal mechanism to link sacred text with sacred space seems consonant with the larger iconographical, intellectual, and ecclesiological thrusts of Christianity after the third century. It created a unity of purpose and function between the physical space of worship and sacra scriptura. It is true that this early significance evolved as the tradition of scribal transmission led over the subsequent centuries to increasingly divergent iterations of the original syntax of the earliest archetypes, and the design convention of the tables gave space for different possible iconographic functions. But in their early versions, it seems clear that the syntactical design and visual elaboration of the canon tables tells us something about the way Christians were working out how to derive a broader and emboldened ecclesiology that linked text, space, and practice in service to their God. [End Page 431]