The Date, Dedication, and Design of Sts. Sergius and Bacchus in Constantinople
The date at which work began on the construction of Justinian and Theodora's church of Sts. Sergius and Bacchus is disputed. It is argued here, on the basis of historical sources, that work probably started after 532, and that the evidence of the dedicatory inscription and the monograms on the column capitals is not inconsistent with this conclusion. The construction of the church alongside the basilica of Sts. Peter and Paul may be explained as an act symbolic of Justinian's hopes to reconcile the non-Chalcedonians with the recently reunified Roman and Eastern Churches. An analysis of the language of the dedicatory inscription suggests that the poem was largely inspired by traditional kingship ideology and the biblical book of Wisdom, and need not necessarily be read as a declaration of rivalry with contemporary members of the aristocracy. It has always been taken for granted that the church's gored dome belongs to the original building. Here, it is argued that the dome is a later repair and that the original dome was seated upon an octagonal fenestrated drum, similar to that of San Vitale in Ravenna. Consequently, Sts. Sergius and Bacchus should not be seen as a prototype for Hagia Sophia.
The church of Sts. Sergius and Bacchus in Constantinople was constructed in the sixth century by the emperor Justinian and his wife, Theodora, and was located in the eastern part of the city, close to the shores of the Sea of Marmara, within a residence called the "Palace of Hormisdas." The building [End Page 62] survives today as a consequence of its conversion into a mosque at the beginning of the sixteenth century.1
The main body of the building (figs. 1–2) measures less than 30 m square, with a projecting three-sided apse to the east and a narthex to the west.2 In the nave stand eight piers, which are located at the corners of a slightly irregular octagon. Each pier is wedge-shaped in plan and preserves the angle of the octagon on the side facing the nave. The piers rise to a height of about 10 m above the nave floor, at which point they are connected by eight arches, which serve as the seating for a dome. The dome's sixteen sides alternate between flat and concave. The concave sides rise above the main piers, and the flat sides, each pierced by a window, rise above the arches. The space surrounding the octagonal core is divided by a floor at a height of about 5 m, so creating two ambulatories. On each side of the octagonal core, except towards the sanctuary in the east, there are, at both ground-floor and gallery level, a pair of columns standing between adjacent piers. The paired columns to the north, south, and west are aligned with the flanking piers, whereas those on the alternate sides of the octagonal core are set back from the nave to create semi-circular recesses (exedras). The ground-floor columns are crowned by so-called melon or fold capitals, which in turn support an inscribed horizontal entablature of 29 marble blocks. The entablature coincides with the level of the gallery floor. The gallery columns are crowned by Ionic impost capitals, which support triple arches. The space between the triple arches and the dome arches is filled, on the straight sides of the octagon, by a lunette, and, on the recessed sides, by a semidome.
The building is known in Turkish as "Küçük Ayasofya Camii," the "Little Hagia Sophia Mosque," a name that compares it with Justinian's much larger church of Hagia Sophia (Ayasofya), the pinnacle of Byzantine architectural achievement, which also survives, having been converted into a mosque in 1453 and into a museum in 1935.3 The fact that both churches are domed and were built by the same emperor has led to conjecture that Sts. Sergius and Bacchus was an architectural experiment prior to building the daring dome of Hagia Sophia with its impressive diameter of 31 m. The suggestion presupposes two things: first, that the smaller church was built before the larger; second, that the nature of the two domes is somehow comparable. Here I argue, [End Page 63]
[End Page 64]
amongst other things, that Sts. Sergius and Bacchus was built not earlier than, but at the same time as Hagia Sophia, and that its gored dome is not—as has always been taken for granted—the original Justinianic roof.
Since 1913, when Jean Ebersolt and Adolphe Thiers published Les églises de Constantinople, it has generally been accepted that Sts. Sergius and Bacchus was built between 527 and 536.4 However, the precise period of construction continues to be disputed. In articles published in 1972 and 1975, Cyril Mango argued that the church had been built between 531 and 536 in connection with the arrival in the Palace of Hormisdas of non-Chalcedonian dignitaries [End Page 65] and refugees.5 In 2000, I claimed that I had found further evidence in John of Ephesus that supported Mango's argument.6 However, in an article published in 2006, Brian Croke rejected the case for the church having been built under those circumstances. Following a suggestion made, but not adequately argued, by Richard Krautheimer, Croke assembled a strong case that the refugees would not have arrived in Constantinople until 536, by which time the church must already have existed.7 Croke therefore set out to re-examine the date and circumstances of the church's construction. Following further in the footsteps of Krautheimer, who had suggested that work on Sts. Sergius and Bacchus might in fact have begun before 527,8 Croke fully developed the case for pushing back the date at which construction started.9
Croke presents the arguments both in favor of his own position and against those of earlier scholars in some detail. For this reason, and because his redating has important implications for both the historical context of the church's construction and our understanding of the monument's significance, his case deserves careful consideration. Here, I intend to argue against Croke's early date, and to show that our most reliable evidence suggests that the terminus post quem should be revised to 532. It is therefore appropriate to address Croke's arguments primarily, while referring to the earlier scholarly contributions to this debate—by Krautheimer, Thomas F. Mathews, Mango, and me—at appropriate points in the discussion.
Croke claims that construction of the church of Sts. Sergius and Bacchus was begun before the death of Justin I on 1 August 527, whilst its patrons, Justinian and Theodora, were resident in the Palace of Hormisdas. His argument relies on his interpretation of the content of the short Greek inscription carved on the ground-floor entablature surrounding the nave of the church, and on the monograms carved on the capitals at both the ground-floor and gallery level of the church.10 I shall postpone consideration of those matters for the moment, and examine first other neglected evidence that is, I believe, of greater significance in determining the church's date.
The first piece of evidence bearing on the date of construction comes from Procopius's Buildings, a work written between 550 and 554 by a knowledgeable [End Page 66] contemporary of Justinian.11 The Buildings has been transmitted in two families of manuscripts, one preserving a long version of the text, the other a short.12 The short recension represents either, in the opinion of Bernard Flusin, an abridgement of the long, with some traces of rewriting to compensate for the effects of deletion,13 or, in the estimation of Federico Montinaro, Procopius's original text, as published in 550/1, which was subsequently expanded into the long text and republished around 554.14 Surprisingly, although Procopius's account is our main source concerning the building of Sts. Sergius and Bacchus, it has not received the careful attention it deserves. I quote first the relevant passage in the long recension:
(1) Ἐς δὲ τοῦ Χριστοῦ ἀποστόλους τὸ πιστὸν ἐπιδέδεικται τρόπῳ τοιῷδε. πρῶτα μὲν Πέτρῳ καὶ Παύλῳ νεὼν οὐ πρότερον ὄντα ἐν Βυζαντίῳ ἐδείματο παρὰ τὴν βασιλέως αὐλήν, ἣ Ὁρμίσδου τὸ παλαιὸν ἐπώνυμος ἦν· (2) ταύτην γὰρ οἰκίαν αὑτοῦ ἰδίαν Παλάτιον εἶναι δοκεῖν τε καὶ πρέπειν τῷ μεγαλοπρεπεῖ τῆς οἰκοδομίας διαπραξάμενος, ἐπειδὴ αὐτοκράτωρ κατέστη Ῥωμαίοις, τοῖς ἄλλοις βασιλείοις ἐνῆψεν. (3) οὗ δὴ καὶ τέμενος ἄλλο ἁγίοις ἐπιφανέσι Σεργίῳ τε καὶ Βάκχῳ ἐδείματο, καὶ ἔπειτα καὶ τέμενος ἄλλο ἐκ πλαγίου τούτῳ παρακείμενον. (4) ἄμφω δὲ τούτω τὼ νεὼ οὐκ ἀντιπροσώπω, ἀλλ' ἐκ πλαγίας ἀλλήλοιν ἑστᾶσι, συνημμένοι τε καὶ ἀλλήλοις ἐνάμιλλοι ὄντες. … (7) … ἐπὶ κοινῆς … τὸ τροσήκειν τοῖς βασιλείοις. (8) οὕτω δὲ ἄμφω ἀγαστὰ τὰ ἱερὰ τάδε ξυμβαίνει εἶναι ὥστε διαφανῶς τῆς τε πόλεως ὅλης καὶ οὐχ ἥκιστα τῶν βασιλείων ἐγκαλλώπισμα τυγχάνει ὄντα.15
(1) His [i.e., Justinian's] faith in the Apostles of Christ is demonstrated in the following manner. First he built a Church of Peter and Paul, which did not previously exist in Byzantium, alongside the imperial residence, which in former times was named after Hormisdas. (2) For he had contrived that this building, which was his private residence, should both seem to be a palace, and by the magnificence of its construction be fitting for one; and when he became Emperor of the Romans he joined it to the rest of the palace. (3) There he built another shrine too to the famous Saints Sergius and Bacchus, [End Page 67] and afterwards another shrine too, which lay alongside this one on its flank.(4) These two churches do not stand face to face, but flanking one another, at once joined and rivaling one another. … (7) … They have in common… that they belong to the palace. (8) Both shrines are so admirable that they manifestly form an adornment of the whole city, and especially of the palace.
As for the short recension, I quote it only so far as is necessary for the purposes of my argument:
(1) Ἐς δὲ τοῦ Χριστοῦ ἀποστόλους τὸ πιστὸν ἐπιδέδεικται τρόπῳ τοιῷδε. πρῶτα μὲν Πέτρῳ καὶ Παύλῳ νεὼν ἐδείματο παρὰ τὴν βασιλέως αὐλήν, ἣ Ὁρμίσδου τὸ παλαιὸν ἐπώνυμος ἦν· (2) ἔνθα τὴν οἰκίαν εἶχεν αὐτὸς πρὸ τοῦ βασιλεῦσαι.(3) ἔπειτα καὶ τέμενος ἄλλο ἐκ πλαγίου τούτῳ παρακείμενον, ἁγίοις ἐπιφανέσι Σεργίῳ τε καὶ Βάκχῳ ἐδείματο. (4) τούτων δὲ τῶν δύο ναῶν ἅτερος. …16
(1) His [i.e., Justinian's] faith in the Apostles of Christ is demonstrated in the following manner. First he built a Church of Peter and Paul alongside the imperial residence, which in former times was named after Hormisdas. (2) Here he had his house before becoming emperor. (3) Afterwards he built another shrine too, which lay alongside this one on its flank, to the famous Saints Sergius and Bacchus. (4) Each of these two churches …
Being more informative than the short, the long recension deserves our particular attention; but before considering its implications, we must address an obvious difficulty. As it stands, the Greek of the long recension gives the impression that, after the construction of Sts. Peter and Paul and Sts. Sergius and Bacchus, Justinian also built a third church, positioning it alongside Sts. Sergius and Bacchus. Yet the fact that the text continues with references to, "These two churches" (ἄμφω δὲ τούτω τὼ νεὼ) and, "Both shrines" (ἄμφω … τὰ ἱερὰ) implies that there were in reality only two churches, and that it was Sts. Sergius and Bacchus that was built alongside the primary church of Sts. Peter and Paul. This inference is supported by the fact that the short recension, the text of which makes perfect sense, mentions only two churches, Sts. Peter and Paul and Sts. Sergius and Bacchus. That the text of the long recension has somehow become corrupted is intimated by the careless repetition of the words καὶ τέμενος ἄλλο ("another shrine too") and by the use of the similar phrases ἐκ πλαγίου ("on its flank") and ἐκ πλαγίας ("flanking"). Therefore, it seems highly likely that Glanville Downey was correct in suggesting that we should delete the words καὶ ἔπειτα καὶ τέμενος ἄλλο ἐκ πλαγίου τούτῳ [End Page 68] παρακείμενον ("and afterwards another shrine too, which lay alongside this one on its flank"), since this would remove both the apparent reference to the third church and the repetitious wording.17
Downey proposed that the words to be excised were those of a copyist, who, distracted by the description of the Palace of Hormisdas (which is more extensive in the long recension than in the short), forgot about the church of Sts. Peter and Paul that had been mentioned previously, and so became confused when he read the words ἄμφω δὲ τούτω τὼ νεὼ ("These two churches"). He therefore tried to make sense of the passage by composing the unnecessary addition.18 Although it is perhaps possible that the copyist became confused as Downey proposed, the inserted words were certainly not the scribe's own, since precisely the same words occur in the short recension, where they relate to the construction of Sts. Sergius and Bacchus itself. Thus, it would appear that the words ἔπειτα καὶ τέμενος ἄλλο ἐκ πλαγίου τούτῳ παρακείμενον from the short recension somehow became incorporated into the long (with the addition of a connective καὶ). This effectively resulted in reference being made twice to the construction of Sts. Sergius and Bacchus, the duplication being reflected in the repetitious wording. However, one can only speculate as to when and how this corruption occurred.19
Therefore, the cause of this textual disturbance can be explained, at least in general terms, and the correct text of the long recension can be suggested with a good degree of certainty. It is also clear that none of this has implications for the text's reliability regarding the events leading up to the construction of Sts. Sergius and Bacchus, to which we may now turn.
From Procopius's account, it is evident that, before he ascended the throne, Justinian built a church of Sts. Peter and Paul in the Palace of Hormisdas, the residence he occupied at the time. This is corroborated by two letters to Pope [End Page 69] Hormisdas, both dated 29 June 519: one from Justinian himself mentioning "the church built in our residence" (basilicam … in domo nostra … constructam), and requesting unspecified relics to adorn it; the other from the papal legate at the court in Constantinople, stating that the basilica of the Apostles has been built, and explaining that Justinian seeks to obtain the chains that bound the Apostles and the gridiron on which St. Laurence was burned as relics for the edifice.20
Procopius next tells us that when Justinian became emperor he connected the Palace of Hormisdas to the neighboring Great Palace, the main residence of the Byzantine emperors located on the east flank of the hippodrome. Justinian was elevated to co-Augustus on 1 or 4 April 527, as the health of Justin I deteriorated, and became sole emperor on 1 August 527, upon his uncle's death.21 The implication of Procopius's statement seems to be that, after the beginning of April 527, and probably after 1 August 527, Justinian moved (as we would have expected) from the Palace of Hormisdas into the Great Palace, and decided to connect his new residence with his former one. At this point, Procopius mentions the construction of a second church in the Palace of Hormisdas, that of Sts. Sergius and Bacchus. Therefore, regarding the events leading up to the construction of Sts. Sergius and Bacchus, Procopius is quite clear.
Procopius presents the construction of Sts. Peter and Paul as an illustration of how Justinian strove to turn his private residence into a palace. If, as Croke claims, Sts. Sergius and Bacchus had also been built while Justinian still occupied the Palace of Hormisdas, it would have been natural for Procopius to adduce this church too as evidence for the adornment of his residence to increase its grandeur. On the contrary, however, Procopius suggests that the construction of Sts. Peter and Paul alone was sufficient to make the Palace of Hormisdas worthy of being joined to the rest of the palace, a process Justinian undertook after he became emperor. Procopius mentions the building of Sts. Sergius and Bacchus as a separate and later act. The natural inference, it would seem to me, is that Sts. Sergius and Bacchus was built—contrary to Croke's claim—after Justinian became emperor in 527 and after the Palace of Hormisdas had been joined to the Great Palace. This text provides the clearest indication we have regarding the date of the construction of Sts. Sergius and Bacchus, and therefore deserves to be taken seriously until better evidence becomes available. What is not clear from this passage, however, is the exact length of time that passed between Justinian's elevation to the rank of Augustus and the construction of Sts. Sergius and Bacchus. [End Page 70]
Those scholars who hold—as I believe Procopius implies—that Sts. Sergius and Bacchus was begun sometime after Justin's death in 527 are faced with the problem of explaining why Justinian and Theodora, having vacated the Palace of Hormisdas, chose to build a new church in that palace. Adopting 527 as the terminus post quem for the construction, Mango held that the explanation lay in the church having been built for the refugee community lodged in the halls of the Palace of Hormisdas by the empress Theodora, and I agreed.22 Rejecting this as a chronological impossibility, Croke attempts to explain the location of the church by arguing that work on the construction of Sts. Sergius and Bacchus must have been started whilst the couple were still residing in the Palace of Hormisdas. "It has," Croke explains, "always been considered highly unusual that Justinian would have begun to build a church such as SS. Sergius and Bacchus within a mansion he no longer occupied, that is, on the assumption that it was more or less entirely built after 527. This problem evaporates if the church was planned and at least commenced, if not largely completed, before the death of the emperor Justin in August 527."23 Having proposed this solution, Croke proceeds, as we shall see, to predicate his interpretation of the inscription and the monograms in the church upon it.
It seems to me, however, that Procopius gives an entirely satisfactory explanation for the location of the church. If we accept the natural interpretation of his account, Sts. Sergius and Bacchus was built after (and possibly immediately after) Justinian had established a connection between his former residence and the Great Palace (described as "the rest of the palace" or the "other parts of the palace"; τοῖς ἄλλοις βασιλείοις). Such a sequence of events would make very good sense, since it would mean that the Palace of Hormisdas was already attached to the Great Palace by the time work started on the church of Sts. Sergius and Bacchus, and that consequently the new church would have been easily accessible to the imperial couple right from the time of its completion.24 Therefore, I suggest that it is wrong to view the church [End Page 71] as having been built in an inaccessible former residence, and that there is no justification for arguing that its location is best explained by assuming that it was built, or at least begun, during Justinian's residency in the Palace of Hormisdas. I hasten to add, however, that this does not mean that we should not ask why Justinian determined to connect his new residence with his former one, and to continue to build inside the Palace of Hormisdas once he had moved into the Great Palace.
At this point, a second passage in the Buildings deserves our attention, for it provides a potentially valuable chronological indication.25 Having described Justinian's ecclesiastical foundations in Constantinople, Procopius mentions the burning of many noteworthy buildings in the city and in the Great Palace during the Nika Riots of January 532, explaining that Justinian "undertook to rebuild them and to restore them all in more beautiful form." Having noted the rebuilding of the propylaea of the Great Palace and the Chalke Gate, the House of the Area,26 the Baths of Zeuxippos, the stoas and other buildings alongside the Mese as far as the Forum of Constantine, he goes on to mention work on the Palace of Hormisdas:
πρὸς ἐπὶ τούτοις δὲ τὴν Ὁρμίσδα ἐπώνυμον οἰκίαν, ἄγχιστα οὖσαν τῶν βασιλείων, παραλλάξας τε καὶ ὅλως ἐς τὸ ἐπιφανέστερον μεθαρμοσάμενος, ὡς τοῖς βασιλείοις ἐπιεικῶς πρέπειν, τῷ Παλατίῳ ἐντέθεικεν, εὐρύτερόν τε αὐτὸ καὶ πολλῷ ἔτι μᾶλλον ἀξιώτερον ταύτῃ ἐξείργασται.
And besides, in addition to these, the house named after Hormisdas, which is very close to the palace, he altered and altogether transformed to greater prominence so as to be properly fitting for the palace, and he has grafted it onto the palace, and it was in this way made broader and much more worthy.27
We know from Procopius's previously discussed account that Justinian's beautification of his private residence began, before he became emperor, with the construction of Sts. Peter and Paul. Here Procopius draws particular attention to Justinian's decision to attach that residence to the Great Palace, so making the palace as a whole much more extensive. Although there cannot be absolute certainty about it, the context in which Procopius has placed this information suggests that he considered the fusion of the two residences to be part of the extensive program of construction that Justinian embarked upon [End Page 72] after the events of January 532. We do not possess any sources stating that the Palace of Hormisdas was itself affected by the fires that were started during the Nika Riots,28 but it is not unlikely that the necessity of extensive repair work on nearby parts of the Great Palace provided an ideal opportunity to connect the two.
There exists further evidence that is of relevance for establishing the date at which the Palace of Hormisdas was connected to the Great Palace. Probably in the spring of 532, and therefore shortly after the Nika Riots, theological discussions were held between Chalcedonians and non-Chalcedonians in the Palace of Hormisdas. A Syriac account—anonymous, but believed to have been written by the abbot John bar Aphtonia, who was present—records that "the order (came) for the two parties to assemble in the hall known as Beth Hormisdas, which is today [sic] joined to the Palace."29 In the first place, this evidence confirms the accuracy of Procopius's claims that some sort of connection was indeed established between the Palace of Hormisdas and the Great Palace. In the second place, it strongly suggests that the two palaces were joined after the conversations took place in 532, but before the summary of the proceedings was written down, which perhaps happened not long after.
I have observed that the passage in which Procopius describes the construction of Sts. Sergius and Bacchus leaves the reader with the impression that the church was built after the connection between the two palaces had been established. Since two sources would lead us to believe that the connection was made after 532, I suggest that the terminus post quem for the church's construction should be revised to 532.30 The terminus ante quem does not require adjustment. It is provided by the facts that the signature of one Paul, abbot of Sts. Peter and Paul and Sts. Sergius and Bacchus, appears on the petitions of the council of 536, and that the missing non-Chalcedonian Patriarch Anthimus was sought "in the blessed chapel of the holy martyr Sergius in Hormisdas' [Palace]" in May of the same year.31 It would therefore appear that Sts. Sergius and Bacchus was built in the period 532–536, which would make it contemporary with Hagia Sophia. [End Page 73]
Industrial Marks and Sculptured Monograms I have presented the strongest evidence we possess regarding the date of the church. Some weaker evidence, which is subject to uncertainties of interpretation, also deserves to be mentioned. It does not refi ne the dating parameters, but is at least not inconsistent with the conclusion already reached. First, some tentative evidence from masons' marks might be taken to add weight to the conclusion that Sts. Sergius and Bacchus was built at more or less the same time as Hagia Sophia.32 Although only a very few marks have been noted in Sts. Sergius and Bacchus (fig. 3),33 two of them (both occurring more than once, and one quite distinctive) have also been recorded amongst the numerous marks in Hagia Sophia: ΖѠ and A.34 Any single mason's mark—which presumably identified a mason or the master mason he worked under—is likely to have been inscribed for many years throughout a mason's or master mason's career. Therefore we would require far larger numbers of corresponding marks at both churches to confirm that the buildings were absolutely contemporary. Nevertheless, the evidence is suggestive.
In the second place, a single brickstamp of sixth-century style found at the church carried a twelfth indiction.35 In the period before 536, this indiction could indicate that the brick was manufactured in 518/9 or 533/4. Clearly, the earlier date could not accurately reflect the date of construction of the church given that the entablature inscription was carved no earlier than 527. By contrast, the later date for the brick's manufacture would agree nicely with the construction period 532–536 suggested by the textual evidence. However, the evidence of a single brickstamp, the published drawing of which is not as careful as we might have wished, must be treated with great caution because we cannot be sure that the brick was found in situ in original masonry, and because there were occasions on which stockpiled or reused bricks were employed—although this is less likely in a period of such intensive construction.36 [End Page 74]
[End Page 75]
Two further pieces of evidence have been brought into the discussion of the dating of Sts. Sergius and Bacchus in earlier studies: the monograms on the capitals, and the dedicatory inscription on the entablature. The essential point is that neither the monograms—some of which, as we shall see, refer to Justinian as "Basileus" and to Theodora as "Augousta"—nor the poem's descriptions of the emperor as "scepter-bearing" (σκηπτοῦχος) and of his wife as "God-crowned" (θεοστεφέος) would contradict the proposition that the church was built between 532 and 536, since the two were ruling together throughout that period.37
I might have finished my discussion of the church's date at this point were it not for the fact that Croke has adduced certain evidence, which cannot be ignored, in support of dating the church before 527. First, Croke stresses that the terms applied to Justinian and Theodora in the monograms and the inscription could have been applied to them as early as April 527, when Justinian became co-Augustus with Justin and when Theodora became Augusta.38 That is undeniably possible. Croke also claims that the poem would have been inscribed towards the end of construction, as the interior was being decorated, so that it is at least conceivable that Justinian and Theodora did not hold imperial rank when work started.39 Again, I am in agreement, at least to the extent that that the inscription must have been carved after the entablature blocks were put in place on top of the piers and columns, because inspection reveals that a number of letters in the inscription were carved across the joints between adjacent blocks. Croke conjectures that building work is most likely to have started several years before April 527, claiming that such a dating would provide a satisfactory explanation for the church's location in the Palace of Hormisdas.40
In support of his early date, Croke cites evidence from the monograms carved on the church's column capitals. This evidence, and the validity of Croke's interpretation of it, therefore deserves attention. The monograms in the church were first published in 1895 by Harold Swainson,41 whose record is curious for a number of reasons. In the first place, all the monograms are [End Page 76] box-shaped, being based around a letter with two verticals (Ν) or two letters each with a single vertical (Β and Ε or Ε and Ρ). The box monograms are to be resolved "Of Justinian" (ΙΟΥϹΤΙΝΙΑΝΟΥ), "Of the Basileus" (ΒΑϹΙΛΕѠϹ), and "Of Theodora" (ΘΕΟΔѠΡΑϹ). Swainson's report contains no cruciform monograms, that is, monograms in which the letters are placed at the center and at the ends of the arms of a cross. This is odd because cruciform monograms reading "Of Theodora," with the letters of her name arranged in a variety of ways, occur in other major Justinianic churches, such as Hagia Sophia, Hagia Eirene (both built soon after the Nika Riots of 532), and St. John at Ephesus.
In the second place, Swainson records only two box monograms reading "Of Theodora," as compared with ten reading "Of Justinian" and nine reading "Of the Basileus." Thus it appears that monograms of the emperor significantly outnumbered those of the empress.
In the third place, Swainson's report suggests that there are no monograms reading "Of the Augousta" (ΑΥΓΟΥϹΤΑϹ). Swainson explains their absence on the grounds that Theodora had not yet been made Augusta when the church was built,42 but this cannot be the case because the poem describes Justinian as "scepter-bearing" and Theodora as "God-crowned." The absence of such monograms is odd, because at Hagia Sophia and Hagia Eirene, there are column capitals bearing box monograms reading "Of the Augousta."
On the basis of Swainson's documentation, I suggested in 2000 that the cruciform style of monogram might not have been introduced by the time the capitals of Sts. Sergius and Bacchus were carved, and that therefore its capitals might have been made slightly earlier than those in Hagia Sophia and Hagia Eirene, probably no later than 533.43 This deduction was not only flawed but also unfortunate, since, seizing upon it, Croke went further, using the apparent absence of cruciform monograms to suggest that Sts. Sergius and Bacchus could have been built considerably earlier than Hagia Sophia and Hagia Eirene. He argued that the evidence of the monograms could be reconciled with his proposal that the text of the entablature inscription was carved fairly soon after Justinian became co-Augustus in April 527.44
A more careful consideration of Swainson's record would have suggested that its curiosities are probably to be explained by the fact that many of the monograms on the capitals in the church have been damaged, and most probably deliberately destroyed. The under-representation of Theodora among the [End Page 77] monograms and the apparent absence of cruciform monograms would both be explained if some of the empress's monograms had in fact been cruciform, since in that case it is likely that they would have attracted the attention of disapproving Muslims.
That there has been a loss of monograms emerges from even a cursory examination of the capitals, whose monograms have suffered badly, especially at ground-floor level.45 There, nine of the eighteen capitals are damaged on the side facing the nave, where monograms would have been expected. In the gallery, where the monograms are again located on the nave-side of the capitals, two of the fourteen are missing.
Since Swainson does not indicate the locations of the monograms he records, since his drawings are not entirely accurate (perhaps because, at the time he was working, their exact form was obscured by layers of whitewash), and since restoration work in 2003–2006 exposed two capitals with monograms—presumably hidden when Swainson wrote—in the ground-floor triple arch in the north wall of the church,46 it is desirable to document them again (fig. 4).47
Swainson records ten monograms reading "Of Justinian," whereas today there are only eight (nos. 5, 9, 15, 18, 20, 22, 23, 27), one of which (no. 18) was probably not visible in Swainson's day. Swainson records nine monograms reading "Of the Basileus," whereas today there are only seven (nos. 10, 13, 16, 19, 21, 24, 28). On the other hand Swainson notes only two monograms reading "Of Theodora," whereas today four are visible (nos. 17, 25, 29, 31), one of which (no. 17) was probably not visible in Swainson's day. The discrepancies may be partly due to the loss of monograms after Swainson made his record, and partly due to inaccurate recording on Swainson's part. Furthermore, and in contradiction of Swainson's denial of their existence, it appears that there are two monograms that should be read "Of the Augousta": one of the box [End Page 78]
[End Page 79] type on an Ionic impost capital in the gallery (no. 30),48 and one of a cruciform type on a melon capital on the ground floor (no. 4).49 Parallels for the box type of the "Of the Augousta" monogram are known at Hagia Sophia and Hagia Eirene,50 but I know of no published parallel for the cruciform example, which is in fact the only surviving monogram of the cruciform type in the church.
The juxtapositions of the surviving monograms might be taken to suggest that they were arranged to form meaningful pairs: "Of Justinian" beside "Of Theodora" (nos. 17 and 18); "Of Justinian" beside "Of the Basileus" (nos. 9 and 10, 15 and 16, 19 and 20, 21 and 22, 23 and 24, 27 and 28); "Of Theo-dora" beside "Of the Augousta" (nos. 29 and 30). Assuming, for the sake of argument, that less meaningful pairings, such as "Of the Augousta" with "Of Justinian," or "Of the Basileus" with "Of Theodora" were avoided, we may very tentatively suggest the readings of some of the missing monograms. Monogram nos. 14, 26, and 32 perhaps read "Of Justinian" or "Of the Augousta"; monogram nos. 3 and 6 perhaps read "Of the Basileus" or "Of Theodora"; and monogram nos. 1, 2, 7, 8, 11, 12 are open to all four possibilities.
In conclusion, given the extensive damage to the monograms in Sts. Sergius and Bacchus, nothing should be deduced about the date of the church on the basis of the absence of certain monograms or monogram-types. All that can be said is that the monograms could have been carved at any time after April 527, when Justinian became Justin's co-Augustus and Theodora became Augusta. Consequently, I see nothing in either the inscription or the [End Page 80] monograms that should cause us to doubt our texts when they suggest the period 532–536 for the construction of Sts. Sergius and Bacchus.
It seems, therefore, that Justinian and Theodora, who had moved into the Great Palace upon the death of Justin in August 527, at first neglected their former residence in the Palace of Hormisdas, which they had adorned with a basilica of Sts. Peter and Paul; but, sometime after 532, they decided to connect that palace with their new abode. After the two palaces had been joined, but before 536, the imperial couple built the church of Sts. Sergius and Bacchus beside the church of Sts. Peter and Paul. Since these two churches are known to have shared an abbot in 536, it is clear that Sts. Sergius and Bacchus was, by that year at the latest, a church for a monastic community, not a church reserved only for the emperor's private use.
Sergius, the Non-Chalcedonians, and the Persian Frontier
At the Council of Chalcedon in 451, a statement of faith was established pronouncing Christ to have two natures (physeis)—one divine, one human—that came together in one person (prosōpon) and one hypostasis. Christ's divinity was established as being consubstantial with God, and his humanity as consubstantial with mankind.51 Many refused to accept the definition, holding that Christ's divine and human attributes, although distinct, were united without separation in one incarnate nature. They were scorned as "Monophy-sites" by supporters of Chalcedon, but are today described, out of ecumenical sensitivity, as "Miaphysites" or "non-Chalcedonians."52
Following the death of the non-Chalcedonian emperor Anastasius, Justin I reconciled with the Roman Church in 519 and enforced the libellus of Pope Hormisdas, forcing non-Chalcedonian bishops to decide whether they would accept it or go into exile.53 In 532, Justinian reviewed his uncle's harsh policy and convened discussions in the Palace of Hormisdas in the hope of achieving a theological formula that would satisfy both sides.54 However, after a council in Constantinople in 536 condemned the non-Chalcedonians, Justinian's ability to make concessions became severely limited.55 Persecution spread quickly through the eastern provinces.56 [End Page 81]
Croke argues strongly that non-Chalcedonians fleeing persecution sought refuge in Constantinople in large numbers only after 536.57 Consequently, doubt has been cast on Mango's proposal that the church of Sts. Sergius and Bacchus was built for the community of non-Chalcedonian refugees that Theodora housed in the Palace of Hormisdas.58 While it must be acknowledged that Croke's position is entirely plausible, it is nevertheless uncertain.59 If accepted, it would require the rejection of my own proposal that John of Ephesus alludes to the construction of Sts. Sergius and Bacchus in his account of the refugee community.60 John tells the story of the construction, in the Palace of Hormisdas, of a vaulted building to replace a hall that collapsed while the refugees were worshipping inside. If the refugees arrived only after 536, this vaulted building could not be Sts. Sergius and Bacchus, which already existed before that time.
In addition, like Krautheimer and Mathews before him, Croke rejects as unlikely Mango's proposal that abbot Paul, whose signature appears on the petitions of the pro-Chalcedonian council of 536, could have headed a community of non-Chalcedonians. Rather, he claims, Paul's monastery of Sts. Peter and Paul and Sts. Sergius and Bacchus was a Chalcedonian one, and the refugees, when they arrived, would have lived and worshipped in the halls of the Palace of Hormisdas, but would not have had access to either of the churches.61 Indeed, John of Ephesus describes the refugees worshipping in one of the palatial halls, thereby giving the distinct impression that they had no church in which to do so.62
Even if we were to accept Croke's position, the possibility of a connection between the construction of Sts. Sergius and Bacchus and the [End Page 82] non-Chalcedonians would remain. Although Procopius and others describe the church as dedicated to both Sergius and Bacchus, the dedicatory inscription in the church in fact only mentions the former.63 Even at Rusafa, where the major shrine of Sergius was located, there was no shrine for his co-martyr; and in the Passio, which tells the story of both martyrdoms, Bacchus plays a minor role.64 Sergius was not a saint venerated exclusively by non-Chalcedonians,65 but he certainly enjoyed great popularity with them. In the 510s, for instance, the non-Chalcedonian emperor Anastasius arranged for the relic of Sergius's thumb to be translated to Constantinople, and elevated Rusafa to the rank of a metropolitan see.66 Sergius's popularity in the East of the empire, where the non-Chalcedonian heartlands were concentrated, means that if, in the period in which Justinian's church was built, we can detect any connection between the Palace of Hormisdas and the non-Chalcedonians, then such a connection should not be overlooked as a possible influence on Justinian's decision to build a church dedicated to St. Sergius within that palace.
I have observed that the decision to locate the church in the Palace of Hormisdas becomes more understandable if we accept that that residence had earlier been joined to the Great Palace. Nevertheless, one is left wondering why the imperial couple chose to build the church in the Palace of Hormisdas in particular, rather than elsewhere, and why they built it at that specific time, having (apparently) neglected the Palace of Hormisdas and its church of Sts. Peter and Paul from 527 until sometime after 532.
In 1975, Mango refined his previously expressed view that the church of Sts. Sergius and Bacchus was built specially for the non-Chalcedonian refugee community in the Palace of Hormisdas, suggesting rather that the church had been built for the non-Chalcedonian bishops who attended the theological discussions held in that palace in 532, and who remained in Constantinople for a year afterwards.67 These discussions involved five pro-Chalcedonian bishops and five or more anti-Chalcedonian bishops. They lasted for three days and aimed at reunification of the Churches.68 I have already argued that Sts. Sergius and Bacchus was built after (but perhaps shortly after) the conversations of 532. Therefore, the church cannot have been built in anticipation of the talks. Nevertheless, if a connection could be demonstrated between the construction of the church and the discussions, it would certainly provide a [End Page 83] plausible explanation for the church's location, dedication, and the timing of its construction.
Croke notes Mango's suggestion but does not engage with it as distinct from the suggestion that the church served a community of refugees resident in the Palace of Hormisdas. Having argued that the refugees did not arrive in Constantinople in large numbers until 536, and that therefore Sts. Sergius and Bacchus could not have been built specially for them, Croke goes on, unjustifiably, to deny any possible connection between the construction of the church and Justinian's dealings with the non-Chalcedonians.69 However, it seems to me that the combination of the following circumstances is unlikely to be entirely coincidental: first, that soon after discussions with the non-Chalcedonians had been held in the Palace of Hormisdas in 532, the imperial couple saw fit to connect that palace to the Great Palace, even though they had been satisfied to have no direct access to it for the preceding five years; second, that in the same palace in which the conversations with the non-Chalcedonians had been held they decided to dedicate a church to a saint who was particularly popular in the East; and third, that this church was built between 532 and 536—the very period of Justinian's détente with the non-Chalcedonians—alongside a church dedicated to Sts. Peter and Paul, of whom the former was considered the founder of the Roman Church.
I do not think it is implausible to imagine that, after the conversations of 532, during the period in which Justinian was engaged in delicate dealings with the non-Chalcedonians to bring them back into communion, he decided to connect the Palace of Hormisdas, where the negotiations had taken place, with the Great Palace, and to build within Hormisdas's palace a church dedicated to a martyr who was particularly popular in the East as a pendant to the existing church of Sts. Peter and Paul. The church of Sts. Peter and Paul was allegedly almost complete in June 519, and it may credibly be suggested that its completion commemorated the restoration of Chalcedonian orthodoxy to Constantinople and the end of the schism with the papacy on 31 March that year, for Justinian is known to have communicated with the pope to negotiate the reconciliation.70 By building Sts. Sergius and Bacchus alongside the church of Sts. Peter and Paul, it seems plausible to suggest that Justinian intended to make an architectural statement. The juxtaposition of the two churches could symbolize the hoped-for understanding between the Chalcedonians (aligned with Rome and represented by St. Peter) and the non-Chalcedonians [End Page 84] (represented by Sergius, guardian of the eastern frontier, where the non-Chalcedonian heartlands were located). It seems to me, therefore, highly likely that both the location and the dedication of the church were influenced by the involvement that Justinian had with the non-Chalcedonians in the years before he resorted to persecution in 536.
In May of 536, a council condemned Patriarch Anthimus for pretending to adopt the Chalcedonian definition.71 He was sought in the "Apostoleion" (the church of Sts. Peter and Paul) and "in the blessed chapel of the holy martyr Sergius in Hormisdas' [Palace]" (ἐν τῷ σεπτῷ εὐκτηρίῳ τοῦ ἁγίου μάρτυρος Σεργίου ἐν τοῖς Ὁρμίσδου), where he had spent time before becoming patriarch in 535.72 Was his prior connection with the Palace of Hormisdas the only reason that his pursuers turned their attention to this location? Does the focus of their search not also suggest that the Palace of Hormisdas was regarded as a likely place of refuge from religious persecution? Is that not because, after the discussions of 532 and the construction of the church of Sts. Sergius and Bacchus, which was symbolic of religious reconciliation, the Palace of Hormisdas was considered a place in which non-Chalcedonians might be accepted?73 Furthermore, is it not because the Palace of Hormisdas was already associated with the hope for religious accord that Theodora decided to use the halls of that palace as makeshift accommodation for the non-Chalcedonians who came to Constantinople from far and wide when Justinian resumed persecution in 536 (even if those refugees were not granted access to the two churches of the monastery)?
This possibility need not be seen as excluding other influences on the dedication of the church. Irfan Shahîd has emphasized that Sergius was a military martyr who was buried near the frontier with Persia, and suggests that Justinian decided to dedicate a church to him to invoke his power against the Persian threat. Since we have established that the church was begun not in 527, when war with Persia broke out, but in 532, when Justinian negotiated peace, the church would instead have to be seen as invoking Sergius's power for the continued protection of the empire.74 The suggestion is not implausible,75 since the dedicatory inscription, which is quoted in full below, asks the saint [End Page 85] to "guard the rule of the sleepless sovereign." That said, Justinian might have invoked almost any saint for protection, since all were perceived as guardians, champions, and allies.76 Indeed, the lengthy dedicatory inscription inside Anicia Juliana's church of St. Polyeuktos invokes the protection not only of Polyeuktos but also of all the other saints to whom Juliana had built shrines: "Wherefore, you servants of the heavenly King to whom she gives gifts and to whom she built temples, protect her readily with her son and his daughters."77
That multiple factors may have influenced Justinian's choice of martyr has been well expressed by Elizabeth Key Fowden: "Seen in the context of two of Justinian's most considerable preoccupations during his reign—reconciliation between Chalcedonians and their opponents, and maintenance of the eastern frontier—it was a wise decision to ally himself and his empress with the most influential martyr in the frontier zone, a military saint who enjoyed great popularity among the region's Arab and mainly, but not exclusively, non-Chalcedonian inhabitants."78
In summary, the decision to allow non-Chalcedonian conference delegates to meet in the Palace of Hormisdas in 532 might well have sparked renewed interest in the Palace of Hormisdas on the part of the imperial couple, and it is possible that the decision to dedicate a church to Sts. Sergius and Bacchus, and to place it alongside a church dedicated to Sts. Peter and Paul, was inspired by, and symbolic of Justinian and Theodora's efforts to reconcile the Oriental Church with the Eastern and Western Churches. The fact that Sergius was a defender of the eastern frontier would have been a significant additional attraction for the emperor.
The Entablature Inscription
Croke, denying that the construction of Sts. Sergius and Bacchus was in any way connected with Justinian's dealings with the non-Chalcedonians, or with the threat from the Persians, seeks another explanation for Justinian's decision to build the church. He suggests that it was built because of Justinian's rivalry with the wealthy aristocrat Anicia Juliana, and that it was dedicated to Sergius because he was a much more potent and popular eastern martyr than Polyeuktos, to whom Juliana had dedicated a magnificent church.79 This theory rests, in part, on Croke's early dating of the church of Sts. Sergius and Bacchus, according to which work was sufficiently advanced by 527 for the carving of the entablature inscription to take place. Croke finds it highly [End Page 86] significant that, according to his dating, Sts. Sergius and Bacchus was erected very soon after Juliana built her church of St. Polyeuktos, the superstructures of which were built with bricks manufactured between 518 and 522.80 However, since it now seems more likely, from the arguments advanced previously, that the church of Sts. Sergius and Bacchus was not built until 532–536, and therefore about a decade after St. Polyeuktos, the force of Croke's argument for a rivalry manifested in architecture is reduced.
The other evidence that Croke adduces for this alleged architectural competition between Justinian and Juliana comes from the short verse inscription on the 29 marble blocks that form the entablature of Sts. Sergius and Bacchus. The poem is inscribed horizontally in a single line of text, running in a clockwise direction from the pier at the east-southeastern corner of the nave. This arrangement is similar to that in the earlier church of St. Polyeuktos, where a much longer poem of 41 verses was inscribed in the nave, beginning at the eastern end of the southern colonnade.81 In that church, however, the entablature was not continuous but carried on two colonnades, one on either side of the nave, each supporting 19 entablature blocks.82 The text of the inscription was therefore broken at the end of verse 21 by the west wall of the nave; it resumed at the west end of the northern colonnade with the beginning of verse 22.83
The poem carved in Sts. Sergius and Bacchus may be translated thus:
Other sovereigns have honored dead men whose labor was unprofitable, but our scepter-bearing Justinian, fostering piety, honors with a gleaming abode the servant of Christ, Begetter of all things, Sergius, whom not the burning breath of fire, nor the sword, nor any other constraint of tortures confounded; but who endured to be overcome for the sake of Christ, the God, gaining by his blood heaven as his home. May he in all things guard the rule of the sleepless sovereign and increase the power of God-crowned Theodora, whose mind is bright with piety, whose constant toil lies in unsparing efforts to nourish the destitute.84
I reproduce the Greek text of the inscription as edited by Silvio Giuseppe Mercati,85 and, since the information has not previously been set forth, I also [End Page 87]
[End Page 88]
indicate how the letters are distributed on the 29 blocks of the entablature (figs. 5–6).86 For each block, I give the number I have assigned to the block, a code indicating the type of block, and the number of letters on the block. The divisions between adjacent blocks are indicated by a vertical line. Where a single letter is carved across two blocks (even if only partially), a vertical line has been placed on either side of that letter. The ivy leaves that mark the end of each verse are sufficiently large to warrant their being counted as a letter (and in some cases possibly two letters) in their own right. They are significantly more prominent than the superscript dots used to indicate the ends [End Page 89] of the verses of the inscription in St. Polyeuktos.87 The first ten letters of the inscription on block 1 are no longer visible, being obscured by the Ottoman mimbar. On block 29, the final letter is only partially preserved, and it seems likely that an ivy leaf has been lost. It may be noted that block 19 carries no punctuation indicating the end of a sentence, nor an apostrophe at ἀλλ' ἐνὶ. This is in contrast to the inscription in St. Polyeuktos, where both an apostrophe and a low dot are attested.88 Long diphthongs are carved with adscript iotas (blocks 10 and 25).
The same type of block can carry varying numbers of letters. If (for the sake of argument) we count the ivy leaf that is used to divide verses as if it were a single letter, pier blocks (p) carry between 18 and 32 letters; concave exedra blocks (e) carry between 9/10 and 15/17 letters; short blocks in the flat-faced entablature between exedras (fs) carry between 9 and 13 letters; and long blocks in the flat-faced entablature between exedras (fl) carry between 8/9 and 16/17 letters. Clearly, there are significant variations between the numbers of letters on blocks of the same type.
The few inscribed entablature blocks that survive from Anicia Juliana's church of St. Polyeuktos indicate that there was a similar variation in the density of letters there. For example, one concave block, on which the inscription arches over a peacock-niche, must, when complete, have carried all 31 letters of verse 31 of the dedicatory poem. However, the surviving left-hand half of the block carries as many as 19 letters to the crown of the niche, leaving only 12 letters to fill the other half of the block, before verse 32 begins at the left-hand end of the surviving adjacent block.89 Clearly, with more careful planning, the letters might have been more evenly distributed on this block.
At St. Polyeuktos, the 41-verse poem that was to be inscribed on 38 entablature blocks in the nave was, evidently, composed with an awareness that it would be divided into two halves when inscribed in the building. The first half of the poem, inscribed on the south side of the nave, opens at verse 1 with the name of Eudokia, and ends at verse 21 with the end of a sentence. The second half of the poem, inscribed on the north side of the nave, begins at verse 22 and has as its third word the name of Eudokia's great-granddaughter, Juliana. When composing a poem as short as that inscribed in Sts. Sergius and [End Page 90] Bacchus, the author would doubtless have found it difficult to ensure that the names of Justinian and Theodora were positioned in such a way that, when carved, they would fall in an architecturally significant position. Nevertheless, it does seem that the distribution of the text on the entablature was contrived so that references to "scepter-bearing Justinian" (blocks 7–8) and "God-crowned Theodora" (blocks 22–24) should fall directly opposite each other on the sections of flat-faced entablature to the south and north of the nave.90 But even if that explains the variation in the spacing of letters on some of the blocks, the spacing of the text on blocks 9–21 might have been better planned to avoid the noticeably congested text in the western section of flat-faced entablature (block 15) and on pier 6 (block 21).
The poem has linguistic parallels with the earlier and much lengthier poem carved in Anicia Juliana's church.91 However, the similarities can reliably be taken only to suggest literary borrowing on the part of Justinian's poet, not necessarily a public power contest between the patrons of the two buildings. Evidence for the latter has been sought in the opening words of the poem: "Other sovereigns have honored dead men whose labor was unprofitable, but our scepter-bearing Justinian, fostering piety, honors with a gleaming abode the servant of Christ, begetter of all things, Sergius." Here the poet employs an ancient rhetorical technique (akin to priamel), using ἄλλος to introduce one thing before another with which it is contrasted.92 Thus he draws attention to Justinian's achievement by first mentioning the lesser achievements of others. Croke claims that the references to "other sovereigns" (Ἄλλοι μὲν βασιλῆες) and "dead men whose labor was unprofitable" (θανόντας ἀνέρας, ὦν ἀνόνητος ἔην πόνος) are not to be "considered as pure literary affectation," asserting instead that Justinian had "particular targets in mind." Croke argues that the "other sovereigns" were Eudokia, founder of the church of St. Polyeuktos in Constantinople and wife of Theodosius II, and Juliana, her great-granddaughter, who completely rebuilt that church. It is a suggestion also made by Shahîd, who makes the claim that the poet was somehow contrasting Justinian's Christian church with the numerous secular (and therefore inferior) monuments for which the House of Theodosius had [End Page 91] been responsible.93 Shahîd even goes on to make the incredible assertion that the poem's praise for the care Theodora gave to the destitute would have been sufficient to make Juliana "look absurd."94 For his part, Croke observes that Polyeuktos was "a relatively obscure saint" by comparison with Sergius, whose cult was rapidly expanding in the East; and, furthermore, points out that there was doubt about whether Polyeuktos had ever been baptized into the Christian faith.95 Thus, Croke argues, Polyeuktos' labor might have been considered unprofitable.
Yet, however attractive it may be to identify these "other sovereigns" with specific individuals, the truth is that the "dead men" whom they honored might be any Christian martyrs or, as we shall see, pagan gods. Consequently, the inscription's opening assertion might rather be read as a fairly standard hyperbolic claim that Justinian's church was better than any place of worship built by any previous ruler.
A text such as that inscribed in Sts. Sergius and Bacchus is particularly hard to interpret because of the difficulty of establishing the extent to which the themes that it addresses are original or traditional. For example, what inspired the poet to describe Justinian with the epithet "sleepless" (βασιλῆος ἀκοιμήτοιο)? Is this a statement of fact, a literary commonplace, or both? In an article discussing this aspect of the emperor's character, Croke acknowledges that some earlier emperors had been praised by panegyrists for working late,96 but argues that it was Justinian who first personally publicized the habit as a virtue—the inscription in Sts. Sergius and Bacchus being one illustration of this.97 This is, I think, essentially true; however, Croke may be mistaken in stating that, prior to Justinian, vigilance "had never been one of the canonical imperial virtues."98 While Dio Chrysostom wrote his Orations on Kingship under Trajan and addressed his remarks to that emperor, he was nevertheless making recommendations for the conduct of any ideal ruler. In his First Kingship Oration, he echoed Homer in describing a good king as "a guide and shepherd of his people" who "ought to be just such a man as to think that he should not sleep at all the whole night through as having no leisure for idleness."99 This connection between wakefulness and the king's [End Page 92] traditional role as shepherd of his people or helmsman of the state seems to have been fairly well established. For instance, in Aeschylus' Seven Against Thebes, Eteocles described a good ruler as a helmsman who possessed "eyes not drowsy with sleep" (βλέφαρα μὴ κοιμῶν ὕπνῳ), and, in his Thebaid, Statius compared the general Adrastus to a solitary wakeful helmsman (solus stat puppe magister | pervigil).100 That this long-standing connection between emperor and wakeful shepherd or helmsman was the origin of the expression in the inscription is strongly suggested by Agapetus's statement, made in the 530s, that "the many-eyed intellect of the emperor [Justinian] remains ever-vigilant, holding secure the rudder of good government and firmly pushing back the torrents of lawlessness, so that the vessel of the universal state may not founder in the waves of injustice."101
It may be added that the prefaces to Justinian's laws regularly refer to the emperor's unceasing toil in drafting appropriate legislation, and Croke duly draws attention to the fact.102 In these instances, again, the claim may have been influenced by a tradition, one that goes back to Callimachus, of conveying the painstaking care taken by authors of both poetry and prose by making reference to their vigilance.103
I do not mean to suggest that Justinian did not deserve his reputation for working long and hard into the early hours, for our sources—including Procopius, who sought to turn the virtue into a vice—leave little doubt about this;104 but it seems likely that it was (in part at least) the established tradition of a good king remaining vigilant through the night that encouraged Justinian to emphasize this aspect of his character to justify his royal status. The term used to describe Justinian's wakeful state was also, as Croke recognizes, the term applied to the "sleepless monks" of the Stoudios monastery, who prayed in shifts to ensure that God received monastic devotion day and night.105 In the mind of the reader of the inscription, this is likely to have evoked a comparison between the emperor and the monks, so emphasizing Justinian's piety. And since respect for God, or the gods, was a traditional attribute of a good ruler because it ensured the security of the state,106 the term can only have served to further strengthen Justinian's claim to his imperial rank. [End Page 93]
The preceding observations illustrate that great care is needed to search out traditional themes in a poem before claiming to have found references to particular historical circumstances. For this reason, I believe that Croke's specific interpretation of "Other sovereigns have honored dead men whose labor was unprofitable" should be re-examined. So far as I have been able to establish, the inspiration appears to have come in particular from the book of Wisdom, in which kings who strive to attain the wisdom of God (by following His law and by worshipping Him) are contrasted with those who fall away from wisdom (into idol-worship).107 At one point in Wisdom, an explanation is given for the emergence of idol worship: "For a father, tormented by untimely grief, having made an image of the child who had been quickly taken away, now honored as a god what was once a dead human being (τὸν τότε νεκρὸν ἄνθρωπον) and handed on to his dependents mysteries and sacred rites. Then, when the impious custom had grown strong with time, it was kept as a law, and at the command of princes (τυράννων ἐπιταγαῖς) carved images were worshipped."108 The theme of worship of the dead was taken up by Lactantius in the fourth century in his Divine Institutes. Lactantius argued that all the ancient gods had once been men, and, as an illustration of how gods were merely the dead who had been divinized by the living, he referred to Cicero's expressed intention to consecrate an image in honor of his dead daughter. "Who," scoffed Lactantius, "is so stupid as to think that heaven [i.e., divine status] opens up for the dead at the consent and determination of a host of idiots [i.e., the misguided mortals who proclaim them as gods]?"109 We find this idea again in Eusebius' In Praise of Constantine and Life of Constantine, where it is observed that Licinius put his faith in "idols of the spent dead in the form of lifeless statues" (νεκρῶν εἴδωλα καμόντων ἐν ἀψύχοις ἀγάλμασι), whereas Constantine armed himself with the "saving and life-giving sign," and so proved victorious.110 In an epigram, Palladas recognized the irony of the situation in which he and his fellow pagans found themselves after Constantine's overthrow of Licinius. It had formerly been the Christians who had been scorned for placing their hopes in a crucified man, but now, "We Hellenes are men reduced to ashes, [by Christians who portray us as] holding to our buried hopes in the dead; for everything has now been turned on its head."111 Thus when the inscription in Sts. Sergius [End Page 94] and Bacchus contrasts Justinian with other sovereigns who honored dead men, there is no need to believe that it is contrasting him with contemporary aristocrats; it may simply be contrasting him with the kings of old who had ordered the worship of idols of dead men.
Elsewhere in Wisdom we read: "for the one who disdains wisdom and instruction is wretched—and their hope is vain, and their labors are unprofitable (οἱ κόποι ἀνόνητοι), and their deeds useless."112 Thus when Justinian's poet refers to kings honoring "dead men whose labor was unprofitable" (θανόντας ἀνέρας, ὦν ἀνόνητος ἔην πόνος), there is no reason to think he is referring specifically to the worship of St. Polyeuktos; rather, he may be making a general denunciation of the worship of dead men who, unlike the Christian martyrs, had lacked the wisdom to worship the Christian God.
The poem claims that St. Sergius stood firm in the face of persecution and was not confounded by fire or the sword, "nor any other constraint of tortures" (οὐχ ἑτέρη βασάνων … ἀνάγκη), so gaining heaven as his reward. According to tradition, Sergius was forced to run in shoes with long nails inside and was then beheaded by the sword.113 Since fire played no part in his martyrdom we should seek a source of inspiration for this imagery in the poem. An explanation for the poet's choice of words can be found in a generalization made in the book of Wisdom: "But the souls of the righteous are in the hand of God, and no torment will ever touch them (καὶ οὐ μὴ ἅψηται αὐτῶν βάσανος). … For even if in the sight of human beings they were punished, their hope is full of immortality … God tested them and found them worthy of himself; as gold in the furnace, he tested them, and as a sacrificial whole burnt offering, he accepted them."114 The description of Sergius remaining unconfounded by fire is therefore metaphorical rather than literal, and the metaphor is biblically inspired.
The rejection of pagan idols and the endurance of various tortures in honor of the Christian God were themes that had become commonplace by the time the inscription in Sts. Sergius and Bacchus was composed. To illustrate this, we need only note Paul the Silentiary's description of martyrs at the opening of his description of the ambo of Hagia Sophia, where he invokes them as: "you who the sword, the life-destroying lash, who the furnaces | gladly endured, and did not bend to dumb | idols your necks, but set the foundation of your spirit | steadfast as witnesses in the honor of the undefiled God."115 Paul, whose speech was composed to mark the re-dedication of the [End Page 95] church during Christmas and Epiphany 562–563, may, of course, have been familiar with the inscription in Sts. Sergius and Bacchus, but the biblical and hagiographical precedents must not be overlooked.
Clearly, then, there is nothing in the verses inscribed in Sts. Sergius and Bacchus that can be satisfactorily explained only as a specific reference to Eudokia, Anicia Juliana, and St. Polyeuktos. The biblical precedents allow the sentiments to be understood in much more general terms. Whilst it must be admitted that Justinian may have perceived Juliana as a rival in architecture even a decade after the latter had completed her church, and that the poetry in Sts. Sergius and Bacchus may carry both contemporary and biblical resonances, the poem does not in itself provide incontrovertible evidence that such rivalry inspired its construction.
Before we turn from the inscription in Sts. Sergius and Bacchus, some aspects of its legacy may be noted. Gianfranco Agosti has suggested that an inscription discovered in Zorava and published in 1960 was influenced by the Constantinopolitan inscription.116 It, too, is the dedicatory inscription of a church honoring St. Sergius, although in this case the building, the work of one Theodore, was formerly a temple. A number of apparent echoes of the poem from Constantinople may be noted. Theodore builds a "house of His [Christ's] servant, the well-mounted Sergius" (δόμον ἑοῦ θεράποντος εὐίππεός τε Σεργίου), just as Justinian "honored with a gleaming abode the servant of Christ, begetter of all things, Sergius" (Σέργιον αἰγλήεντι δόμῳ θεράποντα γεραίρει | Χριστοῦ παγγενέταο); in the Zorava text Sergius is described as having "accepted bitter tortures (πικρούς τε | βασσάνους ἐδέξατο) from head to foot," and in the Constantinopolitan inscription the saint is not confounded by fire, the sword, or any other "constraint of tortures" (βασάνων … ἀνάγκη). The ruler at the time of the church's construction is alluded to in the second line as "the holy king, who has destroyed the works of idols" (ἄνακτ' ἅγιον, ὃς εἰδώλων ὤλησεν ἔργα), and if the inspiration here was indeed the opening of the Constantinopolitan inscription, this would be an indication that the provincial poet understood the "dead men whose labor was unprofitable" (θανόντας | ἀνέρας, ὦν ἀνόνητος ἔην πόνος) to be the idolized pagan dead. [End Page 96] Furthermore, the unnamed ruler may have been Justinian, as the sixth-century style of the letter-forms would allow. All of this leads us to suggest that the Constantinopolitan inscription circulated widely in written form from the time of its composition in the reign of Justinian.
A second and a third inscription provide further evidence for such transmission. The second is the funerary inscription of Basil II, which was presumably composed around the time of the emperor's death in 1025.117 Like the Sergius and Bacchus inscription, which opens with the words, "Other sovereigns have honored dead men whose labor was unprofitable, but our scepter-bearing Justinian … honors … Sergius," Basil's inscription, too, uses the expression ἄλλοι μὲν … δὲ to express a contrast: "Other kings of old ordained other burial places for themselves, but I, Basil, born in the purple, place my tomb on the site of Hebdomon."118 The similarities with the poem from Sts. Sergius and Bacchus continue, since Basil has been endowed with characteristics of Sergius, Justinian, and Theodora: he is at once both an unsleeping (ἀγρυπνῶν) sovereign—compare Justinian as "sleepless sovereign," (βασιλῆος ἀκοιμήτοιο)—and a protective martyr-figure, who endures (ἐκαρτέρουν) endless toils (ἀμετρήτων πόνων) in battle—compare Sergius "enduring to be overcome" (τέτληκεν … δαμῆναι) and Theodora's "constant toil" (πόνος αἰεὶ). Basil guards (ἐρυόμην) the citizens of Constantinople, as Sergius is asked to guard (φυλάξοι) Justinian.
The third inscription is the mosaic dedicatory text, dated 1143, in the Cappella Palatina of King Roger II in Palermo.119 The inscription begins: "Other kings of old erected other places of worship to saints, but I, King Roger, sceptered ruler [dedicate this church] to the foremost of the Lord's disciples, the leader and archpriest Peter, to whom Christ entrusted His church [End Page 97] which He himself consecrated by the shedding of His blood."120 The phrasing of the contrast between Roger and earlier kings (Ἄλλους μὲν ἄλλοι τῶν πάλαι βασιλέων | σεβασμίους ἤγειραν ἁγίοις τόπους· | ἐγὼ Ῥογέριος δὲ), clearly follows the phrasing at the beginning of the tomb inscription of Basil II (Ἄλλοι μὲν ἄλλῃ τῶν πάλαι βασιλέων | αὑτοῖς προαφώρισαν εἰς ταφὴν τόπους, | ἐγὼ δὲ Βασίλειος).121 However, parts of the rest of the text echo the inscription from Sts. Sergius and Bacchus: Roger is a "sceptered ruler" (σκηπτροκράτωρ), just as Justinian is "scepter-bearing" (σκηπτοῦχος); St. Peter is "leader and archpriest" (ποιμενάρχῃ καὶ κορυφαίῳ), as St. Sergius is "servant of Christ" (θεράποντα … Χριστοῦ); Christ "consecrates the church by the shedding of His blood" (ἐκκλησίαν | ἣν αὐτὸς ἔσχεν αἵματος χύσει)122 as Sergius is described as "gaining by his blood heaven as his home" (αἵματι κερδαίνων δόμον οὐρανόν). The author of Roger's inscription apparently had access to both Constantinopolitan texts.
We should, perhaps, consider the possibility that Roger held Justinian's inscription to be a suitable model for the dedicatory text of his private chapel honoring St. Peter because Sts. Sergius and Bacchus flanked Justinian's earlier church of Sts. Peter and Paul, and was located close to, and connected with, the imperial palace. Roger may even have been aware of, and inspired by the fact that Justinian had used Sts. Peter and Paul to mark the restoration of religious unity between Rome and Constantinople,123 since Roger's own chapel was built shortly after Innocent II had been forced to recognize him as King of Sicily.124
How the Constantinopolitan inscriptions came to the attention of the Norman court in Sicily is uncertain. As Mercati has explained,125 the inscription from Sts. Sergius and Bacchus was interpolated into the late eleventh-century Synopsis of Byzantine History by John Skylitzes as part of a description of repairs to the church under Basil I. The interpolation appears in a number of manuscripts of Skylitzes, the earliest of which is the twelfth-century Vindob. [End Page 98] hist. gr. 35.126 It is therefore conceivable that a manuscript of Skylitzes was somehow responsible for the transmission of the text to Sicily. Indeed, there is good reason to believe Skylitzes's work reached the court of Roger, for the illustrated Matr. Vitr. 26–2 is now believed to date to the twelfth century and to have been produced in Palermo, quite possibly at Roger's behest.127 That particular manuscript, however, does not reproduce the interpolated passage with which we are concerned.128
Double-shell Palace Churches
Having established the date and circumstances of the construction of Sts. Sergius and Bacchus, it is fitting to turn briefly to the issue of double-shell palace churches, since, despite the criticism the concept has received, it continues to be defended.129 Krautheimer once argued that the decision to build a church on a double-shell plan—that is, a plan in which the supporting piers or columns for a central dome or tower are surrounded by an ambulatory, as in the case of Sts. Sergius and Bacchus—was to be explained by the location of the church within the palace and its being assigned to the emperor's personal use.130 But is the relationship between form and function really so straightforward?
At the outset, it must be made quite clear that there is no doubt that the church of Sts. Sergius and Bacchus was inside the Palace of Hormisdas and might therefore be accurately described as a palace church. Croke is wrong in claiming, on the basis of the record of the 536 synod, that the church was outside the Palace of Hormisdas,131 for when Paul styled himself presbyter and hegoumenos of the church "of the Apostle Peter near the Palace" (τοῦ ἀποστόλου Πέτρου πλησίον τοῦ Παλατίου), he meant that the basilica of Sts. Peter and Paul (and therefore also the adjacent church of Sts. Sergius and [End Page 99] Bacchus) was near (and therefore outside) the Great Palace, not that it was outside the Palace of Hormisdas.132 Indeed, elsewhere in the same document, as we have already noted, the church is unequivocally described as "in Hormisdas' [palace]" (ἐν τοῖς Ὁρμίσδου).133 Furthermore, the church of Sts. Peter and Paul, to which Sts. Sergius and Bacchus was attached, was said by Justinian to be in domo nostra, at the time when he still inhabited the Palace of Hormisdas.134 Admittedly, according to Procopius's account quoted earlier, Sts. Peter and Paul was built "beside" the Palace of Hormisdas (παρὰ τὴν βασιλέως αὐλήν), but by this Procopius apparently means to convey the idea that Justinian had it built immediately alongside the existing Palace of Hormisdas, so constituting an addition to it, and therefore a part of it.
Given that Sts. Sergius and Bacchus was built between 532 and 536, and was, together with the adjacent church of Sts. Peter and Paul, part of a monastery by 536 at the latest, it would be difficult to claim that it had ever been meant to serve as a private imperial foundation. In order to make such a claim, we would have to argue that Justinian, having connected the Palace of Hormisdas with the Great Palace after 532, and having built the church of Sts. Sergius and Bacchus at some time before 536 with the intention of using it for his own private use, very quickly changed his mind, handing it over to a monastic community by 536 at the latest. Personally, I find that scenario a little difficult to imagine.
It seems more probable that Justinian had decided that his new church of Sts. Sergius and Bacchus would be monastic from the outset, and it is conceivable that the church of Sts. Peter and Paul was already a monastic church before work began on the adjacent church of Sts. Sergius and Bacchus. Indeed, observing that, when the abbot Paul signed the petitions of the council of 536, he sometimes described himself only as the abbot of Sts. Peter and Paul (rather than of both Sts. Peter and Paul and Sts. Sergius and Bacchus), Mathews inferred that abbot Paul's monastery was primarily associated with Sts. Peter and Paul.135 That is certainly possible, and may mean that the monastery had been served by Sts. Peter and Paul alone for some years before Sts. Sergius and Bacchus was added. Mathews went further and claimed that the monastery was founded at the same time as Sts. Peter and Paul,136 a church known to [End Page 100] have been close to completion in June 519.137 That conclusion need not follow, however. Since Paul's signatures of 536 constitute the earliest evidence we have for the existence of the monastery, we cannot be sure how long before that year the monastery was founded. It is, therefore, entirely possible that Sts. Peter and Paul was built by Justinian, around 519, as a private church, and that it was later, sometime before 536, given over to a monastery. Indeed, it may be significant that when Procopius describes the construction of Sts. Peter and Paul, he indicates that it was built to adorn Justinian's private residence and makes no mention of a monastery.138 Nor is any monastery mentioned in the letters written to the pope requesting relics for the new church: Justinian merely states that the church was "in our residence" (in domo nostra).139 It may plausibly be suggested that Sts. Peter and Paul, a private church in 519, became a monastery after August 527, when Justinian and Theodora vacated the Palace of Hormisdas to take up residence in the Great Palace. In that case, Sts. Sergius and Bacchus, built between 532 and 536, would have been part of that monastery from the time of its construction, but, being an addition to that monastery (which had been served for at least five years by Sts. Peter and Paul alone), would perhaps have been considered a secondary church by the likes of abbot Paul.140
Following the construction of the connection between the Great Palace and the Palace of Hormisdas (a project undertaken after 532), Justinian and Theodora could easily have visited the two churches to celebrate the liturgy there.141 [End Page 101] However, since the scholarly definition of an "octagonal palace church" or "double-shell palace church" that was devoted (to use Mathews' words) to the emperor's "peculiar use" is rather vague,142 I am not sure whether such occasional imperial visits would be considered sufficient to place Sts. Sergius and Bacchus in the proposed category—particularly if the monastery had been created in 527 or earlier, so that the church of Sts. Sergius and Bacchus had from the outset been conceived as part of the monastery rather than as a private chapel.
It should be noted that Krautheimer recognized a distinction between churches that were outside palaces but nearby (which he called "palatine churches"), and churches within palaces (which he called "palace churches").143 The fact that churches outside palaces, but close by, have also been brought into the equation is again indicative of the ill-defined scope of the scholarly debate on this issue. Amongst the double-shell palatine churches are the Golden Octagon in Antioch-on-the-Orontes and Hagia Sophia in Constantinople. Krautheimer claimed that the choice of a double-shell plan for these two churches was in some way connected with their proximity to a palace and their function as cathedrals. However, it is doubtful that such a conclusion can safely be drawn from the available evidence.
In the first place, it is demonstrable that emperors did not insist that all churches built within or near palaces should be constructed on a double-shell plan. The plan of Sts. Peter and Paul was basilical, not double-shell,144 yet Justinian occupied the palace at the time of the church's construction, and must have used it as a private chapel if it was turned into a monastery only after he left that residence in 527. It might, I suppose, be argued that the church's basilical plan is to be explained by the fact that Justinian built the church before he had become emperor, and that he therefore refrained from commissioning a double-shell palace church.145 But even during the reign of Justin, Justinian's [End Page 102] future claim to the throne was strong,146 and we might have expected him to have made an architectural proclamation of his imperial aspirations by building a private double-shell church if such a tradition had existed.
The emperor Justin himself is known to have built a huge church in Blachernai, at the foot of a hill on which stood a suburban imperial palace. Close by was a spiral staircase leading up to the palace. This new church might reasonably be classified, to use Krautheimer's terminology, as a palatine church; yet it defies Krautheimer's postulated tradition, since its design was basilical, not double-shell. Its size reflected the importance of the healing cult in Blachernai.147 There was already another church on the site, the Soros, built in the fifth century, which was apparently a rotunda or a double-shell. But even if it was a double-shell, its origins are connected not with the nearby palace, which may have been established only later, in the reign of Anastasius, but with the holy spring at Blachernai.148
At about the same time as Justin built his church in Blachernai, a patron whose family had imperial connections and imperial designs was also building a church. Between 519 and 522, Anicia Juliana built St. Polyeuktos beside her palace on the northern branch of the Mese.149 According to Krautheimer's terminology, therefore, it was a "palatine church"; nevertheless, it was built on a basilical plan, although no expense was spared in its decoration.150 Again, it might be argued that centralized churches were the preserve of ruling emperors, but it seems much more likely that the design of Juliana's church was dictated by her desire to model her church on descriptions of the heavenly temple due to descend from the skies in the eschatological era.151 Had she wanted to do so, one imagines she could have built a centralized church. Indeed, one or more of the other churches she is known to have built may, conceivably, have been centralized.
There are even more questions that this debate raises. If Justinian, five years after he had ascended the throne, had determined to begin rebuilding Hagia Sophia on a double-shell plan because it was near the palace, why did he not choose a double-shell plan when he rebuilt the adjacent church of Hagia Eirene at the same time? Krautheimer might well have explained the difference in design on the grounds that Hagia Sophia was the cathedral, but that is to introduce into the equation a factor beyond the topographical relationship with the palace, and raises another question: if only imperial [End Page 103] cathedrals located in or near palaces were built on double-shell plans, why was Sts. Sergius and Bacchus, which was not a cathedral, built on a double-shell plan? Furthermore, double-shell churches could be built without any palatial association—witness Saint Michael at Anaplous, which stood next to a marketplace (agora) on the shore of the Bosphorus.152 Also, it seems unlikely that Justinian's church of St. John the Forerunner in the Hebdomon, which Procopius says was similar in form to that at Anaplous,153 was inside the Iucundianae Palace. The church has been identified with the remains of an octagonal double-shell church excavated in Bakırköy about 300 m from the shore, whereas the palace is known to have been near the harbor.154 Looking beyond Constantinople, the Justinianic church of San Vitale in Ravenna, a double-shell octagon roughly contemporary with Sts. Sergius and Bacchus, appears to have had no connection with a palace.155
Double-shell churches were probably not unusual, and, given the incomplete nature of the evidence available to us, it would be difficult to demonstrate statistically that they were more commonly located within or near imperial palaces than in locations that had no imperial or palatial links.
In short, Sts. Sergius and Bacchus was within the palace (that is, within the Palace of Hormisdas, which was connected with the Great Palace), and was apparently founded as a monastic church that would be put to occasional imperial use. Although it would be important if some reasoning could be discerned for the choice of either basilical or centralized designs when building churches, it seems to me that, in the case of Sts. Sergius and Bacchus, it is impossible, in the absence of more evidence, to determine whether its double-shell plan was necessitated in some way by its function and location. Rather than seeking functional or topographical reasons for the adoption of the distinct basilical and centralized designs, we should perhaps look for other [End Page 104] explanations, such as the patron's personal taste and available wealth, and the symbolism associated with these architectural forms.156
The dome that today crowns Sts. Sergius and Bacchus, when viewed from the interior, begins immediately above the crowns of the eight arches that connect the eight piers surrounding the nave. It is divided into sixteen sides, which are alternately flat and concave (figs. 1, 7). The flat sides each contain a window, and the concave sides are each located over one of eight piers, whose reentrant angles define the octagonal core of the building. The transition from the re-entrant angles to the concave segments of the dome above is abrupt, and is marked by small ledges.157 The concave segments become shallower and narrower as they rise towards the summit. Ebersolt and Thiers provide sectional elevations that show the dome curving smoothly about a single center.158 However, one need only look up into the dome (fig. 1) to realize that this is not the case, and more recent surveys clearly show the reality (fig. 7).159 In vertical cross-section, the dome's curvature is irregular, rising almost vertically to the tops of the eight windows that pierce it, but then turning tightly inwards before following a very shallow curve to the crown.160
On the exterior, the dome's lower, more vertical part, up to a certain height above the windows, is thickened with masonry so as to create sixteen vertical faces, so reflecting the sixteen sides of the interior of the dome (fig. 8). This [End Page 105]
[End Page 106]
gives the false impression that a very shallow dome springs from the top of a low sixteen-sided drum. What appears to be a drum is, however, strengthening for the lower part of the dome to guard against meridional cracking. The sixteen exterior walls of the "drum" alternate noticeably between short and long (fig. 9).161 The shorter sides are pierced by the windows; the longer sides correspond to the concave segments of the dome, and have substantial buttresses attached to them.162
Architectural historians have always assumed the dome to be the original one that has crowned the building since the reign of Justinian.163 Certainly, [End Page 107] gored domes had long been used in Roman architecture, the best-known examples being at Hadrian's villa near Tivoli: the concrete vault over the vestibule of the Piazza d'Oro, and that over the dining hall (Serapaeum) at the end of the Canopus.164 The plan of the vestibule of the Piazza d'Oro, being octagonal with niches on four alternate sides, approximates to the plan of the inner core of Sts. Sergius and Bacchus.165 Its dome, however, is constructed very differently from that of Justinian's church. In its complete state, it would have had eight gored sections, one corresponding to each side of the octagon, whereas, in the dome of Sts. Sergius and Bacchus, eight gored sections rise above the angles of the octagon and alternate with eight flat elements. The dining hall of the Canopus is semicircular in plan, and in its semidome there are flat elements between the gored sections.166 In neither of the Hadrianic buildings, in contrast to the situation in Sts. Sergius and Bacchus, are the gored sections fashioned in such a way that they continue down to the level of the springing of the vault, so as to recede behind the vertical face of the supporting masonry; rather, they terminate in semicircular lunettes that are flush with the internal walls. But the most obvious difference between such Roman domes and the dome of Sts. Sergius and Bacchus is that they invariably possess a high, smoothly curving, almost semicircular profile, not a kinked curvature. If, however, we turn to later Byzantine and Islamic architecture, it is not difficult to find parallels for gored domes both with an irregular profile, and in which the goring extends downwards as far as the springing of the vault. I mention for instance the ninth-century dome over the mihrab of the Great Mosque at Kairouan in Tunisia;167 the esonarthex, nave, and prothesis domes of Christ in the Chora in Istanbul (Kariye Camii, 1316–1321);168 and the dome in the porch before the main entrance to the Selimiye Camii in Edirne (1569–1575) (fig. 10). Given its irregular profile, I suggest we must seriously consider the possibility that the dome of Sts. Sergius and Bacchus is not original but a later Byzantine or Ottoman addition.
A second piece of evidence that may indicate structural alteration is the observation, already made, that the sixteen sides on the exterior of the dome curiously alternate between long and short, the longer sides measuring about 4–5 m in length, and the shorter being about 2.9–3.4 m long. Finally, it may be noted that the bricks in the lowest part of the dome are reportedly irregular in form and dimensions, measuring approximately 40–45 × 20–25–35 × [End Page 108]
Evidence that appears to support my suspicions that the dome is not original emerged when the brickwork of the lower part of the dome's exterior was briefly revealed to public view in June 2004, after the plaster had been stripped away in preparation for making structural repairs (fig. 11). Photographs show that the curved head of each of the windows in the dome was formed by an arch of small, radially laid bricks. Above each window arch were the remains [End Page 109] of a second brick arch with a greater radius and a somewhat higher center, which was built of longer, radially laid bricks. The external diameter of each of these larger arches was almost exactly the width of the side of the sixteen-sided "drum" in which it was located, but the crown of each arch was lost, truncated by the dog-tooth decoration of the eaves below the lead covering of the dome. These details could be seen most clearly on the eastern side of the "drum" (fig. 12), but were also evident on other sides: northwest, west, southwest, and south (figs. 13–15). The truncation of the large arches clearly indicates that the masonry once rose to a higher level and consequently that the curvature of the dome began, at least on the exterior, at a significantly higher level than it now does.
Without the thorough structural survey that the building has long deserved,171 it is impossible to make firm deductions from this evidence, so the following discussion is necessarily speculative. But clearly we must consider the possibility that, at an earlier period in the church's history, there existed a [End Page 110] true drum. If so, it is possible that the earlier dome did not spring from such a low level as the present one, but was seated on that drum.
One possibility is that the truncated brick arches located over the present window-arches were relieving arches set into the brickwork over the small windows. However, we must consider the likely alternative possibility that the small windows we see today are not original, and that the truncated arches in fact formed the frames of eight large windows in the original drum. If these arches were indeed window arches rather than relieving arches, it would be necessary to suggest that they belonged to large window openings (about 2.5 m wide and about 3.5 m high) that were later bricked up, so as to create the smaller windows (about 1.2 m wide and 2.5 m high) that we see today. This is entirely plausible, since many of the windows in the church have been reduced in size. The three tall windows in the apse, for instance, were reduced in the Turkish period by raising the sill with masonry, and by lining the sides of the window with short bricks, so leaving a small window with a pointed arch of short bricks.172 It is not necessarily the case, however, that the proposed reduction of the windows in the drum took place at the same time as the reduction of those in the apse.
If the truncated brick arches are not relieving arches but the remains of wide windows that pierced the original drum, then an explanation is required for the fact that these arches have an external diameter corresponding to the width of the side of the polygon in which they are set. For structural strength, we would expect such wide windows to have been set into walls that were significantly wider. The logical deduction is that, on the exterior, the original drum was not sixteen-sided but octagonal, having equal sides about 7–8 m wide, each containing a large window about 2.5 m wide (fig. 16).
Assuming this was the case, the buttresses we see today, and the eight sides of the sixteen-sided polygon to which those buttresses are attached were not part of the original design. I suggest that these eight sides were added after demolishing the eight angles of the original octagonal drum from the side of one window to the side of the adjacent window.
The extensive redesign and rebuilding of the octagonal drum helps to explain why the sixteen sides visible in the upper part of the present "drum" are not equal in length. The fact that the springings of the original window arches have survived the demolition of the angles of the drum, may suggest that the windows had already been reduced to their present size before this demolition took place: their filling of brickwork would have provided support for the springings of the original window arches, ensuring their preservation when the angles were demolished. [End Page 111]
[End Page 112]
[End Page 113]
Although we can assess the approximate height of the original windows (and hence the height of the drum) from the architectural evidence, the precise form of the original dome and roof remains conjectural. The original design is likely to have been very similar to the drum and dome of the church of San Vitale in Ravenna, which was founded in the episcopate of Ecclesius (521/2–531/2), and may have still been under construction under Victor (537/8–544/5), since the dedication did not take place till 547.173 The similarities between the plans of San Vitale and Sts. Sergius and Bacchus (both of which have an octagonal core incorporating columnar exedras) have often been pointed out by architectural historians.174 In the light of the evidence [End Page 114]
[End Page 115] presented here, it now seems likely that the elevation of the two buildings was much more similar than has previously been thought. In Sts. Sergius and Bacchus, it is uncertain how the transition was made from the octagonal drum to the circular base of the dome, but, as in San Vitale (fig. 17), there may have been squinches over the eight corners, thus converting the interior of the eight-sided drum into a sixteen-sided figure that approximated more closely to a circle (fig. 18).175 In fact, it seems likely that, at the springing of the present dome (fig. 9), the concavity of each gored segment preserves to some extent the concave plan of a squinch. The semidomes that originally topped the squinches would have been destroyed in the process of lowering the profile of the dome. The external angles of the octagonal drum would not have easily accommodated buttresses, which were probably absent, as at San Vitale. The dome itself, like that of San Vitale, would have curved about a single center, but may have been less than hemispherical, even if it sprang from a point a little below the crowns of the window arches rather than from a cornice above them. Whether the dome was roofed in lead, as was the dome of Hagia Sophia, or whether it was shielded by a pyramidal wooden roof, as was the dome of San Vitale, is also uncertain. Both because of the height of the drum and the curvature of the dome, the nave of the original church of Sts. Sergius and Bacchus would have been considerably higher than it is today; it would also have been illuminated with much larger windows than at present.
The original dome of Sts. Sergius and Bacchus was presumably, like that of Hagia Sophia, built of solid brick. If it indeed sat upon a high, octagonal drum, the dome may not have been adequately buttressed at its base to contain tensile stresses. Meridional cracks may have developed, allowing the diameter of the dome to increase, so causing outward thrusts to act on the drum. Structural problems in the drum seem to be suggested by the decision to reduce the size of the windows. It is therefore possible that weaknesses in both the lower part of the dome and the upper part of the drum caused the original dome to fail. We can only speculate about what might have caused the presumed collapse: structural flaws, damage in an earthquake, the ravages of time, or a combination of these factors. It may be suggested that the drum of San Vitale fared much better because the dome was built not of solid brick but of less dense terracotta tubes, with the result that its stresses were more easily contained.176 [End Page 116]
I suggest that the architects responsible for the postulated rebuilding of the dome of Sts. Sergius and Bacchus determined that the new design should better contain the tensile stresses at its base. Their solution was to dispense with the original drum by setting the springing of the new dome (the dome we see today) at a lower level on the interior of the church—that is, at a level just below the sills of the windows. The original drum was no longer to serve as the seating for the dome; instead, the lower parts of the drum would be incorporated into the external structure of the new dome to serve as buttressing to prevent meridional cracks developing in the lower parts of the dome. The architects cut the height of the drum down to a level just above the springing of the original window arches. Much more of the original drum was then destroyed by cutting off the eight external angles between the windows, in order to create a sixteen-sided form that would surround the dome more tightly and permit the addition of buttresses. This process is likely to have resulted in the destruction of the semidomes over the squinches, if indeed squinches were originally used to bridge the interior angles. It may also have resulted in the destruction of one or two cornices: one crowning the original drum and forming the springing for the original dome; another possibly running above the crowns of the main arches and below the squinches.
Only the very lowest parts of the eight external angles of the original drum were, it seems, retained, being incorporated into the buttresses that were built against the eight new sides. The preservation of these original angles is reflected, in the interior of the church, by the form of today's dome. The eight segments of the dome that are located over the re-entrant angles of the piers are concave. The creation of these concave (or gored) segments was presumably encouraged by the presence of the angles of the original octagonal drum, if not by the presence of concave squinches. Buttresses added to the exterior of the lower part of the dome behind these gored segments served to contain the thrusts of the flattened dome. It was possible to introduce the buttresses because they could be supported from below by the eight main piers of the nave. The fact that flat, rather than concave, segments were used to form the sides of the dome in which the windows are located seems to be the result of a decision by the architect to preserve the central part of the eight straight walls of the original octagonal drum.
Thus it would seem that the alternation of flat and gored dome segments was not merely a product of aesthetic preference but rather suggested or dictated by the form of the original structure. It would also seem that the kink in the profile of the dome just above the tops of the present windows reflects the fact that that was the level at which the original drum was truncated.
Exactly when this substantial alteration of the structure was undertaken is uncertain. We know that the church was repaired, at the prompting of the Patriarch Ignatius (847–858; 867–877), during the reign of Basil I (867–886), [End Page 117] possibly after the earthquake of 869: "Basil decorated this church with sacred icons and made good other deficiencies in the structure."177 However, the extent of Basil's work is unclear and, in the absence of more detailed structural observations, it is unfortunately not possible to eliminate either an earlier or a later date for the rebuilding of the dome. The double row of dog-tooth decoration in brick around the eaves of the building must certainly be post-Justinianic in at least two cases: where it crowns the truncated drum, and where it decorates the top of the buttresses. But, while dog-tooth decoration is more common from the late ninth century, it may be seen crowning the eaves of fifth- and sixth-century structures in Ravenna and therefore provides no clear dating criterion. It may be added, however, that if a longitudinal section drawing of the church, which dates to around 1830 and is preserved in the archive of Charles Texier in the Royal Institute of British Architects, is accurate in showing the dome decorated with figures of angels, the dome cannot have been built after the church's conversion into a mosque at the beginning of the sixteenth century. There are, however, several inaccuracies in Texier's drawing, and these should cause us to hesitate before accepting its evidence.178 The fact that it has been taken for granted by generations of architectural historians that the present dome is contemporary with the rest of the Justinianic structure suggests that the architects who designed the new vault succeeded not only in making it more durable but also in integrating it harmoniously with the existing building.
I find it difficult to see any support for the claim, often made, that the dome of Sts. Sergius and Bacchus constitutes a link in a chain of architectural development leading towards the dome of Justinian's Hagia Sophia.179 Not only was Sts. Sergius and Bacchus a contemporary, rather than a predecessor, of Hagia Sophia, but also the designs of the two buildings were quite dissimilar. The original dome of Sts. Sergius and Bacchus rested on a high octagonal drum, the transition from octagon to circle probably being achieved using a system of squinches. The original dome of Hagia Sophia, like that of Sts. Sergius and Bacchus, no longer survives, and, although there is agreement that the dome probably had the same radius of curvature as the pendentives, there is debate [End Page 118] as to whether there was just a very low drum-like wall to elevate the dome slightly above the dome cornice or whether there was a drum high enough to incorporate windows.180 But even if we imagine that its windows were in a drum rather than in the lower part of the dome itself, that drum would have been cylindrical and low, not octagonal or of the same high proportions as that in Sts. Sergius and Bacchus. Furthermore, the dome of Hagia Sophia is located over a square bay, not an octagonal one, and the transition from that square to the circle at the base of the dome was not achieved using squinches; rather, four pendentives were employed.
A common argument in favor of the contention that Sts. Sergius and Bacchus is likely to have served as an architectural experiment in preparation for Hagia Sophia is that there are "obvious improvisations in its execution."181 Because the present dome is not original, its distortions are not relevant to this discussion; however, the irregularities in the plan of the church are certainly the responsibility of the Justinianic builders. These irregularities have led to the claim that "while the basic design was conceived by a very gifted architect, its execution was entrusted to indifferent builders who, after marking out a lopsided plan on the ground had to improvise as they went along."182 The main justification for this claim is that "there was no reason why the internal octagonal shell should have been placed noticeably askew with regard to the outer square shell."183 On the contrary, there was probably a very good reason for the misalignment: the octagonal core, it seems, was deliberately turned so that its axis passed through the doors in the north and south walls of the church, doors whose positions were dictated by existing adjoining buildings and were therefore not directly opposite each other.
I will explain this point briefly. Towards the center of the north wall of the church, the exterior brickwork reveals the former existence of two triple arches, one at ground and one at gallery level, both now blocked. The capitals of the triple arch at ground-floor level (with monogram nos. 17 and 18) were exposed on the interior during the recent restoration. These triple arches indicate that there once existed connections, at both levels, with adjacent buildings to the north of the church.184
The present south wall of the church has been heavily restored, and it is not possible to discern the distinctive sixth-century construction technique that was used in much of the rest of the building, by which I mean pure brickwork [End Page 119] with single courses of large greenstone blocks after every ten to fifteen courses.185 In the wall, there are traces of what were once five tall, narrow, buttresslike projections (fig. 2).186 These were built using alternating bands of five brick courses and five stone courses, a style that is broadly datable between the seventh and tenth centuries. No earlier work is obvious. It appears that the four spaces between the five projections were originally arched over. In the lower story, three wide brick arches are preserved between the four easterly projections, and there was presumably also a narrower arch to the west. There may originally have been four similar arches in the upper story, and, if they were of the same radius as those in the lower story, it would appear that the height of the south wall is now a little less than it once was. The presence of arches suggests that there was access from the church into another building standing on its southern flank. Later, the arches were blocked. This work may have occurred after the disappearance of the postulated building to the south of the church. However, it may have been undertaken over a period of time, judging by the irregular sizes, types, and positions of the windows that survive in the blocking walls. The result of this blocking was to create the continuous south wall we see today. Because the masonry of the blocking was not as thick as that of the projections between the arches, the effect was to create three shallow bays, each about 4.3 m wide, inside the church at both ground-floor and gallery level.187
The central shallow bay at both ground-floor and gallery level may be entered from the north through a triple arch. The columns of the triple arch [End Page 120] at ground-floor level display fixings for a door-frame,188 and this is again suggestive of the presence of another building to the south of the church. It is important to observe that the triple arches at ground-floor and gallery level on the south side of the church are located further east than the corresponding triple arches in the north wall. It is for this reason that the octagonal core of the church was set askew in relation to its external walls. The north–south axis of the octagonal core was aligned so as to create a line of sight from the doors in the triple arch on one side of the church, through the columns of the octagonal core, to the doors in the triple arch on the other side. The positions of both the apse (which is set slightly to the north of the center of the east wall) and the central entrance from the narthex (which is set slightly to the south of the center of the west wall), are therefore to be explained by the fact that they were located on a second axis aligned perpendicularly to the first. Thus the irregularities in the plan of the building can be explained as deliberate. Furthermore, although the octagonal core is misaligned with the external walls, measurement of its plan reveals that it is carefully laid out.189
Sts. Sergius and Bacchus can no longer be seen as a flawed experiment in preparation for the construction of Hagia Sophia: the two churches were contemporary and the irregularities in the former were not demonstrably due to deficiencies in the ability of the architects. Rather than attempting to detect a chronological increase in architectural sophistication from one church to the other, we should, I think, see both churches as products of a period of experimentation and innovation in church architecture. We may imagine such experimentation as a continuous process among architects, occasionally resulting in highly significant developments, such as Hagia Sophia. At least a decade before Hagia Sophia and Sts. Sergius and Bacchus were built, Anicia Juliana commissioned her church of St. Polyeuktos. Largely built between 519 and 522, it was roughly contemporary with Justinian's church of Sts. Peter and Paul, which was built on a traditional basilical plan. St. Polyeuktos had a highly original design, incorporating six columnar exedras that were not, as might have been expected, arranged in a ring to form a central core that could support a dome, but in a linear fashion to support the timber roof of a basilica.190 Despite its innovative architecture, St. Polyeuktos was not crowned with a brick dome that might have served as a model for Hagia Sophia's. For evidence of earlier developments that may have led to the achievement at [End Page 121] Hagia Sophia, we must look to pendentive domes (domical vaults) over square spaces, such as that crowning the so-called mausoleum of Galla Placidia in Ravenna;191 and we must look to domed rotundas and domed polygons, both in Constantinople (where the foundations of several survive) and elsewhere.192 That, however, is another story.
I dedicate this paper to Cyril Mango, whose contribution to the study of Byzantine Constantinople is both admirable and inspirational. It is a pleasure to thank Mary Whitby for her improvements to the published translations cited in this article, for allowing me access to her translation of Paul the Silentiary, and for much productive discussion. I am also very grateful to Federico Montinaro for kindly sharing relevant pages of his doctoral thesis, which contains a valuable new edition and translation of both the long and short recension of Procopius's Buildings. Thanks are also due to Gianfranco Agosti for generously allowing me to mention his as yet unpublished thoughts on the inscription from Zorava.
2. Descriptions of the architecture of the building are numerous and include: Henderson 1906; van Millingen 1912, 70–83; Ebersolt and Thiers 1913, 19–51; Sanpaolesi 1961; Mathews 1971, 42–51; Mango 1976, 101–107 (= 1986, 58–59); Krautheimer 1986, 222–25.
5. Mango 1972, 189–93 (giving the dates 527–536 on p. 190, but expressing an inclination towards the later limit on p. 192); Mango 1975, 385–92 (giving the limits 527–536 on p. 385, and the more specific period 531–536 on p. 392).
15. Proc. Aed. 1.4.1–4, 7–8. The text of the long recension quoted is that of Haury 1964. So far as I have quoted Haury's text here, Montinaro's text does not differ, other than in drawing attention to the words (to be discussed) that Downey 1948 excludes. The translation is that of Dewing and Downey 1940, adjusted in the light of improvements suggested by Mary Whitby. The words πρῶτα μὲν at Aed. 1.4.1 are balanced by μετὰ δὲ καὶ at Aed. 1.4.9, where the account of Justinian's second church of the Apostles (the Holy Apostles) begins. Mathews 1971, 74 n. 9, correctly observes that ἐκ πλαγίας (Aed. 1.4.4) means "alongside," which is supported by Procopius's later assertion (Aed.1.4.7) that the two churches shared a single narthex.
17. Downey 1948, 44–45. The problem with the text was discussed by van Millingen, who did not, however, suggest any amendment, observing only that Procopius "has not expressed his meaning as clearly as he might have done." See van Millingen 1912, 64–66. Frolow 1949 unconvincingly advocated altering the text to read καὶ ἕπεται καὶ τὸ τέμενος τὸ ἄλλο ἐκ πλαγίου τούτῳ παρακείμενον or καὶ ἕπεται καὶ ὁ μὲν νεὼς ἄλλος. …
19. I mention just two possibilities here. It might be suggested, for instance, that the error goes back to Procopius himself, although such carelessness on his part seems unlikely. Another possibility is suggested by the fact that the margins of text V (Vat. gr. 1065), which best represents the long recension, contain variant readings characteristic of the short text. This demonstrates that the scribe who created V in the twelfth or thirteenth century either had copies of both the long and short texts before his eyes, or had a copy of the long text containing marginal notes derived from the short text (Flusin 2000, 15). Here, therefore, we have a clear example of the two recensions being compared, and it would hardly be surprising if, at some point in the history of the texts' transmission, elements of the short version came to be inserted into the body of the long, either accidentally or deliberately.
24. Mathews 1974, 24 wrote: "La construction de Saints-Serge-et-Bacchus devrait donc être considérée comme faisant partie de ce programme destiné à incorporer l'Hormisdas dans le Grand Palais," an observation that rightly links the two activities, although the connection between the two palaces was established before (if only shortly before) the construction of the church. The extent of the Palace of Hormisdas is not known. As Mango observes, "It is not certain … that the harbour of Hormisdas was identical or co-extensive with the later harbour of the Boukoleon" (Mango 1997, 47). Bolognesi's suggestion that a heptaconch triclinium in the Palace of Hormisdas that was used for meetings of the council in 532 was a precursor of the octagonal Chrysotriklinos of Justin II, which is known to have been located near the Boukoleon harbor, is speculative. See: ACOec. IV.2, 169 (ed. Schwartz and Straub 1914–1983); Bolognesi 2000, 236–37; Bardill 2006b, 6–7, where it is explained that the connection of the Palace of Hormisdas to the Great Palace can be seen in the context of the gradual expansion of the Great Palace in a southerly direction, in particular under Justin II.
25. Proc. Aed. 1.10.2–4.
28. A fire on 15 January 532 affected the area around the House of Probus near the harbor of Julian, but had damage been inflicted on the Palace of Hormisdas we might expect it to have been mentioned explicitly. On the fires during the riots, see Greatrex 1997, 84–86.
30. It may be noted, for what it is worth, that this terminus post quem agrees with a somewhat confused assertion in the often unreliable Patria to the effect that Justinian built Sts. Sergius and Bacchus as his residence after (and perhaps in commemoration of) his suppression of the Nika Riots of 532; until then, the text suggests, Justinian's dwelling had been the church of the Apostles (Sts. Peter and Paul). See: Patria 3.39 (ed. Preger 1907, trans. Berger 1988, trans. Berger 2013); Dagron 1984, 320–21.
33. At Sts. Sergius and Bacchus, twelve marks of four diff erent types were recorded by Deichmann 1974–1976, pt. 2, 217, table 4 (some illustrated on p. 216 fig. 50). Fig. 3 shows marks I have seen myself.
34. Pending the publication of the complete record of the masons' marks in Hagia Sophia preserved in the R. L. van Nice collection in the Dumbarton Oaks Fieldwork Archives, see: Antoniades 1907–1909, 1: 97–103 (= Deichmann 1974–1976, pt. 2, 217, table 4a); Butler, 1989, 136–66; Paribeni 2004, 653–734. See Deichmann 1974–1976, pt. 2, 217–18 (with figs. 50–51), 226, for the correspondences between the marks at Sts. Sergius and Bacchus and Hagia Sophia. The occurrence of the mark ΖѠ at both Sts. Sergius and Bacchus and Hagia Sophia was first noted by Choisy 1876, 248.
36. It is to be hoped that details of any stamped bricks found during the restoration of the structure will one day be published. Details of the restoration on the Byzantine period database of the TAY Project website (http://tayproject.org/Bizanssearcheng.html) include the statement: "Many potsherds and fragments of sealed bricks and architectural elements were revealed among the floor debris during the restorations initiated in 2003" (last accessed 12 December 2016).
37. On the epithet "scepter-bearing" see the discussion of Whitby 2003, 598, 601–602. Mary Whitby further informs me that the term was applied to the emperor as early as the reign of Marcus Aurelius: Opp. Hal. 2.41, 3.1 (ed. Mair 1928).
45. The damage in the lower story was noted by van Millingen 1912, 73: "Most of the monograms have been effaced." He also suggests that a partial monogram of Justinian remained "on the last capital in the north-eastern bay" (i.e. monogram no. 14). In the upper story, he reported (p. 75) only one monogram missing, whereas I note two. The location given for the monogram he claimed to be missing—"the eighth (counting from the north-east)"—is incorrect: the missing monogram is the seventh counting from the north-east (no. 26). Van Millingen's statement suggests that monogram no. 32 was still intact in his day.
47. Suggested equations between the monograms illustrated by Swainson and those illustrated here are as follows. No. 10 possibly = Swainson no. 20. No. 13 possibly = Swainson no. 21. No. 15 possibly = Swainson no. 5 or 6. No. 16 = Swainson no. 18. No. 19 possibly = Swainson no.13. No. 20 = Swainson no. 3. No. 21 = Swainson no. 16. No. 23 possibly = Swainson no. 2. No. 24 = Swainson no. 14. No. 28 = Swainson no. 17. No. 29 = Swainson no. 11. No. 31 possibly = Swainson no. 12.
48. This monogram is not to be resolved "Of Justinian" because of the absence of a clear diagonal rising to the top of one or other of the uprights, as would be required to create a letter Ν, and because a Γ is incorporated into the left-hand upright. The most similar monogram that means "Of Justinian" is no. 5, but by contrast it has a clear diagonal rising to the top of the right-hand upright, it lacks a gamma (other than that which might arguably be read into the square sigma attached to the right-hand upright), and it clearly includes a tau, created by a horizontal bar atop the right-hand upright.
49. The main letters in the monogram are Α, Ο, Υ, and a square sigma, which may be understood as incorporating a Γ. These letters are at the ends of the arms of a cross, which may be understood as incorporating a Τ. Van Millingen 1912, 73, apparently read this monogram as "Of Theodora," since he wrote: "the name of the empress still appears on the capital of the western column in the south bay." But this reading does not explain the presence of the Υ. Bayülgen 2012, 34 n. 28, follows van Millingen's interpretation. The most similar monogram reading (apparently) "Of Theo-dora" from Hagia Sophia, is shown in Antoniades 1907–1908, 2: 356 no. 538, and it has the letters Ο, Ρ, Ε, and Α attached to the arms of a cross.
50. The monograms of Hagia Sophia have been recorded by Antoniades but deserve to be systematically re-cataloged. Box monograms reading "Of the Augousta" at Hagia Sophia appear in Antoniades 1907–1909, 2: 31 nos. 217, 218; 209 no. 288; 231 no. 309; 297 no. 372; 326 nos. 446, 447, 449; 335 nos. 465, 470, 473, 474; 343 nos. 500, 501. That approximating most closely to the example from the gallery of Sts. Sergius and Bacchus (here fig. 4, no. 30) is on p. 231 no. 309. Box monograms reading "Of the Augousta" at Hagia Eirene appear in George 1912, 21 fig. 7, nos. 4 and 6 (see also pl. 16).
57. Croke 2006, 36–40. Menze 2008, 154, reaches the same conclusion, and suggests (pp. 217–18) that Theodora set up her refugee camp in the Palace of Hormisdas around 535/6. On Theodora as a non-Chalcedonian empress alongside Justinian as a Chalcedonian emperor, see Menze 2008, 208–28.
59. Krautheimer 1974, 252, insisted that the refugee situation, "would not have arisen before the Council of 536 which condemned the monophysites and presumably not before early 537 when persecution in the provinces started in earnest." However, Mango 1975, 386, was less categorical, and dated the community's foundation to "535 (or earlier)–548," even allowing a date as early as 531. Despite Croke's arguments, Potter 2015, 250 n. 16, is also prepared to believe that the community began to gather before 536, on the grounds that the Patriarch Anthimus chose to seek refuge in the Palace of Hormisdas in that year.
70. For Justinian's letters, see Vasiliev 1950, 179–83, 199. The connection between the construction of Sts. Peter and Paul and the reconciliation with Rome was noted by: Vasiliev 1950, 378; Dynes 1964, 6; Dvornik 1966, 828–31.
73. Croke 2006, 38 with n. 72, deduces that in 536 Anthimus was sought in the Palace of Hormisdas simply because of his earlier connection with that location. He also suggests dating that earlier connection to a period in the late 520s when there "was evidently no question about his orthodoxy." By contrast, Potter 2015, 250 n. 16, believes that in 536 the search for Anthimus was directed towards the Palace of Hormisdas because a community of non-Chalcedonian refugees was already installed there.
81. The poem is Anth. Gr. 1.10.1–41.
85. Mercati (1925) 1970. Mercati discusses the manuscript tradition of this poem, on which note also Feissel 2000, 89. For metrical observations, see Whitby 2006, 183–84. Previous archaeological records of the inscription include: Dethier 1858; Henderson 1906, 7 with fig. 5; Antoniades 1907–1909, 1: 12; 3: 19 fig. 559 (letter forms); van Millingen 1912, 74 fig. 20; Ebersolt and Thiers 1913, 24.
87. The superscript dots may be seen on surviving blocks 1 a ii and 1 a v, marking the ends of verses 15 and 30, respectively. See Harrison 1986, 118 fig. A, pl. 91 (1 a ii); 120 fig. B, pl. 99 (1 a v).
90. The opposition of the names is also noted by Eastmond 2016, 225, who comments: "It is likely that this division corresponds to the gendered spaces in the church, where the two rulers stood during the liturgy."
92. Compare, e.g., Hom. Il. 2.1–2; 2.211–12; Eus., HE 5. pr. 4 (ed. Schwartz, Mommsen, and Winkelmann 1999); an inscription in the mid-tenth-century Leo Bible (Cod. Vat. Reg. gr. 1), fo. 2v (Mango 2011, 65). Further references in Rhoby 2014, 815 n. 15.
96. Thus Plin. Pan. 49, observed, "Your sleep is short and sparing, for in your love of us there is no period of your time so short as that you spend alone."
99. Dio Chrys. Or. 1.13. Dio alludes to Hom. Il. 2.24–25, concerning Agamemnon. The term "shepherd of the people" is applied by Homer to several kings and chieftains, including Agamemnon.
100. Aesch. Sept. line 3. Stat. Theb. 8.269–70.
108. Wisdom 14:15–16.
112. Wisdom 3:11.
114. Wisdom 3:1; 3:4; 3:6.
115. Paul. Sil. Amb. 37–40, ed. De Stefani 2011, trans. Mary Whitby.
116. I thank Gianfranco Agosti for allowing me to mention his observations on the Zorava inscription before their publication. I quote the text from Mondésert 1960, 125–30: ☩ καὶ νῦν σωτῆρος δεσπότου θεοῦ δύναμιν ὁρῶν | δόξασον ἄνακτ' ἅγιον, ὃς εἰδώλων ὤλησεν ἔργα· | οὗτος γὰρ δόμος τὸ πρὶν γλυπτῶν δαιμόνων ἐτέτυκτο | ἀχρίστοις λάεσι vac. δεδμημένος, οὓς λόγος Χριστοῦ | λῦσεν, ἠδ' ἀνήγειρεν εὐξέστοισι λάεσι | δόμον ἑοῦ θεράποντος εὐίππεός τε Σεργίου, | σπουδῇ καὶ ἔργοισι παίδων ἐσθλοῦ Θεοδώρου, | Σέργιν αὐτὸν ἅγιον ἔχειν ἀρωγὸν θελήσαντες, | ὃς χθόνιον κράτος ἀνῄνετο ἠδὲ πικρούς τε | βασσάνους ἐδέξατο κεφαλῆς ἄπο μέχρι ποδῶν τε· | πόδας γὰρ ἡλωθεὶς κεφαλῆς οὐκ ἐφίσατ' ὁ κλῖνος, | ἀλλ' θανάτῳ προὔδωκεν ψυχὴν ἑῷ δεσπότῃ δώσας | σωτῆρι ἠδ' ἀντὶ χθονίας οὐρανίαν ἔλαχεν ζω ̣|ήν. For an English translation, see Key Fowden 1999, 110.
117. The text is preserved in various manuscripts and in different versions: see Mercati (1921) 1970. I quote the edition of Asdracha 2003, no. 102: Ἄλλοι μὲν ἄλλῃ τῶν πάλαι βασιλέων | αὑτοῖς προαφώρισαν εἰς ταφὴν τόπους, | ἐγὼ δὲ Βασίλειος, πορφύρας γόνος, | ἵστημι τύμβον ἐν τόπῳ γῆς Ἑβδόμου | καὶ σαββατίζω τῶν ἀμετρήτων πόνων | οὓς ἐν μάχαις ἔστεργον, οὓς ἐκαρτέρουν· | οὐ γάρ τις εἶδεν ἠρεμοῦν ἐμὸν δόρυ, | ἀφ' οὗ βασιλεὺς οὐρανῶν κέκληκέ με | αὐτοκράτορα γῆς, μέγαν βασιλέα· | ἀλλ' ἀγρυπνῶν ἅπαντα τὸν ζωῆς χρόνον | Ῥώμης τὰ τέκνα τῆς Νέας ἐρυόμην | ὁτὲ στρατεύων ἀνδρικῶς πρὸς ἑσπέραν, | ὁτὲ πρὸς αὐτοὺς τοὺς ὅρους τοὺς τῆς ἕω, | ἱστῶν τρόπαια πανταχοῦ γῆς μυρία· | καὶ μαρτυροῦσι τοῦτο Πέρσαι καὶ Σκύθαι, | σὺν οἷς Ἀβασγός, Ἰσμαήλ, Ἄραψ, Ἴβηρ· | καὶ νῦν ὁρῶν, ἄνθρωπε, τόνδε τὸν τάφον | εὐχαῖς ἀμείβου τὰς ἐμὰς στρατηγίας. Lauxter-mann 2003, 236–38, reprints the edition of Mercati (1921) 1970, provides an English translation, and makes useful comments on the text. For another English translation, see Stephenson 2005, 230–31.
120. For the text, see Rhoby 2014, 813–17, where various parallels are noted. I quote the relevant part: Ἄλλους μὲν ἄλλοι τῶν πάλαι βασιλέων | σεβασμίους ἤγειραν ἁγίοις τόπους· | ἐγὼ Ῥογέριος δὲ ῥὴξ σκηπτροκράτωρ | τῷ τῶν μαθητῶν προκρίτῳ τοῦ Δεσπότου | τῷ ποιμενάρχῃ καὶ κορυφαίῳ Πέτρῳ | ᾧ Χριστὸς ἐστήριξε τὴν ἐκκλησίαν | ἣν αὐτὸς ἔσχεν αἵματος χύσει ξένῃ | πάντ᾽ αὐτὸς [. … Translation adapted from Demus 1949, 60 n. 11 and Borsook 1990, 17.
123. Borsook 1990, 43 n. 4. It is uncertain why the inscription in Justinian's church of Sts. Peter and Paul (Anth. Gr. 1.8) was not chosen for imitation instead of that from Sts. Sergius and Bacchus; perhaps it was not known, or perhaps it was not considered worthy of imitation.
128. Mercati (1925) 1970 explains that the inscription may also be read on fo. 117 v. of Pal. gr. 141, a manuscript of the fourteenth or fifteenth century containing letters and epigrams of Maximus Planudes and verses of Manuel Philes. Having read it there, a sixteenth-century scholar added it into the margin of a first edition of the Planudean Anthology (Florence, 1494) in the monastery of Patmos.
130. The debate concerning the existence of such a tradition is to be found in Mango 1972; 1975; Krautheimer 1974, 251–53; Mathews 1974, 22–29. In the face of Mango's criticisms (1975, 390–91), Mathews 2006, 141, appears to have abandoned his claims that St. Stephen in Daphne was an example of a centralized palace church; he clings, however, to St. John in the Hebdomon, despite the fact that Mango had observed that it was outside the Iucundianae palace.
134. Coll. Avell. 187.5 (CSEL 35.2: 645).
137. In Justinian's letter to the pope of 29 June 519, Sts. Peter and Paul is described as basilicam… constructam; and in the letter of the papal legate of the same date, it is said that Justinian basilicam … constituit. This ought to mean that at least the shell of the edifice was finished. See Coll. Avell. 187.5, 218.1 (CSEL 35.2: 645, 679); van Millingen 1912, 64–65; Mango 1972, 189; Vasiliev 1950, 377–78.
138. Proc. Aed. 1.4.1.
139. Coll. Avell. 187.5 (CSEL 35.2: 645).
140. Cedr. pp. 642–43 (ed. Bekker 1838–1839) states: "And [he built] the church of Sts. Sergius and Bacchus, next to the palace towards the sea, and hard upon it the church of the Holy Apostles [Sts. Peter and Paul], which was previously his own private house, and dedicating all his former property to these two churches he endowed a monastery of glorious men." The account is certainly inaccurate in giving the impression that the two churches were built simultaneously or that Sts. Sergius and Bacchus was built first. It therefore cannot reliably be deduced that the two churches became a monastery at the same time, as might also be inferred. The passage would, however, support the view that Sts. Peter and Paul had initially been a private church. This account, as Mango (1972, 192 n. 11) observes, "is part of a general characterization of Justinian's activity, and has no chronological value."
141. One might imagine that Justinian's connection provided a route that passed from the Great Palace, south of the curved end of the Hippodrome, to the Palace of Hormisdas. Presumably this connection was used by the future emperor Tiberius II, who, when still Caesar, is said to have gone each evening from the Great Palace to the Palace of Hormisdas in order to visit his wife, who felt herself to be in constant danger there (Joh. Eph. Hist. eccl. 3.7–8, ed. and trans. Brooks 1952). However, for the imperial visit on Easter Tuesday (Const. Porph. de cer. 1.11, ed. Reiske 1829, 86–89, trans. Moffatt and Tall 2012), the emperor seems to have used a more circuitous route. He left the Chrysotriklinos (located in the lower palace) and passed through the Lausiakos, Justinianos, Skyla Gate, and Covered Hippodrome. He then entered the Hippodrome (crossing it, I presume, from the Karea Gate under the Kathisma to a gate on the southwestern side [the Nekra?]) and passed through the old bureaux (asēkrēteia) to reach the church. The part of the route that lay within the palace may be appreciated by consulting the diagram of the Great Palace in Bardill 2006b, fig. 1. On the Karea and Nekra gates, see Bardill 1999, 223 n. 35.
144. Proc. Aed. 1.4.1–8, esp. 6, where the basilical form of Sts. Peter and Paul is contrasted with the double-shell form of Sts. Sergius and Bacchus.
145. Mathews 1974, 24: "Jugeant insuffisante la basilique qu'il avait construite comme simple particulier, Justinien éleva à côté d'elle une église à plan centré en forme d'octogone inscrit dans un carré."
152. Proc. Aed. 1.8.2–14.
153. Proc. Aed. 1.8.15–16, tells us that Justinian's church of St. John in the Hebdomon closely resembled St. Michael at Anaplous.
154. On the church excavated in the Hebdomon, see: Mathews 1971, 55–61; Kleiss 1974 For the palace's location near the harbor, see Joh. Mal. 18.114 (ed. Thurn 2000; trans. Jeffreys et al., 1986). For discussion of the church's location in relation to the palace, see: Mango 1975, 391; Mathews 2006, 141. Mango, 1975, 391, suggests that the blue–grey granite column of Theodosius II, found fallen between the church and the shore, is that which fell in 558 (Joh. Mal. 18.124; Theoph. a.m. 6050 (ed. de Boor 1883–1885, trans. Mango and Scott 1997)) and believes Theophanes made an error in describing the latter as a porphyry column. Since Theophanes says the column that fell was "in front of the palace," this would constitute additional evidence that the church was outside the palace. However, Taddei 2014, 81–82 (following R. H. W. Stichel) plausibly argues that the column that fell in 558 was indeed of porphyry and was not that of Theodosius but rather that of Justinian, which is mentioned by Joh. Lyd. de mag. 3.35 (ed. and trans. Dubuisson and Schramp 2006; trans. Bandy 1982), as having been located "in front of the agora of the so-called Hebdomon."
156. This is a theme I hope to explore elsewhere.
157. Henderson 1906, 4, writes: "a small segmental ledge is formed between the groin and angle of octagon, two to each compartment." Van Millingen 1912, 77, notes that "the concave compartments are set slightly back from the octagon's inner face, leaving, at the springing line, portions of the wall-head to appear as little flat ledges on each side of the angles." See Antoniades 1907–1909, 3: 72 n. 271 with fig. 593; van Millingen 1912, fig. 28; Ebersolt and Thiers 1913, fig. 14; Sanpaolesi 1961, fig. 31; Bayülgen 2012, fig. A.43.
159. The inaccuracy of claims that the dome has a smooth profile was pointed out by Sanpaolesi 1961, 142 with fig. 31, 160. The irregularities are clear from the dome profiles published in, for example: Alkis et al., 2001, fig. 5; 2003, fig. 13; Svenshon 2013, figs. 3–4; Bayülgen 2012, fig.A.44. In preparing fig. 7, I have followed Bayülgen 2012, fig. A.44.
160. Van Millingen 1912, 76, observed: "The eight compartments directly above the dome arches are flat, and flush with the inner surface of the octagon, and in each of them is a semicircular-headed window. They rise perpendicular to a point a little above the windows, and then curve with a radius to the centre of the dome." This is broadly correct, although it tends to obscure the fact that the dome in fact begins below the windows. A more precise description, elaborating on Henderson 1906, 4, is given at van Millingen 1912, 78, where it is observed that the radius of the dome changes three times. Initially, from the dome-springing to a point above the tops of the windows, it is 8 m, then the radius becomes just 3.75 m (i.e., a tight curve) to a point about 2 m higher, and finally, from there to the crown of the dome, the radius is 9.5 m, its center being about 2 m below the springing of the dome.
175. For photographs of San Vitale's drum, dome, and squinches, see Deichmann 1958, 280, 281, 287–88, 289–90, 291–92. For scale drawings of the same, see Deichmann 1976, 33–36, 38, 40, 41, 42, 43. For discussion, see Deichmann 1974–1976, pt. 2, 64–67 with pl. 20.
178. The drawing is inaccurate in several respects; for instance, it shows a dome cornice, gives the dome a single radius, shows all segments of the dome as gored, and shows each gored segment terminating in a semicircular lunette. The angels in the illustration (RIBA41474) are noted by Mango 1965, 317 n. 48. Briefly on the building's history as a mosque, see Müller-Wiener 1977, 182.
186. The structural phases indicated in fig. 2 (and also in figs. 6, 9, 16) broadly follow those indicated by Grossmann 1989, fig. 1, and Svenshon and Stichel 2000, fig. 1, who believe the entire narthex to be a later addition. The plan in these figures follows Bayülgen 2012, fig. A.41.
187. See Mathews 1971, 44–46; Grossman 1989, 157–58; Svenshon and Stichel 2000, 405; Kostenec 2005–2006, 105–114. Mathews claims that the structure of the southern wall of Sts. Sergius and Bacchus belonged to Sts. Peter and Paul, and that the few remaining sections of banded brick and stone that can be detected in that wall appear similar to fifth-century work. Yet, Sts. Peter and Paul was certainly built in the sixth century, and in that period is likely to have been built using all-brick construction with occasional greenstone courses, as was Sts. Sergius and Bacchus itself. That is not to deny the possibility that Sts. Peter and Paul was located to the south of Sts. Sergius and Bacchus. No parts of the south wall of Sts. Peter and Paul have been detected in the sea-wall (Bolognesi Recchi Franceschini 2004). However, in October 2006, several sixth-century Ionic impost capitals decorated with peacocks (compare Mendel 1912–1914, 3; 466–67 no. 1242 = Fıratlı et al., 1990, 115–16 no. 215) were unearthed at the sea-wall south of Sts. Sergius and Bacchus and might be adduced as evidence in support of a southerly position for Sts. Peter and Paul. It should be noted that the four piers on the south side of the central octagon of the church are inclined significantly towards the south, whereas those on the north side are vertical (Alkis et al., 2003, 496–501). If this distortion occurred during construction, it might suggest the presence of a large building buttressing the church on its north side, and the absence of one to the south.
190. On the excavated remains, see Harrison 1986 and 1989. On the architecture, see Bardill 2006a and 2011. For the date, see Bardill 2004, 62–64, 111–116. The final results of my research on the architecture of St. Polyeuktos will appear in a dedicated monograph. The author and A. Tayfun Öner are collaborating to produce a digital 3–D reconstruction of the church of St. Polyeuktos.