Wine, Water and the Missing Symposium in Justin’s First Apology

Justin’s First Apology contains the longest extant description of an early Christian meal. This description (ch. 65-67) poses several problems, of which this short article singles out only two. On the level of textual criticism, an oft-discussed variant, rejected in all editions, suggests that the blessing is made over a cup of water, not wine. On the level of liturgical history, Justin’s Eucharist seems to contradict the view that early Christian meals resembled Graeco-Roman symposia. By combining the textual and the historical approach, this article offers a compromise. It is argued that water and not wine was indeed used during the opening ritual, but that the rest of the event did unfold as a symposium and hence included wine.


Introduction
In recent decades, much work has been undertaken to situate early Christian meals in their Graeco-Roman context. While traditional views on the structure and the symbolism of the Eucharist drew a straight line between the New Testament and contemporary practice, this new approach challenges such claims to continuity. In many corners, the symposium-in all its diversityhas replaced the medieval mass as the authoritative model for at least the first three centuries of Christian banqueting culture.1 This has led to reinterpretations of all major meal descriptions in the New Testament and patristic literature, some of which are inevitably more plausible than others. The fact that even popular treatments now make the case for the early Eucharist as a symposium indicates that this is no longer an "alternative" or "radical" view but has become a mainstream paradigm for understanding early Christian meals.2 While numerous observations on individual texts have of course been made, the core of the "Eucharist as symposium" hypothesis remains an assumption about general plausibility: how else would Christians have eaten in a world where all banquets of a certain size seem to look roughly the same? Once the problem is formulated in these terms, explicit proof is not required: it is enough to show that the sources do not explicitly contradict an interpretation that understands the Eucharist as one variant among the regionally and socially diverse practices assembled under the umbrella term "symposium".
If this is accepted, it should be seen as all the more problematic that the earliest descriptions of Christian meals have rather little resemblance to sympotic practice as it is commonly understood. While it is relatively easy to get around Pliny, who remains vague in his description and in any case has not seen a Christian meal himself,3 Chapters 65-67 of Justin's First Apology certainly pose a challenge. Purportedly sent to Antoninus Pius in the 150s CE,4 Justin's description on first sight looks more like a medieval mass celebration  than a Greek symposium. A ceremonial leader (προεστώς) performs a ritual act over bread and a cup, "eucharistizing" the elements before they are distributed by the diakonoi (and even brought to the homes of those who could not attend). Justin also includes a complicated-and much debated-reference to the "change" (μεταβολή) of the meal elements that could be taken to imply transubstantiation; and as if that was not enough, his text has also been understood to distinguish between baptismal and Sunday Eucharists.5 Based on such readings of Justin, those earlier assumptions of continuity do not seem all that misinformed. The first cracks in this picture would appear if we were to accept the old theory that Justin's Eucharist involved water but not wine. This has been the subject of heated debates since Harnack made the case for water in 1891, and it leads directly to problems of textual criticism.6 A positive answer would increase the distance between Justin's Eucharist and later orthodoxy, but it would also seem to further remove his description from ancient sympotic contexts, where the central element was of course wine. This paper makes two claims: 1) While the arguments commonly adduced in favour of the water hypothesis are flawed, it is still likely to be correct. 2) This very same hypothesis can support rather than weaken the case for a sympotic context for Justin's meal. As textual criticism is at the heart of the matter, the argument can serve as an illustration of the entanglement of textual and ritual variants in early Christianity.

The Manuscripts and the Editions
Justin's Apologies are preserved in an almost complete form in two manuscripts: Parisinus Graecus 450 (A) of 1363 and Phillippicus 3081 (a) of 1541. As the latter has been proven to be a direct copy of the former, we are left, for the most part, with a single manuscript.7 That manuscript is not very good. available thus supports the view that the Parisinus contains a version of the Apologies that has not only suffered from numerous corruptions but has also undergone considerable embellishment. Unlike the rest of the Apologies, the description of the meal in 1 Apology 65-67 is preserved in four additional manuscripts, two in Greek and two in Latin, which seem to have been created in the context of 16th century-debates about the Eucharist, particularly the Council of Trent.12 The only one that has played any role in scholarship so far is Ottobonianus Graecus 274, written between 1548 and 1555 (i.e. possibly after the first edition of Justin in print, published in 1551). The manuscript contains several obvious mistakes, but also one variant that has given rise to wide-ranging interpretations, to be discussed below. Other ancient Christian authors do not offer anything of use for our purposes. 13 There are three relatively recent editions of Justin's Apologies, each following different principles. 1) Marcovich's edition of 1994 includes a large number of emendations, drawing the logical conclusion from Harnack's assessment of the quality (or rather lack thereof) of the Parisinus. These emendations, which also affect the description of the meal, are often of a stylistic nature and have been criticized for this reason.14 2) Munier's edition for the Sources Chretiennes, published in 2006, is different in that it rejects the harsh verdict about the Paris manuscript. 15 It follows the Parisinus as closely as possible, with only occasional emendations. This approach is now rendered problematic by the discovery of P. Oxy. 78.5129.

3) In 2009, Minns and Parvis have published their edition for Oxford Early
Christian Texts, again based on different principles. Although they take a critical stance towards Marcovich's tendency to interfere with the text, their own approach is the most intrusive of the three. They differ from both Harnack/Marcovich and Munier in that they see the Parisinus as an intellectual contribution in its own right. On this view, the Parisinus contains not just a flawed copy of, e.g., a hypothetical 10th century manuscript (Harnack's argument), but a deliberate attempt to make sense of a tradition that had already been corrupted at an earlier point in time.16 Their ambitious aim is therefore to cleanse the text from both those 14th-century attempts to create cohesion and from the earlier corruptions. The result is a sometimes radical rewriting of Justin, a method that may find partial justification in the new fragment from Oxyrhynchus. While these editions thus work on different assumptions and present somewhat different texts, they all agree that the Parisinus (henceforth A) is the only manuscript worth considering. The manuscripts containing only chapters 65-67 are not seen as having any independent value, and there is reason to believe that this judgment is indeed correct for most of them.17 The Greek Ottobonianus has a special status because some earlier editions have considered it to preserve an independent tradition,18 and its reading of 65.3 has caused a major controversy; however, all recent editions ultimately conclude that it is irrelevant for reconstructing Justin's (or any ancient) text.19

The Value of the Ottobonianus
There is no doubt that O contains mistakes not present in A. It may be possible to attribute some sense to a casus deviation like 65.5: καλούμενοι παρ' ἡμῖν A; ἡμῶν O. However, there are also passages that make no sense at all in O-most blatantly 67.3, where the assembly of people living "in the cities or 16  in the countryside" (κατὰ πόλεις ἢ ἀγροὺς A) turns into an assembly "towards the breads" (κατὰ πόλεις εἰς ἄρτους O). In this sense there is every justification for the assertion that "O is by far inferior to A".20 However, this alone does not tell us much about the value of O as a witness to a potentially ancient text. Things only change when we argue that O not only has a generally worse text than the Parisinus, but ultimately depends on it. Whereas a higher number of errors would still allow for individual cases where O preserves an ancient (and perhaps even a more original) text than A, O's dependence on A would indeed reduce its value for reconstructing an ancient text to zero.

(O)
A's text reads better than O's (but neither is easy to understand). Minns and Parvis argue that the version of the Parisinus was at an earlier stage corrupted by casus assimilation (ἔργων ἀγαθῶν), and that at a later stage, πολιτευταί was changed to πολίτευσθαι to make the text grammatical again. If that later stage was O, it would be removed from A by two stages, but it might well be more. The development as such is certainly plausible. It is also true that a construction with διά is an odd way to qualify πολιτεύ(ε)σθαι. However, πολιτεύ(ε)σθαι is used by Justin elsewhere while πολιτευταί is not; in one instance, πολιτεύεσθαι even appears in combination with καταξιῶσθαι. We therefore have to evaluate the one variant that actually matters (the content of the ποτήριον in 65.3) on its own merits, without the backing of a redactional model that ultimately cannot be proven.

The Cup
The cup brought to the ceremonial leader in 65.3 is described as a ποτήριον ὕδατος καὶ κράματος in A, but as a ποτήριον ὕδατος in O. Both texts are problematic. If the cup contains only water (as in O), how are we to explain the inclusion of water and wine in 65.5 and 67.5? If it contains water and "mixed wine" (as in A), this would not only appear redundant (as the wine would of course be mixed with water); it also causes similar problems of coherence because κρᾶμα is never mentioned in later passages.
Options to make sense of this abound, but they all depend on the assessment of the manuscripts. If we see O as dependent on A, all that needs to be done is explain the redundant formulation in A. Solutions have indeed been proposed: we can postulate that κρᾶμα is simply synonymous with οἶνος (the problem being that this is not true),25 or we can change the text until it suits our expectation: to "a mixed cup of water and wine" (poculum aqua et vino temperatum),26 to "a cup containing a mixture with water" (ποτήριον <ἔχον> ὕδατος κρᾶμα),27 to two "cups of water and mixed wine" (ποτήρια ὕδατος καὶ κράματος),28 or to a "cup mixed with water" (ποτήριον ὕδατι κεκράμενον).29 There would then be no need to think of a possible ritual context that involves only water. Seeing O as independent of A opens up the possibility of seeing ποτήριον ὕδατος as a 25 Schmid, "Textüberlieferung" (n. 6) 125 discusses the relevant arguments. version in its own right, perhaps even the original text. We would then have to explain A's ποτήριον ὕδατος καὶ κράματος as a later modification. This was the view of Harnack, who thought that Justin's use of water in the Eucharist had caused problems for a later orthodox redactor, who added wine.30 This view has been endorsed more recently by a number of scholars,31 leading to renewed attempts at refutation.32 Before venturing into a possible reconstruction of ritual, the arguments brought forward in support of Harnack's theory therefore deserve a closer look.
The first argument is that wine has mistakenly been inserted in two passages that do not concern the Eucharist, as a sort of "hypercorrection". In 1 Apology 54 and Dialogue 69, Justin discusses the prophecy in Genesis 49:11 ("Binding his foal to the vine and his donkey's colt to the choice vine, he washes his garments in wine and his robe in the blood of grapes"). Justin notes that the demons have imitated this prophecy by calling Dionysus the inventor of the vine and using wine in his mysteries. The idea of hypercorrection is supposed to explain the fact that the wine is mentioned twice while the donkey is left out. On this view, Justin actually wrote that the demons use a donkey (ὄνος) in the mysteries, but a corrector replaced it with wine (οἶνος), because he had just done (or had set out to do) the same thing in the description of the Eucharist. Leaving aside questions regarding the plausibility of a concerned orthodox turning suspiciously looking donkeys into wine, the argument is unconvincing. Right after this passage, Justin goes on to state that the demons have identified Bellerophon with the Messiah, because they did not know if the expected riding animal was a horse or a donkey.33 The donkey is thus part of the explanation; it is just not part of the discussion of Dionysus. Justin uses two Greek myths to explain two different demonic imitations of an ancient prophecy. There is no room for the assumption that ὄνος was corrected into οἶνος.
The second argument is based on two passages where Justin compares the Eucharist to contexts involving only water, not wine. The more pertinent one The theory of the orthodox corrector is thus difficult to uphold. But this just leaves us with the same problems as before. Two vantage points for reopening and perhaps refocusing the debate can nevertheless be identified, one general and one very concrete: 1) So far, the debate has been informed by a desire for coherence. All solutions presuppose that if 65.3 did indeed mention only a cup of water, the same would have to be the case in the other passages mentioning the cup (and vice versa: if these do not mention only water, it cannot have been mentioned alone in 65.3). However, absolute coherence may be a modern rather than an ancient preoccupation. An attempt should at least be made to imagine a ritual context where it is not required.
2) Focusing on the variant in O can easily result in the impression that water is the problematic element of the equation. It is therefore important to keep in mind that water as such is not problematic at all in 65.3: it is present throughout chapter 65-67 (and in fact earlier, as chapter 65 continues where 61 on baptism had left off). The element that really causes problems is κρᾶμα, which is only mentioned in 65.3, and only in A-a text that has been discredited by the discovery of P. Oxy. 5129, regardless of our view on O. There is thus much to suggest that καὶ κράματος is indeed a later embellishment or, perhaps, a corruption. From this, we can deduce two questions that remain to be answered. Can we develop a scenario where ποτήριον ὕδατος in 65.3 is indeed an ancient reading, but where the community involved-perhaps: Justin's community-also used wine in the Eucharist? And can we explain the introduction of mixture in an ancient context?

Reconstructing Ritual
Two observations should inform any attempt to reconstruct a meal ritual from Justin's Apology. The first relates to the relative importance of the meal elements mentioned throughout chapters 65-67. Including the comparison with Mithraic meals, there are four enumerations of meal elements (65.3; 65.5; 66.4; 67.5). Bread and water are mentioned in all cases, in all versions. Wine is missing twice, in 65.3 and 66.4. This suggests that bread and water are more important to the author than wine. The second observation is that wine can be present during the ritual even when it is not explicitly mentioned. This becomes evident when wine (οἶνος) appears in 65.5 after it has not been mentioned in 65.3 (where O has only water and A adds κρᾶμα but not οἶνος). Clearly, wine must already have been in the room for it to be affected by the blessings spoken over the bread and the cup. But it is less important for the ritual than bread and water, and it is mentioned only when the text tries to give an impression of the event as a whole, without the focus on the opening ritual performed by the προεστώς. This distinction between an opening ritual and the unfolding banquet may indeed be crucial for understanding the scenario envisaged in 1 Apology 65-67. The evidence is consistent with the assumption that before the beginning of the meal, Justin's text envisages a ritual that involves bread and water. The passages mentioning only bread and water focus on this ritual alone. The similarity of this ritual to a ritual performed in Mithraic circles may have informed Justin's case for the demonic imitations in 66.4. It is overwhelmingly likely-and must have been known to Justin-that Mithraists drank wine during their banquets.35 However, he does not mention wine in 66.4 because it is not important for his argument: he compares what he regards as the respective core rituals, not the banquets as such. In a similar vein, the interpretation of Isaiah 33:16 in Dialogue 70 focuses only on this ritual, not on everything else that is going on at a Christian (or Mithraic) meeting. When this "everything else" comes into view, wine immediately appears. In 65.5, its appearance marks the transition from the ritual act performed by the ceremonial leader, i.e. a special moment of the meeting with only one active participant, to the diakonoi and their interaction with all the people who are present. What these people then do is not told in any detail, but we know that it involved wine. We can assume that they held a symposium.
There are two obvious arguments against this neat dichotomy. The first is that in 65.5, wine is listed among the elements that have been "thanked for" (or "eucharistized") earlier, i.e. in 65.3, where wine is not mentioned. However, an explanation for this seeming contradiction (which of course informs the textual emendations discussed above) can be found in the interpretation of Christian meals as symposia. For any ancient symposium, wine is an obvious prerequisite. It is stored in a big kratēr, in the center of the room. During the opening ritual described in 65.3, the wine would already be in the room, ready to be distributed once the ceremonial leader has finished the blessings. These are uttered specifically over bread and water, but the prayer may easily have encompassed all meal elements.36 Comparison with other ancient rituals introducing meals shows that this basic structure would have been common. When Plutarch says that we render everything we eat holy by making an offering from it before the meal, we are probably not supposed to sacrifice a piece of every possible meal element; rather, a bit of food must have been treated as representative of the whole meal.37 The introductory libations of the potos (the part of the symposium that focused on drinking) also treat a small bit of substance-in this case, unmixed wine-as representative of everything that is consumed later (i.e. mixed wine and snacks).38 As these traditional rituals involved idol worship, Christians would certainly have found them problematic, but it is plausible to assume that instead of doing away with them completely, they replaced them with structural equivalents. If opposition to the use of wine can indeed be explained by its use in sacrificial rites,39 the only situation where this would become relevant would be the beginning of the meal. There is thus no contradiction between the use of wine in Justin's meal and its exclusion from the ceremonial leader's cup.
The second counter-argument may be derived from a later passage in Justin's account. Our solution requires the assumption that the wine is not to be found in the cup that is ritually treated by the ceremonial leader, but is merely an element of the meal, transformed into a different status through a ritual that explicitly involves only bread and water. However, in 67.5, "bread, wine and water" are brought in again, Justin points to his earlier description, and he even uses the same word προσφέρεται. This might be taken to imply that at least on this occasion (and by extension also in 65.3), wine is among the elements involved in the ceremonial leader's ritual. However, it is important to note that while there is indeed much overlap between 65.3 and 67.5, προσφέρεται is used differently in both cases. 65.3 explicitly states that bread and cup are brought to the ceremonial leader (and only to him). In 67.5, "bread, wine and water" are merely brought into the room, without a direct object.40 It is clear enough that τις οὖν αὕτη παρατεθῇ, ταύτης πρὸ τῶν ἀπολαύσεων πάντας ἀπάρχεσθαι μικρὸν μέν, ἀλλὰ τῷ μικρῷ τούτῳ παντὸς μᾶλλον μεγάλη τίς ἐστι τιμή ("That the god is pleased not by the size of the sacrifices, but by anything offered to him, is clear from the fact that all people make an offering of their daily food before they enjoy it, no matter what is served; it is a small offering, but this small offering is worth more than anything else"). See also Nicolaus of Damascus FGrHist 90 F 103 (l): Πισίδαι δειπνοῦντες ἀπάρχονται τοῖς γονεῦσιν, ὡς ἡμεῖς τοῖς θεοῖς παρασπονδίοις ("When the Pisidians dine, they make an offering to their parents, just as we make an offering to the gods presiding over libations" the wine needs to be brought into the room at some point for it to be available for consumption later. This does not imply that bread, wine and water are all treated equally. The difference between the two descriptions is that Chapter 67 focuses on the communal context throughout. It does not specifically focus on the ritual involving the bread and the cup; instead, 67.5 jumps almost immediately to the effect of the ritual on the community. This latter observation leads to the old question why Justin offers two descriptions of the Christian meal. Attempts to divide Chapters 65 and 67 into two different occasions (baptismal Eucharist and Sunday Eucharist) appear problematic in light of 67.5, where both descriptions are explicitly linked to the same event.41 One solution may be to see the Apologies as an assortment of sometimes unconnected notes,42 provisionally glued together by remarks like ὡς προέφημεν in 67.5. Justin (or a redactor) might then have mistakenly included two notes on the same event. The version given in 65-66 approaches the topic from the perspective of the newly baptized, who have just undergone an initiation ritual where water is important; this may explain the focus on the opening ritual, which also involves water. The version in 67 approaches the same topic from a different angle, offering a description of what Christiansas a community-do every Sunday; from this perspective, the water ritual is less interesting. While this explanation may have some validity, it is probably wise to admit that there are many possible reasons why an author would choose to change focus within a given argument. We should not overestimate our ability to uncover them.
If this reconstruction is accepted, reading ποτήριον ὕδατος in 65.3 does not cause difficulties anymore. A plausible ritual scenario can be developed based on this reading. The remaining question is how we can account for what should now be regarded as the later introduction of mixture into this picture. We have already seen that the mere addition of καὶ κράματος in A simply does not yield a satisfactory meaning and should be seen as either a corruption or a misinformed embellishment. That A is likely to contain embellishments has become clear through the publication of the Oxyrhynchus fragment. However, a change to a "mixed cup of wine and water" (Périon's plausible Latin translation) or the (n. 19) 518 simply notes that προσφέρεται is used "identisch" in both passages; he also has to assume that "wine and water" are "ein Wein-Wasser-Gemisch". 41 Justly stressed by A. Lindemann, "Die eucharistische Mahlfeier bei Justin und bei Irenäus," in like may have occurred well before 1363 for A to be able to have what would then not be an embellished, but quite simply a corrupted text. When exactly this might have happened is impossible to guess. The issue would certainly have become more pressing after the Eucharist had lost its sympotic character.43 In a situation where all that remained was the ritual that we have interpreted here as an introductory performance before the meal, the symbolism of that ritual and the content(s) of the cup inevitably attracted more attention. Debates about the nature of Christ as reflected in the elements of the cup were not limited to the council of Trent, and one could easily see a late antique reader of Justin's Apologies worrying about the lack of wine in the ceremonial leader's cup. This assumption does not merely revive the "orthodox corrector" hypothesis, because it does not claim that wine has been added anywhere else in Justin's text. As our solution has detached 65.3 from the main part of the meal, no coherence between the elements contained in the cup and those mentioned in the later parts of Justin's description is required. This very fact made the content of the cup all the more prone to modifications that would bring it in line with the ritual practice of a given community, or, perhaps, with the particular version of that practice envisaged by a redactor.