Lex orandi, lex credendi; it is a solid and worthy theological principle. But if it is interpreted to mean that theology always follows a pre-existing liturgical practice, it becomes questionable. In periods of great theological upheaval, the ambiguity of certain liturgical expressions comes to light in new ways, and there is pressure to clarify: the lex credendi moulds the language of prayer, just as much as it has itself been moulded by earlier devotional practice. An examination of how the metaphors of sacrifice in (especially) eucharistic worship were redeployed by Protestants in the Reformation period is illuminating; in the English-speaking world, we could also point to the ways in which Archbishop Cranmer and his colleagues so often subtly modified in their translations the idiom of the mediaeval collects in order to reinforce a reformed doctrine of grace. More recent liturgical revision in many churches shows exactly the same processes at work: questions about offering or oblation are still live issues in Anglican liturgical revision, and recent debates in the Church of England Synod exposed deep anxieties about the implied doctrine of grace in certain new collects.

My chief focus of interest for the present, though, is on an earlier age of theological revolution, the fourth century. The significance of liturgy in the controversies around trinitarian doctrine in this period is being more fully recognised as time goes on, and it has been suggested that some of the most problematic phrases and ideas in the teaching of Arius himself may be based on archaic elements in the liturgy of his church (which are duly phased out in the course of the fourth century). The difficulty faced by all controversialists of the period who want to appeal to worshipping practice is that the formulae of public worship are so often capable of interpretation in opposite senses. We may see the traces of this in the use of baptismal practice and formula in the polemic of the day (as I have argued elsewhere); and Basil of Caesarea effectively grants the seriousness of the difficulty in the de spiritu sancto, when he appeals to the dogmata of routine practice and teaching as a means of rightly understanding the kerygmata of public formulae.

It is a text associated with Basil that prompts this particular study. Recent scholarship is still inclined to look favourably on the conclusion that the Byzantine eucharistic liturgy called by the name of Basil is indeed in substantial part from his hand. Of the relations between the Byzantine text and the Alexandrian 'Basilian' liturgies, I do not intend to speak here: it is the Byzantine text that poses the more interesting questions in this context. If it is true that, as Kenneth Stevenson says, the Greek Alexandrian text shows Basil updating an earlier liturgy 'with the theological developments of the fourth century', the longer Byzantine text is a further and quite extensive continuation of the same process. However, I want to suggest, on the basis of an examination of one part of the text, that this process is a little more complex than a simple updating with fresh theological elements; and that the results of such an examination throw some more general light on the interrelations between theological and liturgical formulation in the fourth century.

The portion of the liturgy I shall be examining is the pre-Sanctus prayer which follows the sursum corda dialogue. The Byzantine Basilian prayer is notably long and rhetorically rich (though, as we shall see, it is not alone in this), twice the length of the corresponding text in the liturgy of St John Chrysostom; and the passage I want to discuss is the opening invocation:

Truly-existing One, Master, Lord God, Father, Ruler of all, worthy of worship, it is truly fitting, just and proper to the greatness of your holiness to praise you, to sing hymns to you, to bless you, to worship you, to give you thanks, to glorify you, the only truly existing God, and to offer you with a repentant heart and a spirit of humility this service of ours that belongs to our rational nature. For it is you who have granted us knowledge of your truth, and who is fit to speak of your mighty doings or make your praisesto be heard, or tell of all the wonders you do in every age? Master, Master of all, Lord of heaven and earth and of every creature, seen and unseen, you who are seated on a throne of glory, beholding the depths, without beginning, invisible, ungraspable, unlimited, unchangeable, the Father of our Lord Jesus Christ, the great God and Saviour, our hope, who is the image of your goodness, the exact seal that in itself reveals you as the Father, living Word, true God, Wisdom before all ages, Life, Sanctification, Power, True Light, from whom the Holy Spirit has shone forth...

This is a text densely woven with biblical allusion and citation, but it is also notable for the long sequences of adjectives and titles ascribed in turn to the Father and the Son (and the following phrases treat the Spirit in much the same way), some biblical and some not. This is manifestly a doctrinally 'saturated' prayer, and there are elements, as we shall see, of special relevance to fourth century debates; but what is most striking is that a good deal of the rhetorical elaboration echoes both liturgical and credal formulations that do not belong to quite the same theological world as we might expect Basil to inhabit. The fact is that the phraseology shows many signs of being rooted in an idiom common throughout the Antiochene sphere of theological influence and shared by both pro- and anti-Nicenes in the fourth century, and with clear origins in the pre-Nicene period.

Take a sample of liturgical and credal texts from the region across the fourth century - the statement of faith by the Antiochene Synod at the beginning of 325 (surviving only in Syriac; there must be some caution about Greek retroversions, but the standard Greek text of Schwartz has a high degree of plausibility for the most part), the second creed of the Synod of Antioch in 341 (associated with the name of the martyr Lucian, though it is impossible to assess the reliability of this tradition), the eucharistic liturgy in the eighth book of the Apostolic Constitutions and the confession of faith of Eunomius, Basil's arch-opponent – and some patterns emerge. The address to God as ontos on, 'truly existing', is common to Basil, the Apostolic Constitutions liturgy (AC henceforth) and Eunomius. Basil and AC share the plain phrase ontos onta theon (Basil prefacing it with monon), Eunomius refining a little with ontos onta phusei te kai doxe theon hena, 'truly existing by nature and glory as one God', an expansion fitting the emphasis in this formula on the impossibility of any sharing of the divine nature. Anarchos, 'without beginning', is there in Basil and Eunomius; AC has it in association with and directly following monon agenneton and also in a rather later position in its pre-Sanctus prayer, where God is addressed as anarchos gnosis, 'knowledge without beginning'. Akataleptos, 'ungraspable' is common to Basil and (probably) the Antiochene formula of 325; it appears also in the liturgy of Chrysostom, but – unsurprisingly – is absent from AC and Eunomius, both of which have serious theological objection to phrases suggesting that God is radically unknowable. Paradoxically, the word seems to have been important to Arius, and was the major point at issue between his theology and that of other opponents of the Nicene formula. Analloiotos, 'unchangeable' appears in Basil and is paralleled in the Antiochene creed of 325 (with the usual caution about Greek retroversion).

So in regard to the epithets applied to God the Father, there is some significant overlap between Basil and these other texts. Aperigraptos, 'unlimited' is unique to Basil, and aoratos, 'invisible', is shared only with Chrysostom. The agennetos of the AC formula is not a word we should expect in Basil, given the polemical importance of the phrase monos agennetos among virtually all non-Nicene groups. What of the series of titles for Christ? Here the pattern is if anything even more marked: of the sources listed, the Antiochene creed of 341 and Eunomius's confession have very similar lists of titles. Logos zon, 'Living Word', appears in the Antiochene creed of 341; AC has simply logos theos. Sophia, 'Wisdom', is in Antioch 341, AC and Eunomius; in all three instances, what we have is actually sophia zon, 'Living Wisdom', and it is perhaps significant that in Basil sophia is immediately followed by zoe, 'life'. It is not impossible that the formula may have originally been the same. Dunamis, 'power', appears in Basil and Eunomius. Phos alethinon is shared by Basil, Antioch 325, Antioch 341 and Eunomius, and in the first and last of these is the final member of the series of titles. The sequence logos – sophia – phos is common to Basil and Antioch 341; Antioch 325 has only logos and phos. It is perhaps worth mentioning that the 'Creed of Gregory Thaumaturgus' has a logos – sophia – dunamis triad; but the questions around this text are manifold. It seems to be a regula fidei summary for catechesis rather than a strictly liturgical formula, and its connections with any of the other texts here reviewed are not evident.

Moving on, it is again striking that Basil's use of eikon, 'image', is parallelled in the same family of texts. For Basil, the Logos is the image of God's goodness; Antioch 325 initially has simply has 'the image', though we should probably supply 'of the Father', with Schwartz; later in the text we read that the Son is image of the Father's hupostasis, not just of his will or any other attribute or activity. Antioch 341 has 'the exact image of the substance and will and power and glory of the Father's divinity'; Eunomius has 'the image and seal of all the pantocrator's action and power, the seal of the works and words and counsels of the Father'. It is ironic that the image-seal connection appears both in Basil and in his arch-enemy, but not elsewhere in the texts we have been examining. However, the identical collocation can be found in Gregory Nazianzen's 38th Oration (13), complete with the adjective aparallaktos attached to eikon, as in Antioch 341; eikon and sphragis also appear together in his Third Theological Oration (17), with reference to Jn 6.27 as a biblical justification ('On him has God the Father set his seal'). It is just possible that there is another irony as well: Basil uses theos alethinos of the Logos; and the direct manuscript tradition of Eunomius's confession does the same. When Gregory of Nyssa quotes it, he has huios alethinos, 'true Son'. Given that theos alethinos was one of the flagship phrases of Nicene orthodoxy, it may well be that the codices are wrong (as Eunomius's most recent editor notes, they have a common exemplar). The shadow of a doubt may remain: Eunomius was a bold controversialist who might not have hesitated to carry the battle into the enemy's territory by giving to a Nicene phrase a wholly non-Nicene sense. But the fact that Jn 17. 3 uses monon alethinon theon for the Father as opposed to hon apesteilas Iesoun Christon probably settles the matter.

In summary, the evidence suggests quite strongly that Basil's prayer is part of a recognisable pattern of credal language in the Antiochene world (West Syria and Asia Minor). If we allow that the AC text and Eunomius have a clear interest in playing down one element, the stress on divine unknowability, we can still see the broad lines of a sequence beginning with the ascription to God of 'true' existence (i.e existence independent of any contingent being), in line with the Septuagintal version of God's self-naming to Moses in Exodus 3.13 as ho on, 'the existing one'; continuing with reference to God's eternity, incomprehensibility and immutability; naming the 'Living Word' as Wisdom, Power and True Light, and quoting scripture on the nature of the Logos as exact image. The 341 expression aparallaktos eikon is used by Asterius the Sophist, and it is possible that its popularity with a noted opponent of Nicaea may have prompted Basil to modify it (although Gregory Nazianzen does not apparently share such scruples), and to favour an Old Testament locution ('the image of your goodness', referring to Wisdom 7.26) over the wording of Antioch 341, 'the image of the Father's substance', with its clear grounding in Heb. 1.3. The unusual use of sphragis in both Basil and Eunomius strongly suggests some common point of reference; it wold be hard to imagine either a casual coincidence here or a deliberate echo one way or the other. And, although the use of 'seal' is not common in speaking of the Logos, Gregory Nazianzen's use of the sphragis metaphor in proximity to eikon language tends to reinforce the conclusion that this is not just an idiosyncracy of Basil's. And indeed a glance at Nazianzen's Fourth Theological Oration (20) gives us a sequence of titles obviously within the same general territory – logos, Wisdom, Power, Truth, Image, Light, Life, Justice, Sanctification, Redemption and Resurrection. In this list, sphragis appears as a gloss on 'Truth', just prior to 'Image'. The relative order of Word, Wisdom and Light is identical to Basil and to Antioch 341, with the added insertion of dunamis in its Basilian position.

It looks, then, as though Basil's pre-Sanctus represents something of a mosaic of familiar and sanctioned theological phrases in the Antiochene tradition. Specifically, there is a clear relation between these expressions and the kind of theology – classically found in Antioch 341, the 'Creed of Lucian' – that represented not so much an anti-Nicene but a para-Nicene style, a theological idiom that was very shy of the homoousios but disdainful of the speculations of Arius; it is roughly the theology associated with 'homoiousians' in the mid-century, for whom, as has often been said, Antioch 341 was something of a rallying point. In such a perspective, we can see the language of AC and Eunomius moving off decisively in one direction, eradicating reference to the unknowability of God – though, as I have argued elsewhere, I am cautious about concluding simply that AC is as it stands a 'Eunomian' composition. The absence in AC of eikon language is a little puzzling, but may reflect caution about one of two possible dangers from the probable theological standpoint of the compiler – either the risk of a contrast between the unknowable Father and the knowable Logos (Arius's theology) or that of a robust claim that the 'content' of image and prototype must be identical (Athanasian theology).

Basil's development of the same tradition takes it in the opposite direction, intensifying the language of divine incomprehensibility and reinforcing, though not all that strongly, the Son's perfect representation of the Father's nature; the language is still intelligible within the framework of Antioch 341. The omission of 'firstborn of all creation' (quoted both in Antioch 341 and the AC) from Basil's text may also suggest a conscious avoidance of what had become a highly problematic phrase. What appears to emerge is a careful political strategy in the composition of the prayer: Basil is elaborating a liturgical formula which is sufficiently deeply rooted in local theological tradition to be a plausible bridge between Nicene Christians and the conservative majority in Asia Minor and North West Syria. Here is a composition comparable to AC, in that it is a selfconsciously traditional production, signalling to those who read or hear a clear continuity with earlier formulations, yet slanting the tradition in a specific direction. The AC liturgy is an immensely prolix text, whose sources in Jewish prayer as well as local Antiochene usage have been explored a little (there are some parallels too with the credal formula found in the Epistula Apostolorum); Basil is more economical – though much more expansive than in the presumably earlier essays in liturgical composition represented by the Alexandrian texts. It is tempting to speculate that the appearance of something like the AC liturgy, or the drawing of it to Basil's attention, triggered a longer composition from Basil; but this is probably pushing the evidence further than it will persuasively go. It is a text that generally avoids manifestly controversial phrases: theos alethinos is the likeliest example of a pushing out of the boat in a clearly Nicene direction (and the reference to God as creator of things 'seen and unseen' could be a specific allusion to Nicaea), but on the whole, as I have said, the phraseology would not have raised too many hackles at Antioch in 341.

Basil's conciliation of the 'homoiousian' clergy is one of the clearly evidenced themes of his ministry. Like Athanasius, if less overtly, he is working to convince conservatives that Nicaea is the only real alternative to the Eunomian theology that has shocked and unsettled them (and indeed cost them their jobs, in many instances). If we can allow something of the same impulse in Nazianzen's theological orations also, especially in the passage referred to earlier in the fourth of these, we have a strategy that is more than Basil's individual concern: Nazianzen too, in the passage cited, is drawing on what looks like a traditional list of titles for Christ and giving each of them a clear Nicene 'steer'. In the overall context of Cappadocian theology, the Liturgy of Basil may have a more significant place than has sometimes been thought.

The whole of this discussion also raises a wider issue. If the reading proposed is right, there is a clear interaction between the credal formulae of the Antiochene Councils and this particular portion of the liturgy in that region. The pre-Sanctus, in other words, is expected to carry a heavy load of catechetical significance. It is generally, of course, a place in the liturgy where some recital of the nature and attributes and acts of God is thought appropriate. What tends to mark out this early Antiochene tradition (contrasting with both the primitive Egyptian models that survive and the early Western usage) is the considerable extension of the narrative of God's works to constitute a sort of foreshadowing of the anaphora proper, a full account of creation, as in AC, or a detailed recital of trinitarian relations, as in Basil; and Chrysostom anticipates the prayer following by summarising briefly the history of Christ's work. There is in this tradition an obvious expectation that this is the point where the lignes maitrises of theological orthodoxy are traced.

Remember that in the fourth century there is no liturgical recital of the creed during the Eucharist; if, then, there is to be a presentation of orthodox teaching to the faithful, this seems to be the place where it is done. We have found two related and 'politically' significant liturgical compositions doing important pieces of implicit catechesis here. But we might pursue the question a little further. Is the movement simply one-way, from creed to liturgy? It is not just that credal material is absorbed into liturgy; the credal material itself has a strongly doxological character, most evident perhaps in the 341 Antiochene text but by no means absent from the 325 statement of faith, which has a slightly perfunctory allusion to the biblical titles of the Logos. What we know of baptismal professions and the various versions of a regula fidei in the literature suggests that these formularies are not likely to include the piling-up of attributes or titles for Father and Logos, or the complex tissue of biblical allusion found in all the texts we have been examining: the contrast with, say, the baptismal creed preserved by Cyril of Jerusalem is striking. There is, in any case, as recent research has stressed, little solid evidence for anything like detailed baptismal creeds prior to the fourth century. But might we suppose that the credal statements in this Antiochene group are in part based on existing liturgical texts or at least conventions other than individually-formulated summaries (what have been called 'private creeds') or baptismal formulae only? Was there an Antiochene style of praying the pre-Sanctus which provided a framework for conciliar credal formulation?

The case is worth considering. The location and rationale of the pre-Sanctus mean that the actual shape of the formula is governed by the need to express why it is 'right and just' to praise God; thus it naturally requires a recitation of divine attributes and actions, whereas a baptismal interrogation requires only an identification of the three persons as objects of belief. To take what is at first sight the most economical of the Antiochene texts, the 325 creed, we find the attributes of divine transcendence named and the Son's work as image and revealer of the unknown divinity spelled out with a scriptural citation; the slightly later reference to the saving work of Christ is prefaced by a list of titles and a reiteration of what it means to call the logos the Father's image.

The overlap with the rules of faith found in so many early authors is obvious; rehearsing the occurrence and the climactic events of the incarnate life is both an identifying of who it is in whom the believer professes faith, and a justification for giving thanks to God. I do not suggest that we can identify a credal formula that is somehow wholly independent of the structure and wording of the baptismal questions or the regulae fidei; simply that such a template does not account for the overall style of the Syro-Asian statements of faith as we have them, and that a eucharistic context for their Vorlage does so rather better. If we delve a bit further into eucharistic traditions, we find within the same general geographical area a pattern in which there is emphasis upon the utterance or revelation of the divine name spoken in or through the Son. It is there in the Didache, in the Liturgy of Addai and Mari, but also in very striking form in the eucharistic prayer preserved in the Acts of John – another product of Asia Minor. The invisible Father is made visible (but only to the 'pure') through the humanity of the Son who speaks the Father's name and his own, and is characterised in a long list of epithets – a very different set from those we have been considering, but reflecting a thorough quarrying of scriptural metaphor (seed, salt, pearl, treasure, plough, net, crown). The invocation of these revealed names in the prayer of thanksgiving is what makes the goodness of God manifest here and now (in the bread of the sacrament, presumably, as the instrument of enlightenment).

Thus the solemn recitation of the mystery of God's transcendence and incomprehensibility followed by the recitation of the names ascribed to the revealing Son is the basis of the eucharistic miracle. Although the Acts is a text with strong gnostic elements, we should not assume that it represents a eucharistic theology peculiar to gnostic groups, given the foregrounding of the revealed name in the Didache and the later texts we have been examining. If the eucharistic prayer is understood as the invocation of God's revealed name(s), we should indeed expect the opening of the prayer to have just the quasi-credal style identified. Leaving aside the vexed question of where and when an Institution Narrative became intrinsic and essential to the prayer, there seems a strong case for supposing that initially what mattered was the calling upon God in the name of Jesus, apostrophising both Father and Son by certain revealed titles.

If this was in fact the case, particularly in Syria and Asia Minor, then one of the most obvious places from which to derive an authoritative statement of faith in times of controversy would be the recital of the names of God and the Logos in the foundational invocation of the eucharistic prayer, a context in which emphasis upon the inherent mystery of the divine being and upon the multiplicity of names associated with the saviour, and a creative use of scriptural allusion would be entirely appropriate. A council seeking a clear statement of what was taken for granted as a measure of orthodoxy might well look not only to earlier regulae or to the shape of the baptismal interrogation but to the eucharistic prayer; a theology identified as failing to comply with the pattern of such a prayer would stand self-condemned as a failure in piety, not simply in accuracy. And, if other studies of the impulse behind non-Nicene theologies are on the right track, Nicaea was very probably seen by many as undermining the essential dynamic of this central prayer in the Church's life by weakening the dialectic of the transcendent Father's self-revelation through the Word (both in creation and in redemption). All of which in turn gives plausibility to the enterprise Basil is undertaking, the revision of a traditional pattern of Antiochene liturgical language to produce a degree of compatibility with Nicaea, without sacrificing the central imaginative logic of that traditional pattern: Basil's text makes much of the divine incomprehensibility and of the Son's perfect manifestation of his prototype, though steering away from a simple contrast between invisible Father and visible Son by stressing the exact correspondence of image to prototype and the idea that the Logos reveals his source specifically as Father.

In the recent and admirable survey of scholarship on credal origins, Kinzig and Vinzent argue that all fourth century creeds are in some degree dependent on the exchange of formulae of belief found at the beginning of the controversy over Arius, the statements of the two initial principals in the debate, Arius and Alexander. They also conclude that these two formulae set the style for later work in that they represent a deliberate attempt at mutual 'colonisation' – i.e. each side uses 'building blocks' drawn from the other's text. Arius appeals to what Alexander has been teaching, Alexander employs phrases central to Arius's argument and offers alternative readings of them. Thus, according to Kinzig and Vinzent, the synodical formulae of the fourth century should be seen as grounded 'in the attempt of their authors directly to reply to their respective opponents in the latter's own terminology and to emphasize one's own readiness for compromise and consensus in order to defend and to assert one's own views ever more effectively'.

Broadly speaking, I am very sympathetic to this account, and entirely agree that it is a mistake to look for 'declaratory creeds' before the fourth century. The strategy described is precisely what I have attributed to Basil, a careful incorporation of material which has had a vigorous life in the theology of opponents. But what I should want to do as a refinement of the argument is to point to a further possible source of 'building blocks' in the solemn naming of God and the Logos in Asian and Antiochene liturgy. It is, of course, rash to press any firm conclusions about the wording of the liturgy in the pre-Nicene period; yet there are – as has been argued elsewhere – enough odd and awkward archaisms in our surviving witnesses, enough phrases that could not easily have been freshly and deliberately composed after the fourth century controversies, to give us some indications of primitive theological emphases - the Alexandrian interest in the worship of the heavenly sanctuary (not indeed unique to Alexandria, as AC shows, but given a distinctive colouring there), and the Antiochene and Asian interest in the naming of God so as to reveal the divine power in the elements of the sacrament. How some of these diverse eucharistic concerns later shaped the Christological controversies is a subject often touched upon, but still yielding new possibilities of interpretation.

It would be entirely natural for bishops in synod to turn for help in formulating belief to the language they used at the altar. Indeed, it may even seem surprising that some of the most important formulae – such as the creed of Nicaea itself – reflect relatively little of the doxological idiom we have been examining. It could be – and here I would enter a slight caveat about the Kinzig-Vinzent argument – that the use of variegated scriptural imagery in the language of liturgy simply gives too many hostages to fortune at a time of bitter exegetical conflict, so that the formulae with the most staying power prove to be those with the most modest repertoire of biblical imagery. And meanwhile, within the liturgy itself, corrections, judicious omissions and insertions, qualifying phrases and further scriptural citations are variously deployed by both pro and anti-Nicenes. The outcome in the long run (in the eastern Christian world) was the combination of a fairly minimal creed with a more elaborate eucharistic idiom (it is worth remembering that the Liturgy of Basil was the Byzantine norm for about three centuries, only slowly being displaced by that of Chrysostom); the creed is very sparing indeed with metaphorical and scriptural language ('light from light' is the most obvious exception, an image canonised by long use and helpful to Nicene theology in expressinmg derivation without diminution), and the style of utterance found in Antioch 341 drops out of credal use and finds its definitive place in the eucharistic thanksgiving. And if there is any force to the case here set out, that represents a return home.

I began with a reminder that the relation between creed and worship is not always as simple as we might like it to be. Liturgy has a history, and an enormous part of that history is bound up with the processes of theological discernment that go on in the whole life of the Church. But that being said, we can salvage one point of cardinal importance for understanding how the public proclamation of dogma works. I have suggested a model for credal composition that looks for its primary model in thanksgiving. The pre-Sanctus acclamations spell out exactly why God is worthy of praise, in terms of what he is as well as what he does: doctrine seeks to display a God who in his very essence merits our utmost adoration, both because of his irreducible transcendence – so that we can never speak adequately of him – and because of his unreserved self-giving in the life of the Word and the power of the Spirit – so that we can nonetheless speak truthfully of him. Without that dual acknowledgement, the enterprise of doctrinal formulation becomes sterile and even oppressive. But doctrine as essentially 'eucharistic' language retains its celebratory energy and its capacity to augment our joy.

© Rowan Williams 2004