I

It seems to me, although I shall not argue it here, that even if you regard the New Testament as an infallible source of doctrine, you cannot derive from it a doctrine of the Trinity. Although there are many passages in the New Testament which speak of Christ as divine, and passages which speak of ‘The Spirit’ of God or of Christ and of ‘The Comforter’, there are non-Trinitarian ways of interpreting the latter passages which are just as plausible as interpreting them as expressing the doctrine that the Holy Spirit is a divine person and so entailing a doctrine of the Trinity. So unless Christians today recognize some good a priori argument for a doctrine of the Trinity (and most of them do not recognize such an argument), or unless they consider that the fact that the subsequent Church taught a doctrine of the Trinity is a significant reason for interpreting the relevant passages in a Trinitarian way, it seems to me that most Christians today (that is, those not acquainted with any a priori argument for its truth) would not be justified in believing that doctrine. Those who recognize the Church's authority to teach normally recognize the Nicene Creed promulgated by the First Council of Constantinople as the first binding authoritative statement from which a doctrine of the Trinity can be derived, and indeed the only authoritative statement on this matter binding alike on Catholics, Orthodox, monophysites, Anglicans, and some Protestants.Footnote ¹

The Nicene Creed (Tanner (Reference Tanner1990), 24) affirms belief in one God (θεóς, deus). Then, since it speaks of ‘The Son’, who is ‘begotten of the Father’ as ‘true God from true God’, of one essence (ὁμοούσιον) with ‘The Father’, it explicitly affirms that both the Father and the Son are ‘God’. It goes on to speak of ‘The Spirit’ ‘who proceeds from the Father’, as ‘to be worshipped jointly with the Father and the Son’. Although the Council of Constantinople was unwilling to say so explicitly in the Creed, it follows from the Creed, that since only the divine can properly be worshipped, the Spirit is also ‘God’; and the Council's synodical letter speaks of Father, Son, and Holy Spirit as having a ‘single Godhead, and power, and substance (οὐσία) (ibid., 28). It is normally acknowledged that there is an apparent contradiction here. The Creed claims that there is one God, and also three apparently different beings who are each God. And the contradiction seems more obvious in the contemporary western world when ‘God’ is usually used as a proper name of the (supposedly) one and only being who has the traditional divine properties, and so the word ‘one’ seems redundant. Then ‘the Son is God’ would mean that the Son is the same individual as this being. But in the ancient world where it was open to serious question how many ‘gods’ there are and how powerful they are, θεóς and deus can both be understood either as the name of a unique individual, or as a predicate (‘divine’) which can be predicated of more than one individual. While no doubt the Fathers of the Council did not have a clear view of what was the sense in which there is just one ‘God’ and the sense in which each of the three beings is ‘God’, the distinction between the two senses of the crucial words makes available one obvious way of resolving the apparent contradiction. This is by thinking of these words as having the former sense when the Creed says that there is ‘one God’, and as having the latter sense when it claims that each of the beings is ‘God’. Thus understood, the Creed is saying that there is one unique thing which it names ‘God’, which consists of three beings.

The Creed states that the Father is the originating source of the other two beings – he ‘begets’ the Son, and the Spirit ‘proceeds from’ the Father. The three beings came to be called (in Greek) ὑποστάσεις or πρόσωπα, and (in Latin) substantiae or personae; and what they had in common (in Greek) their οὐσία, (in Latin) their essentia, natura, or substantia. (Confusingly, Latins sometimes used substantia as a translation of οὐσία and sometimes as a translation of ὑπόστασις.) I will call the beings ‘persons’ and what they have in common their ‘essence’. The Council provided no elucidation of what ‘proceeding’ meant, but in saying that the Son was ‘begotten’, it added that he was ‘not made’; and I assume that the latter meant at least that he was not made out of any pre-existing matter, and perhaps also that he was not made out of anything non-existent (see my Appendix). All this is the framework which subsequent ‘theories’ of the Trinity seek to fill out; a theory which is such that there is no ancient sense of the relevant words under which the Creed is true that cannot justifiably be called a Christian theory of the Trinity.

There seems to be further agreement from the fifth century onwards until modern times among all Catholic and Orthodox theologians (and, I would think, most early Protestant theologians) on various further matters which I will now spell out. The first such agreed matter is what is meant by saying that the members of the Trinity are ‘persons’. The fifth-century Christian philosopher Boethius (Reference Boethius and Stewart1968, sec. 3) classically defined a ‘person’ as ‘an individual substance (substantia) of a rational nature’; and he adds that ‘by this definition we Latins have described what the Greeks call ὑπόστασις’ and he claims that the Greek usage is much clearer, since persona is borrowed from a different Greek word πρόσωπον which originally meant ‘mask’, a meaning which Boethius rejects. A ὑπόστασις, Boethius claims, is what ‘underlies’ and ‘supports’ accidental properties, and is best translated into Latin as substantia, an individual thing, not a genus or species. All the Latin mediaevals, including Aquinas (Reference Aquinas1963–1980), Ia.29.1 and ad.1), seem to have accepted this definition as giving a necessary condition for something being a ‘person’. While Latins accepted that there were three individual substances in the sense in which the Greeks were using the term, which they called ‘supposits’ or ‘subsisting things’, they were reluctant to say explicitly that there were three ‘substances’. Aquinas (ibid., Ia.30.1 ad.1) gives as the reason for this, that ‘substantia’ was often used to mean ‘nature’ or ‘essence’, and Latins and Greeks all denied that there were three essences in God. Aquinas (ibid., Ia.29.3) held that – like all words used of God – persona is used in an analogical way, that is as expressing a higher degree of what is attributed to created things. He held, however, that all analogical usage of words is ‘literal’ usage; and would surely have denied that calling any member of the Trinity a ‘person’ was using the word ‘metaphorically’.

The second agreed matter is that each of the persons has the same essential intrinsic properties which make each of them divine. I understand ‘properties’ as including both intrinsic properties which are properties that can be possessed by one substance independently of its relations to any other substance, and relational properties which are properties that relate two or more substances to each other. Having a mass of 10 kg is an intrinsic property; a substance can have a mass of 10 kg, independently of how it is related to other substances. ‘Being taller than’ is a relation which can relate many different pairs of humans to each other; and so ‘being taller than John’ is a relational property which relates its possessor to John. An essential property of a thing is a property which it must have in order to exist; thus ‘occupying space’ is an essential property of my desk. So, to repeat, it was agreed that each of the persons has the same essential intrinsic properties which alone make them divine.

Having exactly the same essential intrinsic properties which make them divine, all theologians assumed, followed from having the same οὐσία, the same (unknowable) essence. But in what sense? Οὐσία, alas, has several meanings, of which two are crucially relevant to the present discussion. Aristotle (1983, 2a) distinguished between οὐσία in the primary sense of an individual thing, a particular man or horse; and οὐσία in the secondary sense of the essence common to all members of a particular species or genus. While fourth-century theologians may have understood οὐσία in different senses and so not have had a clear agreed view about the sense in which the persons of the Trinity are ὁμοούσιος, ‘of the same essence’. But it seems to me clear that many theologians then and almost all theologians after the fifth century understood it in the secondary sense. Thus Basil of Caesarea (1895, Letter 236.6) wrote: ‘the distinction between οὐσία and ὑπόστασις is the same as that between the general and the particular; as, for instance, between the animal and the particular man . . . The Godhead is common, the fatherhood particular.’ In the later Eastern tradition John of Damascus (Reference John of Damascus and Salmond1899, 3.4) said exactly the same: ‘οὐσία signifies the common and general form . . . such as God, man, while ὑπόστασις marks the individual, that is to say Father, Son, Holy Spirit, or Peter, Paul.’Footnote ² But here it might seem that the Western tradition began to diverge from the Eastern tradition, for the former vigorously asserts that the persons of the Trinity have more in common than Peter and Paul. In the thirteenth century Joachim of Fiore was believed to have claimed that the unity of the Godhead is a collective unity ‘in the way that many persons are said to be one people, and many faithful one church’. The Fourth Lateran Council of ad 1215 (Tanner (Reference Tanner1990), 232) rejected his supposed view and declared it to be Catholic doctrine that ‘there exists a certain supreme reality (res)’, ‘that is, divine substance, essence or nature’ which ‘truly is the Father and the Son and the Holy Spirit’. Yet it had earlier said that the Trinity is ‘undivided according to its common essence, but distinct according to the properties of its persons’. So it allows that the persons have distinct properties, but share a common essence. So what is it saying more than Basil said?

What is at stake here is a different understanding of ‘essence’ in the secondary sense between many Eastern and Western theologians.Footnote ³ For many Easterners, an essence of a kind was simply an abstract universal; to say that being an embodied rational animal is the essence of humans would be just to say that to be human is to be an embodied rational animal. But for many Western thinkers an essence of a kind was a concrete thing, located in each of the things of that kind; it was divided among the things of that kind, and to be found in the places where they were. Peter had a bit of the essence of humanity, and Paul had a different bit. But of course each of the members of the Trinity does not occupy places, and so the essence of divinity is not spread among different places. Hence the essence of divinity was not divisible. Hence it was una res fully instantiated in each of the three persons. But no Eastern theologian was going to deny that in that sense the essence of each of the persons was una res. John of Damascus (Reference John of Damascus and Salmond1899, 1.8) wrote that Peter and Paul ‘do not dwell in one another, and are separated’; we recognize ‘by the mind’ that they have a common nature. But the members of the Trinity ‘dwell in each other’, ‘nor can one admit difference in will or judgement or energy, or power, or anything whatsoever which may produce actual and absolute separation in our case. So Lateran IV was not making any theological point with which the East might disagree; it was merely expressing an agreed point in terms of a philosophical category, ‘concrete universal’, which Easterners did not use although they would have agreed with the point made in terms of that category.

The third agreed matter is that the persons differ from each other in their relations of origin, and that this was the only essential difference between them. Thus Boethius (Reference Boethius and Stewart1968, sec. 6): ‘the many-ness of the Trinity is expressed by the category of relation . . . Only terms belonging to relation may be applied singly to each [member of the Trinity]’. The Father was the cause of the existence of the other two, who are distinguished from each other by the Son being ‘begotten’ of the Father, and the Spirit ‘proceeding from’ the Father. But this led to three different views about what the relations were. The first view was that they were simply relations of causation as such; and so the difference between the persons is that the Father is the person who is directly the cause of the Son, and – either jointly with the Son, or through the Son – the cause of the Spirit; the Son is the person who is caused by the Father alone and the Spirit is the person who is caused by the Father either with or through the Son. I cannot see any significant difference between the formula of the Roman Catholic version of the Creed, ‘the Spirit proceeds from the Father and the Son’, and the formula acceptable to many Orthodox:Footnote ⁴ ‘the Spirit proceeds from the Father through the Son’. Both Aquinas (Reference Aquinas1963–1980, 1a.36.3) and Duns Scotus (Reference Scotus and Balic1950–, Ord. 1.26 q1) held that it was a mere verbal difference, and that the latter was an acceptable formula. I find this first view in Gregory of NyssaFootnote ⁵ Aquinas,Footnote ⁶ and other thirteenth-century Dominicans (see Freidman (Reference Friedman2010), ch. 2). The second view, the view of John of DamascusFootnote ⁷ and many modern Orthodox, was that both Son and Spirit were caused by the Father alone. In that case the causal relations of the Father to each would have to be relations of different kinds – ‘begetting’ would have to be a different relation from ‘spirating’ (that is, causing to proceed). The third view, which I find in both Western and Eastern theologians (see Friedman (Reference Friedman2010), ch. 2), combined the other two views. It held that the Son and the Spirit differed from each other both in virtue of their different causal relations to the Father, and because the causal relations themselves were of different kinds. Those later Western mediaeval theologians who advocated the third view tried – following one of Augustine's analogies of the Trinity – to explain the difference in terms of the difference between being produced by unintentional copying (as when we recall some event) and being produced ‘voluntarily’.Footnote ⁸ But what none of the Fathers would advocate is that Son and Spirit differ from each other in virtue of any intrinsic properties, a view advocated by Lossky,Footnote ⁹ for in that case they would not have a common οὐσία.

Because all agreed that the persons were distinct only because of their different relations to each other, it follows that they do not have thisness (in Latin haecceitas, a word invented by Duns Scotus). To say that a thing has thisness is to say that – it is logically possible – there could have existed, instead (or in addition to) that substance, a different thing which has all the same properties as the former thing. Human beings, in my view (see Swinburne (Reference Swinburne2013), esp. ch. 6), have thisness. Instead of me writing this today there could have been another person who has the same body and thinks the same thoughts and had the same (mental and physical) life history. Aquinas and – I suspect – all mediaeval thinkers held that relations are only ever accidental properties of created things; created things are distinct from each other only in virtue of their ‘essential principles’,Footnote ¹⁰ i.e. essential intrinsic properties and/or (in the case of material objects) by their matter, which entails that matter has thisness. So by denying that the Father, the Son, and the Spirit have thisness, as Aquinas would have done if he had had that concept, thinkers would have meant that each is not such that there could have been instead of him a different person with all the same intrinsic properties and all the same relations to other beings as the actual Father, Son, or Spirit. Aquinas (Reference Aquinas1963–1980, 1a.11.3) wrote that ‘it is in virtue of the same fact that [God] is God and is this God’, which implies that God himself does not have thisness, from which it follows that no divine person has thisness; for if a substance does not have thisness, no part of that substance has thisness. In holding that all created things which have the same essential intrinsic properties have thisness I believe that mediaeval thinkers were mistaken. Following most physicists, I believe that (probably) all fundamental particles of a kind (e.g. all electrons) have the same essential intrinsic properties as each other and lack ‘thisness’; they are distinct from each other only in virtue of their relations. An electron is the particular electron it is in virtue of its location, which is a matter of its spatial relations to other fundamental particles (or, if there is an absolute space, its relation to some point of that space), and its temporal relations to past particles. If we accept this, we find less difficulty in understanding Aquinas's claim that ‘with God relations are subsisting’. All that he means by relations being ‘subsistent’ is that relations are part of what makes a thing that thing. So, he claims, what makes an individual divine person that individual person is his kind nature – that is, his divine nature, and his relations to other divine persons. That the persons of the Trinity do not have thisness is important for my later a priori argument for the truth of the doctrine of the Trinity.

Further, all agree that there being three divine persons does not entail that God is ‘compound’ or has three ‘parts’. One reason for this, writes John of Damascus (Reference John of Damascus and Salmond1899, 1.8), is that the persons have the same essence; and Aquinas (Reference Aquinas1963–1980, 1a.42.4. ad.3) also gives as a reason why there is only one God, that ‘the whole fullness of the divine nature is present in each of the persons’.Footnote ¹¹ Another answer given by Anselm in Monologion (Anselm (Reference Anselm, Davies and Evans1998), 79) is that the ‘persons’ are not ‘independently existing things’. John of Damascus (Reference John of Damascus and Salmond1899, 1.8) wrote that ‘compoundness is the beginning of separation’. But such reasons do not rule the persons being ‘parts’ in the sense that together they constitute the Trinity, and – although they would be unwilling to express the point in this way – I suggest that most of the Fathers and mediaevals would have accepted that. Aquinas gave the explanation which I suggested earlier is ‘one obvious way of resolving the apparent contradiction’ that both the Trinity and its members are said to be ‘God’. ‘God’, Aquinas (Reference Aquinas1963–1980, 1a.39.4) wrote, is ‘a name’, yet it means ‘having divinity’ which is ‘an adjective’; and that is why ‘whereas there are three having divinity, that still does not entail there being three gods’. In some cases ‘God’ stands for the ‘essence’; it is by reason of the form signified that ‘God creates’. It is the whole divinity present in the three persons which ‘creates’; thus, it would seem, ‘God’ so used is the name of the three taken together. But in other cases it stands for the person who has divinity, as in ‘God generates’ (that is, God the father generates the Son). So the fact that the Father is God, and also the Son is God, does not, Aquinas claimed, imply that there are two gods, simply because ‘God’ is used in two different ways.

Finally, all patristic and mediaeval thinkers agree, all the actions of each member of the Trinity are actions of the whole Trinity, except the ‘characteristic acts’, which are the acts by which one member causes another member. Omnia acta Trinitatis ad extra indivisa sunt, ‘all the outward-directed acts of the Trinity are indivisible’, that is, each such act is one united act of the whole Trinity. And, wrote Gregory Palamas (Palmer (Reference Palmer1979–), IV, sec. 112), referring to their outward -directed acts, ‘the three divine hypostases mutually coinhere and interpenetrate naturally, totally, eternally, inseparably, and yet without mingling or confusion, so that their energy is also one’. For Palamas, writes John Meyendorff (Reference Meyendorff1974, 216) commenting on this sentence, ‘the Trinity is the union of three hypostases who each keep, in a real way, their personal identity’. The ‘characteristic acts’ of ‘begetting the Son’ and ‘spirating’ the Spirit are not – as Aquinas (Reference Aquinas1963–1980, 1a.41.2) puts it – ‘acts of will’ in the sense of being ‘the source of effects which can be one way or another’; rather, they are ‘acts of nature’ (i.e. it follows from the nature of the relevant person that he will beget or spirate; he has no free choice whether or not to beget or spirate). But they are not consequences of a necessity which acts independently of the person's will. ‘The Father begets the Son willingly’, writes Aquinas. He cites Augustine as claiming that ‘The Father begat the Son neither by will nor by necessity’; this saying repeats the Creed of Damasus (Denzinger (Reference Denzinger1968), no. 3225), that ‘The Father begat the Son, not by will, nor by necessity but by nature.’

So with these precisifications I find in Western mediaeval thought the same doctrine of the Trinity as in Eastern thought, and this theory is a ‘social theory’ in the sense that it postulates three ‘persons’ in somewhat Boethius's sense as ‘distinct centres of knowledge, love, will and action’.Footnote ¹² That the Western doctrine was that of a social theory is made evident especially in the theory of Richard of St Victor, to which I shall come later, to whose content (although not to its provability by natural reasonFootnote ¹³), later theologians do not seem to have objected. So the agreed theory holds that the persons of the Trinity are separate persons, since at least once they perform separate intentional acts – the Father willingly begets the Son; and so must have had separate consciousnesses (though ‘how many consciousnesses?’ was not the kind of question which the Fathers or mediaevals asked). The consciousnesses would be type-identical (that is have the same content), except in respect of the ‘characteristic acts’, but be token-distinct (that is, there are three of them). The three persons have exactly the same divine ‘nature’ (in the secondary sense of οὐσία) as each other, but what makes them separate persons are merely their different relations to each other. I see no evidence of any theory of the Trinity between the fifth century and modern times, recognized as within the bounds of orthodoxy, which is not a ‘social theory’ in this sense. It is true that the ‘models’ of the Trinity which Augustine discusses in the middle chapters of De Trinitate, such as the models of the mind, the mind's love of itself, and the mind's knowledge of itself, and the model of the mind's memory, understanding, and will, are not models of a social theory. For they are all models of one person, and of parts or properties of that person which could not exist without the person. In Boethius's sense of ‘person’, they are theories of one person. But Augustine himself rejected these analogies as totally inadequate in chapter 15 of De Trinitate, on the grounds that each of the persons of the Trinity has love, understanding, etc. quite apart from their relation to the other members. ‘Who’, Augustine (Reference Augustine and Dods1873, 15.7) wrote, ‘would dare to say that the Father understands neither himself, nor the Son, nor the Holy Spirit, except by the Son, or loves them except by the Holy Spirit?’ Augustine, unlike his mediaeval successors, does not seem to me to have a theory of the Trinity.

II

The unanimity of church tradition in the major respects which I have just analysed about the kind of Trinity affirmed by the Creed seems to me good reason for adopting it. But it needs to be spelled out in order to show that it does not contain any inconsistencies, especially in order to show that the claim that each of the persons of the Trinity is divine and so omnipotent does not entail that one person might try to cause some good state of affairs (e.g. Uranus moving in a clockwise direction round the Sun relative to some framework) and at exactly the same time another member might try to cause an incompatible equally good state (e.g. Uranus moving in an anti-clockwise direction). Since both persons could not succeed, both could not be omnipotent, and – if not omnipotent – then surely not divine. To deal with this, we need to consider what it is for a person to be divine.

A divine person is naturally understood as one who is essentially eternally omnipotent and exists (in some sense) ‘necessarily’. If ‘eternal’ is understood as ‘everlasting’, then at each moment of everlasting time, the Father is the cause of the existence of the others at the subsequent moment; if ‘eternal’ is understood as ‘timeless’, then the Father timelessly causes each of the others. (I shall subsequently understand ‘eternal’ as ‘everlasting’, but what I have to say can easily be rephrased in terms of ‘eternal’ being ‘timeless’.) Being divine, each is also ‘omnipotent’. If we understand a person being omnipotent not merely as having the power intentionally to do any logically possible action but also as knowing the nature of the alternative actions between which he can choose, and as being free from non-rational influences, then omnipotence entails perfect goodness, in the following sense. An omnipotent person will know of each available action whether it is good or bad, and whether it is better than some incompatible action. Recognizing an action as good entails having some motivation to do it, and recognizing an action as better than another entails having greater motivation to do it, and recognizing an action as bad entails having motivation not to do it.

So, if there is in some situation a best possible action (or kind of action) for him to do, any omnipotent person will always do that action (or kind of action). But in many situations, for an omnipotent person as also for us, there will be two or more incompatible possible actions (or kinds of action), each as good as each other and better than any other action (or kind of action). In such a situation an omnipotent person will freely choose which of them to do. But surely there will be frequently before an omnipotent person, as there is not before us, a choice between an infinite number of incompatible possible actions (or kinds of action), each of which is less good than some other action (or kind of action), but where there is no best. In such a case his perfect goodness can only consist in choosing to do one of these actions, even though the action the person chooses will not be the best (because there is no best). Thus suppose that it were a best possible kind of action to create planets, and the more of them the better, then an omnipotent being will be perfectly good if he creates exactly one planet, but not a worse person than he would be if he created exactly two planets, although an act of creating planets would be a better act the more planets are created. Now given that any omnipotent person is in this sense perfectly good, there could only be more than one omnipotent person, if each of the omnipotent persons believed that it would be bad for him to bring about effects of certain types, types which only some other omnipotent person had the right to bring about. That is, there must be some agreement among them about which area each is entitled to control. Otherwise we would get the incompatibility which I described. But given that the omnipotence of a person is only a power to do good actions and so actions of a kind which that person had the right to do, more than one person could be omnipotent. Only one person would initiate actions affecting the motions of the planets, and the others would put their power behind his choice; and similarly for other areas. Hence the unity of the Godhead in ‘power’ affirmed by the Council of Constantinople's synodical letter. This demarcation of areas of influence could not be reached by discussion between the persons, for while the discussion was taking place none of them would be omnipotent. In causing the others to exist (at each moment of time) the Father would have to lay down the rules determining who has the right to do which actions; and the other members of the Trinity would recognize his right, as the source of their being to lay them down.

Given my understanding of what it is for each of the persons to be ‘divine’, and given that the Father caused the Son and Spirit, everything else agreed by the traditional social theory follows. In order to know what actions are available to them, each of the three persons would need to know the whole past and present of the world (for everything that has happened in the past or is happening in the present determines whether some otherwise available action is logically possible. For example, in order to know that he cannot now bring about a second world war, an omnipotent being would need to know that a second world war has already happened.) So each divine person would be fully aware of all the thoughts and actions of the others. This common omnipotence, omniscience, and perfect goodness in their community of action makes it the case that in a natural sense there is one God.

Son and Spirit could only exist with the same necessity as the Father, if their causation by the Father was not a voluntary act; and so it would have to follow from the nature of the Father that he would bring about Son and Spirit. It follows that Son and Spirit do not have ‘thisness’; that is, each is not such that there could be instead of him a different person with all the same properties (including relations to other beings) as the actual Son and Spirit. For if for example the Son had thisness, then – since there could have been innumerable possible persons with exactly the same properties as the actual Son, there would have been no reason to cause one to exist rather than another; and so the Father would have had to bring about one of them by a voluntary act; and in that case the actual Son would not exist with the same necessity as the Father. And if Son and Spirit do not have thisness, they would only have the same kind of divinity as the Father, if we suppose that the Father also does not have thisness. And anyway it is simpler and so a more probable true hypothesis to suppose that everything depends on a being who does not have thisness; and so the question does not arise of ‘why this Father, rather than any other?’ Since the three divine persons do not have thisness, they can be distinct only if they have different (intrinsic or relational) properties from each other. But given that it is simpler to suppose that they each have no more intrinsic properties than those following from essential eternal omnipotence, the difference must consist in their having different relations to each other. And inevitably the Father will have a different relation to the other two persons – since he causes them, and they do not cause him. And his doing this must entail his having a certain initial primacy, which puts him in a position to do this. The difference between Son and Spirit could therefore consist in each having a different kind of relation to the Father; and the simplest form of that would be if one (the Son) is caused by the Father alone, and the other (the Spirit) is caused by the Father and/or through the Son. But the difference could also – as we have seen – consist in each being caused by the Father in a different way (signified by the unexplained difference between ‘being begotten’ and ‘proceeding’). Simplicity favours the view that the difference consists solely in their asymmetric relations of origin to the Father – the Son being caused by the Father alone, and the Spirit being caused by the Father and/or through the Son. And I shall shortly give an a priori argument in favour of that view.

So, given that it is coherent to suppose that there is one divine person, understood as a person who is ‘essentially eternally omnipotent and existing necessarily’, I suggest that I have shown that it is coherent to suppose that there are three such persons who form a Trinity of the traditional kind. Given that they are individuated by their relations to each other, it follows that each person could not exist without the others. It is a consequence of the nature of the Father that he will continually sustain in existence Son and Spirit and of the nature of the Son that he will help to sustain the Spirit. I shall now argue that not merely is this coherent, but that there is a strong a priori argument to show that it is true: that necessarily, given one divine person (the Father), there will be two and only two more divine persons.

III

The claim that there is a Trinity of the above kind needs filling out in a crucial respect. In most of the authors whom I have discussed so far there is not much about why it is a good thing that God is triune; yet on a social theory, the answer is obvious – because if there are three divine persons, they can exhibit within the Godhead that characteristic of God which Christians have always regarded as so important – love. Important to mediaeval thought was the Dionysian principle, that goodness by its very nature is diffusive of itself: it seeks to produce more good things. Augustine (Reference Augustine1861, q. 50) wrote with respect to the Father generating an equal ‘if he wished to but could not, he is weak; if he could but did not wish to, he is envious’. In the twelfth century Richard of St Victor echoed this, and – claiming that the fullest kind of love is love of an equal (and so it would not be enough for God to create humans), deduced that the Father must generate the Son.

Perfect love must be love of someone worthy of perfect love, wrote Richard (1958, bk 3.2), and only a divine person could be worthy of perfect love. I wouldn't myself put the point in quite that way. Rather, I would say that perfect love must be fully mutual love, reciprocated in kind and quantity, involving total sharing, the kind of love involved in a perfect marriage; and only a being who could share with him the rule of the universe could fully reciprocate the love of another such. While of course the love of a parent for a child is of immense value, it is not the love of equals; and one thing which makes it as valuable as it is, is that the parent is seeking to make the child (as she grows up) into an equal. To put the point in the terminology I used earlier, it would be a unique best action for the Father to cause the existence of the Son, and so inevitably he would do so. If we think of God as temporal, as Richard of St Victor did not, it follows that the Father could not begin to cause the existence of the Son at some moment in time, say a trillion trillion years ago. That would be too late – for all eternity before that time the Father would not have manifested his perfect goodness. Rather, at each moment of everlasting time the Father must always cause the Son to exist, and so always keep the Son in being.

A twosome can be selfish. A marriage in which husband and wife are interested only in each other and do not seek to spread the love they have for each other is a deficient marriage. (And of course the obvious way, but not the only way, in which they can spread their love is by having children.) Perfect love for a beloved, Richard (ibid., 3.11) argues, must involve the wish that the beloved should be loved by someone else also. Hence it will be a unique best action for the Father to cause the existence of a third divine being whom Father and Son could love and by whom each could be loved. Hence the Holy Spirit. And I suggest that it would be best if the Father included the Son as co-cause (as he is of all other actions of the Father) in causing the Spirit. And again they must have caused the Spirit to exist at each past moment of everlasting time. Hence the Trinity must always have existed.

So why only three divine persons? Do not these arguments suggest that there should be more than three divine persons? Richard of St. Victor's answer (ibid., 5.15) is that there would be no need for a fourth since three divine persons would satisfy the demands of love fully. I argued earlier that when there is a unique best action a divine person must do it. It was a unique best action for the Father to bring about another divine person – for perfect love requires another with whom to share; and it was a unique best action for Father and Son to bring about a third divine person, for the love of Father for Son and Son for Father required that they should cause there to be some divine person other than themselves for the other to love and be loved by. Three persons is the necessary minimum for unselfish love between persons of some kind. Every member of the Trinity could show unselfish love without there being a fourth divine person.

However, Richard did not consider an objection that even so, surely the more divine persons the better, and so surely good divine persons would produce lots more divine persons; and hence he has not proved that God must be triune. His argument needs a crucial step, which I now add, as follows: if the objection were correct, then however many divine persons the Father (in conjunction with others) brought about, it would be still better if he brought about more. But, as I argued earlier, when there is an infinite series of incompatible possible good actions, each better than the previous one, available to some agent, it is not logically possible that he do the best one – because there is no best action. An agent is perfectly good in that situation if he does any one of those good actions. So since to bring about only three divine persons would be incompatible with an alternative action of bringing about only four divine persons, and so generally, the perfect goodness of the Father would be satisfied by his bringing about only two further divine persons. He does not have to bring about a fourth divine person in order to fulfil his divine nature. To create a fourth divine person would therefore be an act of will, not an act of nature. But then any fourth divine person would not exist necessarily in the sense in which the second and third divine persons exist necessarily – his existence would not be a necessary consequence of the existence of a necessary being; and hence he would not be divine. So there cannot be a fourth divine person. There must be and can only be three divine persons.Footnote ¹⁴

IV

Among modern writers Jürgen Moltmann (Reference Moltmann1981), Cornelius Plantinga (Reference Plantinga, Feenstra and Plantinga1989), William Hasker (Reference Hasker2013), and John Zizioulas (Reference Ziziolilas2002) have all advocated versions of a social theory. I find myself in agreement with the first two of these authors, and largely in agreement with Hasker. But whether I agree with Zizioulas's ‘social Trinitarianism’ depends on just what he is claiming. He claims (Zizioulas (Reference Ziziolilas2002), 41) that the Father ‘freely’ begets the Son and brings forth the Spirit. If ‘freely’ entails that these acts are acts of will, and not acts of essence, then that is a long way out of line with Christian tradition and to be rejected also on the ground that Son and Spirit would not then exist as necessarily as does the Father. But ‘freely’, which in this sentence he explains as ‘out of love’, may just mean ‘willingly’ or ‘gladly’. But he also claims in the same place that the Father ‘perpetually confirms “through being” his free will to exist’,Footnote ¹⁵ which seems to imply that he could commit suicide – which, if I have interpreted him correctly, would make him even further out of line with Christian tradition.

The ‘social theory’ is sometimes called the ‘Greek theory’ in contrast to the ‘Latin theory’. But if I am right, there was no other theory from the fifth century until modern times, above all because they all have roughly the same understanding of what a ‘person’ is; and agree that the members of the Trinity are different persons from each other, and that the divine essence or the Trinity itself are not persons. It is only in respect of the less central issue of the filioque that there are differences. Only when we come to modern times do we find, among theologians not officially deemed heretical, theories which are in Boethius's sense one-person theories, and in which the Trinity of ‘persons’ is a Trinity of ‘persons’ in a quite different sense. I'll call them unitary theories. Barth and Rahner are both unitary theorists. Barth (Reference Barth and Thomson1936, 412–413) thinks of the persons of the Trinity as three ‘modes of existence’ of the one God, and Rahner (Reference Rahner1970, 109) thinks of them as God's three ‘distinct manners of subsisting’. For Rahner, the Trinity is simply ‘The threefold quality of God in himself’, his triune ‘personality’.Footnote ¹⁶ These theories all seem to me to be ‘modalist’ theories; but to differ from Sabellianism, which was always deemed heretical, in that they claim that the threefoldness of God is not merely a feature of the ways in which humans become aware of God – as creator, as redeemer (in Jesus Christ), and as sanctifier (in the ‘Holy Spirit’) – but is a necessary division in his operations which lies behind these contingent manifestations. Among modern analytic philosophers I would classify Brian Leftow (Reference Leftow2004),Footnote ¹⁷ William Lane Craig (Reference Craig, McCall and Rea2009),Footnote ¹⁸ and Keith Ward (Reference Ward2015)Footnote ¹⁹ as unitary theorists.

The two primary positive reasons given in favour of unitary theories are the reason that it provides the most natural interpretation of the Christian monotheistic tradition, and that it alone safeguards the ‘simplicity of God’. I have argued that on the contrary that social Trinitarianism provides the obvious natural interpretation of the Nicene Creed, and the way in which all theologians not deemed heretical developed it in the following 1,000 years. And as regards the claim that a unitary theory alone safeguards the ‘simplicity of God’, note that simple theories may have complicated consequences; Newton's simple theory of gravitation has very complicated consequences for how many bodies interact, but it is no less simple for having such consequences. Likewise the simplicity of the claim that there is one divine person is not diminished if it follows from the existence of such a person that he would produce two other divine persons, and that is what I, following Richard of St Victor, have argued. On the other hand, no unitary theory seems to give a satisfactory account of why there should be three rather than one or any other number of ‘modes of being’ or ‘persons’ in any other highly metaphorical sense, let alone anything amounting to a proof that there must be such. I conclude that there is nothing in favour of a unitary theory, and everything in favour of a social theory.Footnote ²⁰