Why exactly three divine persons?

Every account of God as Trinity has to answer a question that sounds childish and turns out to be very hard: why three? Not why more than one — that has answers. Not why not infinitely many — that has answers too. Why exactly three, rather than two, or four, or seven?

The question is old. It was put to the fourth-century Cappadocian fathers, and it is put on YouTube livestreams today. What follows is an account of the best answers that exist, assembled from the primary texts rather than from apologetic restatements of them, together with an honest report of where each one stops.

The short version: the best argument available is considerably better than its popular versions suggest, and it fails for a reason that is now precise enough to state in a sentence. It proves that relating has exactly three roles in it — a source, something derived from that source, and the rejoining of the two. It does not prove that three roles require three distinct someones to fill them. That gap is not a detail. It is the whole question.

The question, stated exactly

The demand has four parts, and an argument has to meet all of them.

That last row is the one that decides whether an answer counts as an argument at all. A Christian is perfectly entitled to say that the number is revealed and not derived — a great many have said exactly that, and the last section of this page takes that answer seriously on its own terms. But it is a different kind of answer, and conflating the two is how the question gets to look settled when it is not.

One more distinction matters throughout. Ruling out four is much harder than ruling out one, and almost every attempt quietly stops after the easy half. A reason for thinking a solitary God insufficient — that love needs an object, that a mind needs something to know — is not yet a reason for thinking a fourth person impossible. Watch for the switch.

The oldest answer: a movement that comes to rest

The canonical Christian formulation is a sentence of Gregory of Nazianzus, the fourth-century bishop the Eastern church calls “the Theologian.” It is quoted constantly, and it is worth reading exactly.

Because of this, the Monad “from the arche” when it moves into a Dyad comes to rest at a Triad. And for us, this is the Father and Son and Holy Spirit.¹

Maximus the Confessor, writing three centuries later in a commentary on that very passage, restates it and adds the crucial negations:

For this reason the Monad from the beginning moved toward a dyad and at the Trinity came to a halt… since the Godhead is a Monad (but not a dyad), and a Trinity (but not a multitude)… The Trinity is truly a Monad, for such it is; and the Monad is truly a Trinity, for as such it subsists.²

This is the source-family that almost every later “monad, dyad, triad” gesture is reaching for, usually without citation. Notice what it is and is not. It is a description: the movement comes to rest at three. It says where the movement stops. It does not say why it could not have gone further. Maximus’s “but not a multitude” is a denial, not a demonstration.

That is not a criticism of Gregory or Maximus, neither of whom was trying to supply a proof. It is a criticism of the modern habit of citing them as though they had.

A living proponent, twice interrupted

There is a reason to start with a livestream rather than a monograph. It shows, better than any argument summary can, exactly where the question currently sits among the people most committed to answering it.

On 27 July 2026, Fr. Deacon Dr. Ananias Sorem — an Orthodox deacon with a philosophy doctorate, one of the more capable presuppositional apologists working in English — hosted a two-hour-forty-three-minute question-and-answer livestream.³ A viewer asked, in the chat: “I am very confused on how can I prove the trinity. I watch jays video but it’s very difficult.”

Sorem’s answer was methodological, and it was good. Proof, he said, does not run from the parts to the whole. (“Tag” in the transcript below is TAG, the transcendental argument for the existence of God — the argument that certain things must already be true of God for reasoning to be possible at all.)

I don’t understand how tag proves the trinity — tag does not. You’re backwards. You cannot use tag or any argument to prove any part. Why? Because tags about the whole. So you don’t move to prove the whole. You don’t go this way. You don’t go from parts to whole.³

And, a minute later:

No, you look, you get a standard, you get a criteria, you get a whole then proves the parts. So once you prove orthodoxy, that’s how you know the trinity. You don’t prove the trinity and then ecumenical councils and then I get some of this…³

That is a clean statement of a real position — you argue for the whole framework, and the doctrines come with it — and it is the strongest material in the segment. But something else happens in the same six minutes, and it is the reason this stream is worth reporting. Sorem begins the exact sentence this page is about, and breaks off:

It’s the system. It’s the whole… that’s proved. So how do you prove the trinity? By proving the whole. What does the whole say? Trinity. That’s how you prove. I mean there’s other ways too but there’s not some natural theology independent argument of how you prove three over well like — I mean there is I mean I could go through it but you guys will get confused. You’re not going to get the it’s too difficult.³

Five minutes later he declines a second time:

But there is another argument, but for some reason people I just people can’t get it. So, I’m not going to go over it cuz they don’t understand what monad, diad, and triad mean for some reason, and they get super confused. So, just stick with that. That’s an easier argument.³

So: a serious proponent, on the record, says twice that an argument from monad, dyad and triad to exactly three exists, and twice declines to state it. Across the whole stream the words hypostasis, essence and filioque are never spoken once; the only cardinality-relevant claim offered is that “a monad doesn’t refer to anything,” which even at maximum strength is an argument against one, not against four.

A search of Sorem’s own published corpus — essays, papers, podcast appearances, social posts — did not locate the argument stated in full anywhere. The most likely explanation is not that he is concealing something, but that the argument he has in mind is the patristic movement-formula above, which is a description rather than a derivation, plus a piece of Neoplatonic number theory that the rest of this page examines. He is the fifth independent Orthodox source in this line of research — after Richard Swinburne, Beau Branson, Joshua Sijuwade and Jonathan De Young — to leave “exactly three” stipulated or revealed rather than derived.

The strongest version anyone has built

Since no proponent has written the argument out, the honest thing to do is build it for them, at maximum strength, from the sources they are drawing on. That is what this section does. The result is materially stronger than any popular version, and it deserves to be met at that strength.

The classical premise

The taproot is Aristotle, and it is not about God at all.

A magnitude if divisible one way is a line, if two ways a surface, and if three a body. Beyond these there is no other magnitude, because the three dimensions are all that there are, and that which is divisible in three directions is divisible in all. For, as the Pythagoreans say, the world and all that is in it is determined by the number three, since beginning and middle and end give the number of an “all”, and the number they give is the triad. … Further, we use the terms in practice in this way. Of two things, or men, we say “both”, but not “all”: three is the first number to which the term “all” has been appropriated.⁴

Two things carry the weight there: that there is no fourth spatial dimension, so “three ways” exhausts “every way”; and that ordinary language reserves “all” for three and up. Both are offered in support of a conclusion about complete magnitudes. Neither is about how many gods there are.

The formal engine

The strongest formal version is Proclus, the fifth-century Neoplatonist, and this is a genuine argument rather than a gesture:

Every effect remains in its cause, proceeds from it, and reverts upon it. For if it should remain without procession or reversion, it would be without distinction from, and therefore identical with, its cause, since distinction implies procession. And if it should proceed without reversion or immanence, it would be without conjunction or sympathy with its cause, since it would have no communication with it. And if it should revert without immanence or procession, how can that which has not received its being from the higher revert existentially upon a principle thus alien?⁵

Proclus is not counting. He is running an exhaustive case analysis over the possible combinations of three structural roles — abiding, proceeding, returning — and eliminating every incomplete subset. A companion proposition supplies the closure mechanism: the return conjoins the end to the beginning, so the structure is a cycle, and a cycle needs no fourth term to terminate. It terminates by arriving back where it started.⁶

The argument, assembled

Call it Argument E. Two definitions first. A role here is a position in a structure, counted as distinct only if no combination of the other positions yields it. A structure is complete when it contains every role it needs in order to be what it is without depending on anything outside itself.

1. Any internally related plurality has a source — some term not derived from another within the structure. Otherwise you have an infinite regress or a circle of mutual derivation, and neither is a first principle.
2. A source alone is not a plurality, so real distinction requires at least one derived term. That rules out one.
3. A source plus one derived term does not close. It is bare opposition: two things standing apart, with nothing in virtue of which the derived is rejoined to its source. That rules out two.
4. A third term, occupying the rejoining role, closes the structure. End meets beginning; the structure becomes a cycle and needs nothing further.
5. A fourth term introduces no new role. It must occupy source, procession or return, all three of which are taken. It adds a member, never a kind. This is what the tradition means when it says numbers past three are “repetition.”
6. In God there is nothing redundant. No potency, no unrealised capacity, nothing merely permitted-but-unrequired. What is in God is there because the divine life requires it.
7. In God each role must be a someone. Divine simplicity forbids a real distinction that is not a subsistent bearer — there is no substrate in which a mere “aspect” could inhere. So role-count and person-count coincide.

Conclusion: exactly three divine persons.

That is a real argument. It is valid in form, it rules out one and two by mechanism rather than by assertion, it has an actual anti-four step, and step 7 — the step where most versions cheat — is defended rather than smuggled. Three hard objections can be answered from inside it.

Why not a return of the return? Because returning has nowhere further to go. Reversion is defined as reversion upon the source, and the source is unique within the structure; a return-upon-return is a redescription of the same reversion, not a new role. The cycle closes by identity of terminus with origin. This is the best thing in the argument.

Why exactly three roles rather than some other number? Because the roles are generated by a single two-place primitive — derived-from — plus its completion: the term it starts at, the term it ends at, and the act of joining them. There is no logically prior fourth notion in a derivation.

Does this count persons or only kinds of relation? The proponent concedes that steps 1–5 concern roles, and defends step 7 as the bridge: in a simple being there can be no real relational distinction lacking a subsistent term, because there is no substrate for a mere aspect to inhere in. So in the divine case, and only there, role-count and person-count coincide.

Where the strongest version breaks

Four independent failures. Any one of them would be serious; together they close the line.

Its historical premise is false in its own source

The claim that numbers past three are “only repetitions, adding no new principle” is usually credited to Neoplatonic number theory, by way of Eustratius of Nicaea, a twelfth-century Byzantine bishop who built an anti-filioque triadology on pseudo-Iamblichean arithmetic.⁷

The credit is misplaced twice over. First, Eustratius’s own surviving sentence says nothing of the kind. Here it is, in the specialist’s translation:

According to the nature, when the Monad is going forth by itself, it normally makes its proceeding to the first Dyad, because it would not step forward into generation of the numbers without previously having the Dyad appeared from it.⁸

That is a lower-bound claim — a source cannot generate numbers at all without a pair appearing from it first — with no upper bound in it anywhere. There is no “and it stops here” clause. Read strictly, it licenses four, five and six rather than excluding them. Eustratius’s actual argument is a symmetry argument against the filioque: Son and Spirit must be perfectly co-ordinate, or they cease to form a pair distinct from the Father’s unity.

Second, the number theory he drew on positively contradicts the closure claim. The pseudo-Iamblichean Theologumena arithmeticae is organised as a treatment of ten distinct numerical principles, one through ten, each with its own theological profile. The number four is not a repeated pair in that book:

The tetrad, again, is a very great marvel to them, another God, Many-God, or rather All-God. For to them, it is the fount of physical perfections, and the key-holder of Nature. But it is also the forerunner and cause of the constitution and stability in accordance with mathematics.⁹

In the wider Pythagorean tradition the completion-figures are the tetraktys (four) and the decad (ten), not the triad. A proponent may not appeal to this tradition for the closure premise, because this tradition denies it.

“Adds no new kind” is not “cannot exist”

Grant step 5 at its strongest anyway. Grant that a fourth term instantiates no new role. It does not follow that a fourth term cannot exist, and nothing in the argument closes that gap.

Every composite number factors into primes, and composite numbers exist. Every polygon can be cut into triangles, and quadrilaterals exist. “Decomposes without remainder into the three” and “is impossible” are simply different predicates.

The only bridge offered is step 6, divine non-redundancy, and it fails twice. Parsimony is a norm on explanations, not a modal constraint on beings; “God contains nothing redundant” is a substantive theological premise, exactly as contestable as the conclusion it is meant to secure. And it cuts symmetrically: the intuition running the other way — divine fecundity, the very rationale Gregory of Nazianzus uses against a solitary God — favours more rather than fewer. You cannot invoke fecundity to get past one and parsimony to stop at three without an independent principle telling you when to switch. No such principle is on offer.

Run the argument on its own author

Proclus produced the strongest formulation of the triadic-completeness claim. What did he conclude from it?

He held that every effect abides, proceeds and reverts — so his system is triadic at every level, with triads nested inside triads without limit. And above and beyond that triadic structure he posited an indefinite plurality of henads: divine individuals, uncounted.

This is a control result, and it is decisive. The strongest known version of the triadic-completeness argument, in the hands of the man who devised it, yields unboundedly many triads and unboundedly many gods. Whatever it establishes, it is not a count of persons — and we know this not by philosophical suspicion but by inspecting what its own author took it to show.

The dilemma at step 7

This is where the argument dies, and it dies by dilemma rather than by a missing premise.

Proclus’s three are moments of a single causal relation. One effect, related to one cause, exhibits all three: it abides in, proceeds from, and reverts upon that cause. Two entities, three moments. So:

• Read the three as moments. Then step 4’s closure argument goes through, but the “three” are three aspects of one process. Identifying them with persons is straightforwardly modalist — the third moment just is the first returning to itself, so the third “person” is the first under another description. That is Sabellianism, condemned by the tradition the argument is meant to serve.
• Read the three as subsistent someones. Then step 7 is satisfied, but step 4’s closure argument no longer applies, because it was an argument about the moments of a process, not about three things. Closure at three would have to be re-established for persons, and it is not — not by Eustratius, not by Proclus, not by any source located in this research.

Valid on a reading its proponents must reject as heresy; orthodox on a reading where the closure step is missing. That is a structural defect, not a repairable gap.

Peirce: the premise proved, the inference refuted

Here the story takes an unexpected turn, and it is the most interesting thing on this page.

The structural half of Argument E — the claim that three is a closed and irreducible inventory of relational kinds — is not superstition. It is very nearly a theorem, proved in modern logic, and nobody in the Orthodox or presuppositional literature surveyed here has ever cited it.

Charles Sanders Peirce’s Reduction Thesis has two clauses. Necessity: some three-place relations cannot be built out of one- and two-place relations alone. Sufficiency: one-, two- and three-place relations suffice to generate relations of any number of places whatever. Robert Burch proved both in 1991 within a Peircean algebra of logic; Hereth Correia and Pöschel later removed a restriction in the proof.¹⁰ Peirce’s own statement of the intuition is the one Proclus was groping at: bringing two things together requires a third, and “triadic relation cannot be reduced without a use of triadic relation.”

A 2023 result comes closer still to the tradition’s claim. Defining the ternarity of a relation as the minimum number of three-place relations needed to build it, the authors prove that for a non-degenerate n-place relation on an infinite domain, ternarity is always n − 2: one unit of three-ness per unit of arity above two.¹¹ A seven-place relation is exactly five triads’ worth of relating. More triads; no new kind of relating. The Neoplatonists’ structural intuition was, on this narrow point, correct — and provable.

And then the same formalism kills the theological inference, three times over.

The flagship irreducible triad is satisfied by one object. The relation that carries the whole reduction thesis is teridentity: the three-place relation that holds of ⟨a, a, a⟩ and nothing else. It is irreducibly three-placed, and it is a relation a single individual bears to itself three times over. If irreducible three-ness were readable off into a count of persons, the flagship case would yield one person, not three. This is a formal demonstration that number of places does not count individuals.

The sufficiency clause affirms the tetrad rather than forbidding it. The thesis says four-place relations are constructible from triads. That is an existence claim. Nothing in it says a four-place relation is redundant, unreal or excluded. The objection two sections above is not merely intuitive; it is what the theorem literally states.

Places are not occupants. A three-place relation may hold among three individuals, or four, or infinitely many in its field, and may hold of one individual repeatedly. The number of argument-places and the number of relata are independent quantities.

Two honest caveats about the thesis itself. On infinite domains, a 1915 result of Löwenheim shows that every relation of three or more places reduces to a combination of two-place relations; the reduction thesis is true only under Peirce’s more restrictive notion of composition, which is why the thesis was, in one survey’s phrase, all but universally rejected for most of the twentieth century. And there is a live objection that the base operations were chosen to force the result, answered in 2022 by a strengthened formulation but not settled.¹²

There is also a literature connecting Peirce to the Trinity, and it is worth knowing what it claims. Andrew Robinson maps Firstness to the Father, Secondness to the Son, Thirdness to the Spirit, and proposes that the sign-processes fundamental to life are “vestiges of the Trinity in creation.”¹³ The vocabulary is decisive: a vestige is something identified in creation after the Trinity is known by revelation. Robinson offers a model that makes the doctrine coherent and fruitful. He does not offer a proof that God must be three, and does not claim to.

The crux, stated exactly

Everything above converges on a single sentence, and stating it precisely is worth more than another round of argument.

Let R be a relation. Write slots(R) for its argument-places, and occ(c) for the set of individuals occupying those places in some particular completion c of the relation.

• What every argument on file proves: |slots(R)| = 3.
• What the doctrine needs: |occ(c)| = 3.
• The bridge required: |occ(c)| = |slots(R)|.

That bridge is not an analytic truth and not a theorem. It is a substantive further claim, and in the general case it is false — and the leading contemporary theory of argument-places says so as an axiom rather than as a concession.

Slot theory, the main realist account of argument-places, holds that a universal is n-adic exactly when it contains n slots.¹⁴ That is precisely the notion needed here. Its occupation axioms settle the question directly:

The first axiom is a precise statement of what sets slot theory apart from pocket theory: the claim that each slot may be occupied by at most one entity in a given partial completion. The second ensures that a single object may occupy more than one slot in a property or relation in a given partial completion. This is to accommodate the fact that a single object can occupy more than one slot in a property or relation in a given partial completion.¹⁵

The illustrating examples are a is exactly as tall as a and a is taller than a. The axiom is not wrung from the theory by an objector; it is built in from the start, and the theory’s main rival keeps it while dropping the other one. Both leading accounts of argument-places agree on exactly the point a slot-counting Trinitarian argument needs denied.

So Peirce’s teridentity is not an isolated logician’s curiosity. It is the general case as the metaphysics of relations understands it. Any argument that needs slot-count to equal occupant-count is arguing against a settled axiom of the relevant field, and owes a reason.

Philosophy of mathematics discovered the same gap independently, from the other side. Ante rem structuralism — the view that mathematical objects just are places in structures, the purest possible version of “the slots are the individuals” — has been in crisis over this for twenty-five years. Jukka Keränen’s identity problem observes that structures typically contain positions that are indiscernible within the structure, and that every account of place-identity available to the structuralist then forces those positions to be identical: the theory says 1 = −1 in the additive group of integers.¹⁶ Two independent disciplines, no shared stake, converging on the same finding: structure alone does not individuate its occupants, and the default failure mode is collapse — which is the direction of modalism, not of three distinct persons.

That last point yields a genuinely new reading of a twelfth-century argument. Eustratius’s diagram placed the Father at the apex of an isosceles triangle and the Son and Spirit at the two base vertices — perfectly symmetrical, which was the whole anti-filioque point. But that structure has a non-trivial automorphism: the reflection swapping the two base vertices. On structuralist criteria the two base positions are therefore indiscernible and must be identified, collapsing the triad into a pair. Eustratius’s own symmetry requirement is what creates the problem. And Nicholas of Methone’s objection to him — that Son and Spirit are not from the Father in the same way, one begotten and one proceeding — turns out to be the structurally correct repair. In modern terms, Methone was breaking the automorphism to make the positions discernible. He was right, and for a better reason than he could have known.

The bridge that does exist, and it is not about counting

There is a real principle that forces distinct occupants. It has nothing to do with how many places a relation has. It is about irreflexivity — a relation nothing can bear to itself.

…in cases like the above examples irreflexive relations are instantiated: relations entities cannot bear to themselves. This irreflexivity is the key to proving that (a generalized version of) [the identity of indiscernibles] is satisfied after all: if an entity stands in a relation that it cannot have to itself, there must be at least two entities.¹⁷

This is Quine’s “third grade of discriminability,” used by Simon Saunders to rescue the identity of indiscernibles from Max Black’s two-spheres puzzle and to distinguish otherwise identical fermions. Stated as the bridge this case would need: if a relation is irreflexive across each pair of its places, then any completion of it has exactly as many distinct occupants as places.

That principle is valid — near-trivially so. It is also exactly what the arity arguments cannot deliver, because irreflexivity is a fact about which tuples a relation holds of, not about how many places it has. Teridentity is the maximally reflexive three-place relation; a maximally irreflexive one has the same arity and the opposite verdict. Arity does not constrain reflexivity in either direction.

And the theological bridges that actually work in the tradition are all of exactly this kind. Every one of them is a content-level claim about what a particular relation is. None counts places.

Augustine states it as a flat axiom: nulla omnino res est quae se ipsam gignat ut sit —

…whoever thinks of God as having such power as to have generated Himself, is so much the more in error, because not only does God not so exist, but neither does the spiritual nor the bodily creature; for there is nothing whatever that generates its own existence.¹⁸

Nothing begets itself. That is irreflexivity applied to one specific relation, and Augustine offers no argument for it; he treats it as self-evident and moves on. Aquinas later cites it in exactly that axiomatic role.

Aquinas meets the general principle head-on. He records the objection in almost the words used above — “every real relation requires two really distinct extremes”¹⁹ — and, rather than rejecting it, accepts it and denies that the divine relations are self-relations. A thing’s ordering to itself under two aspects, he says, is a relation of reason only, not a real relation.²⁰ That category — relatio rationis, the order of a thing to itself under two descriptions — is Peirce’s teridentity, thirteen centuries early. The scholastic tradition sees the problem, has vocabulary for it, and resolves it by a substantive claim about which relations are real.

Then comes the datum that settles the matter. Aquinas counts four real relations in God — paternity, filiation, active spiration, procession — and three persons. He explains the mismatch explicitly:

Now the relations cannot distinguish the persons except forasmuch as they are opposite relations; which appears from the fact that the Father has two relations, by one of which He is related to the Son, and by the other to the Holy Spirit; but these are not opposite relations, and therefore they do not make two persons, but belong only to the one person of the Father.²¹

The most rigorous Trinitarian theology on offer explicitly denies that relation-count equals person-count. Aquinas gets three from four. His bridge is not slot-counting; it is opposition, a pairwise content-level condition on which relations can be borne by one and the same subject. Anyone reaching for “three slots, therefore three persons” is not using the tradition’s best apparatus. He is using something the tradition’s best apparatus contradicts.

That bridge has costs, and they should be stated. It presupposes what it individuates, since the persons are constituted by the very relations said to distinguish them. It is confessionally loaded: relationis oppositio is the Florentine formula deployed to license the filioque, and Aquinas’s argument here is an argument for the filioque, so an Eastern Orthodox proponent cannot adopt it without either accepting that or supplying a different account of what opposes Son to Spirit. And it never yields a cardinality on its own — getting to exactly three still needs the separate claim that there are exactly two processions, which Aquinas grounds in intellect and will, a psychological analogy that Orthodoxy has generally declined.

Richard of St Victor offers the strongest positive bridge located anywhere, because it is the only one that supplies irreflexivity and a cardinality in the same argument. Charity, he argues, is essentially other-directed: “no one is properly said to have charity on the basis of his own private love of himself.” That gets past one. Then:

When one person gives love to another and he alone loves only the other, there certainly is love but it is not a shared love. When two love each other mutually and give to each other the affection of supreme longing… there certainly is love on both sides, but it is not shared love. Shared love is properly said to exist when a third person is loved by two persons harmoniously and in community, and the affection of the two persons is fused into one affection by the flame of love for a third. From these things it is evident that shared love would have no place in Divinity itself if a third person were lacking to the other two persons.²²

Richard is the only figure in this entire search whose argument has the right shape: an irreflexivity premise grounded in the nature of a specific relation, a lower bound at two, a distinct further argument to three, and an explicit upper-bound argument against four.²³ Swinburne’s modern perfect-love argument is a descendant of it, and the published critique of Swinburne’s version targets precisely the step from three to no-more-than-three. Note carefully what Richard is not doing: he is not counting argument-places. His bridge is a claim about what charity is.

The reference literature concedes the state of play. The Internet Encyclopedia of Philosophy states the irreflexivity premise plainly — “the Father begets the Son, but the Son does not beget the Son” — and then, of the project of justifying it, says it is “at best speculative.”²⁴ That is this case’s crux, conceded by a neutral encyclopedia.

One failure in eight costumes

Step back from the detail and something worth pausing on comes into view. The positions surveyed above were produced in six different disciplines, by people who in most cases had never heard of one another, over roughly fifteen hundred years. A fifth-century Athenian Neoplatonist, a twelfth-century Byzantine bishop, two medieval theologians, an American logician, an Oxford philosopher of religion, a philosopher of mathematics, and two contemporary metaphysicians of relations. They were not answering the same question and most of them were not interested in each other’s field.

They arrive at the same place. Every one of them runs into the fact that the number of places in a relationship does not fix the number of individuals occupying them — and the ones who get past it get past it in exactly one way, by saying something about what a particular relation is rather than by counting anything at all.

They all arrive here

An inventory of kinds is not a count of individuals

|slots(R)| = 3 does not give |occ(c)| = 3

c. 480Neoplatonic metaphysics

Proclus

A derivation has exactly three stages — remaining, procession, return — and the return closes the circuit, so no fourth is needed.

Collides

early 1100sByzantine arithmology

Eustratius of Nicaea

The Son and the Spirit must be perfectly co-ordinate: an isosceles triangle with the Father at the apex.

Collides

before 1173Medieval theology

Richard of St Victor

Charity cannot be self-directed; mutual love between two is not shared love; shared love needs a third.

Routes around

c. 1268Scholastic theology

Thomas Aquinas

Four real relations in God and three persons: relations distinguish persons only where they stand opposed.

Routes around

c. 1903Formal logic

Charles Sanders Peirce

Three-place relations cannot be built from two-place ones, and anything larger can be built from three-place ones.

Collides

1994Analytic theology

Richard Swinburne

A perfectly loving God must share love, and shared love requires a third person to be loved by two.

Routes around

2001Philosophy of mathematics

Jukka Keränen

Structures routinely contain positions that nothing inside the structure can tell apart.

Collides

2013–2018Metaphysics of relations

Gilmore and Dixon

A universal is three-placed exactly when there are three slots in it. Argument-places are real entities.

Collides

Press enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.

Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Collides — establishes a shape, and stopsRoutes around — answers with what a relation isClockwise from the top is chronological.

1. c. 480

   Neoplatonic metaphysics

   Proclus

   A derivation has exactly three stages — remaining, procession, return — and the return closes the circuit, so no fourth is needed.

   CollidesThree stages of one process. The case analysis never asks how many individuals run through them.

2. early 1100s

   Byzantine arithmology

   Eustratius of Nicaea

   The Son and the Spirit must be perfectly co-ordinate: an isosceles triangle with the Father at the apex.

   CollidesThe symmetry the argument needs is a symmetry that makes the two base positions indiscernible — and indiscernible positions collapse.

3. before 1173

   Medieval theology

   Richard of St Victor

   Charity cannot be self-directed; mutual love between two is not shared love; shared love needs a third.

   Routes aroundRoutes around it. The premise is an irreflexive relation, not a count of places. He is exposed on the ascent instead — why a pair must be better than one.

4. c. 1268

   Scholastic theology

   Thomas Aquinas

   Four real relations in God and three persons: relations distinguish persons only where they stand opposed.

   Routes aroundRoutes around it, and denies it outright. The tradition's most rigorous exponent gets three from four and says why.

5. c. 1903

   Formal logic

   Charles Sanders Peirce

   Three-place relations cannot be built from two-place ones, and anything larger can be built from three-place ones.

   CollidesHis own teridentity is an irreducible triad with exactly one occupant. The proof and the counterexample share an author.

6. 1994

   Analytic theology

   Richard Swinburne

   A perfectly loving God must share love, and shared love requires a third person to be loved by two.

   Routes aroundRoutes around it by the same irreflexivity as Richard — then stops at three with a silence rather than a barrier.

7. 2001

   Philosophy of mathematics

   Jukka Keränen

   Structures routinely contain positions that nothing inside the structure can tell apart.

   CollidesEvery structuralist account of identity then makes those positions the same thing — so the theory says 1 = −1. Structure collapses occupants; it does not multiply them.

8. 2013–2018

   Metaphysics of relations

   Gilmore and Dixon

   A universal is three-placed exactly when there are three slots in it. Argument-places are real entities.

   CollidesThe theory's own axioms let one object occupy more than one slot at once. The field states the denial as a starting assumption.

What relative identity does and does not do

A natural thought is that relative-identity theories close the gap: “the Father is the same being as the Son but not the same person as the Son” looks like a formal way of requiring distinct occupants without landing in tritheism.

It does not close the gap, and the reason is structural rather than a matter of the theory’s known costs. Relative identity is a permission, not a requirement. Peter Geach’s and Peter van Inwagen’s systems show that the sentence above can be stated without contradiction. That removes an obstacle to counting three. It supplies no pressure whatever toward three: a relative-identity system in which the “not the same person as” clause is simply never asserted is equally consistent, and describes a God of one person. Relative identity answers “can you coherently say it?” This case asks “must it be so?”

Worse, on one of its two main formulations it lands on the collapse side. Michael Rea’s objection to the pure Geachean strategy is that “orthodoxy will not permit us to say that the very existence of Father, Son, and Holy Spirit is a theory-dependent matter… The latter appears to be a form of modalism.”²⁵ The theory does not merely fail to supply the bridge; pressed, it reproduces the exact failure the bridge was needed to prevent.

The tradition’s other answer: refuse the question

There is a second answer, and following the evidence honestly makes it look stronger, not weaker, than the derivation.

Return to Sorem’s livestream. Amid the two declines, he says something else: ”So I could say the question’s wrong. Your question is incorrect.” Read as an improvisation, that is a dodge. It is not an improvisation. It is the mainstream position of the tradition, continuously attested for sixteen centuries.

Evagrius Ponticus, Gregory of Nazianzus’s own disciple, wrote that the Holy Trinity differs from the numerical triad in that, in God, “three” is not preceded by “two” and not followed by “four.” Gregory himself explains the “overstepping” of two by reference to the material world, where two is inevitable because of the form/matter division.²⁶

Nicholas Mouzalon, later Patriarch of Constantinople, objected to Eustratius in the twelfth century — and his reason turns out to be this very case’s objection, stated from inside Orthodoxy. On the numerological approach, he argued, the Trinity decomposes into a monad causing a pair which in turn causes another monad, and such a sequence would go on generating numbers past three, leading to a bad infinity.²⁷ He was not refusing to reason. He was pointing out that the reasoning over-generates. In his own words:

Nowhere is a dyad applicable to the unique divinity. There is no pairing in the Trinity. The monad is not followed by a dyad and then by a triad, so that you would think a dyad before the monad, but the monad is the fountain of the two monads which are from it … the triad is preceding the dyad, so that, in this way, it will show forth its supernaturality.²⁷

And Nicholas of Methone, refuting Proclus in the 1150s:

Therefore, the Trinity we are worshipping is not a multiplicity either … neither the dyad is before it, nor the monad is before the dyad that is within it, but the paternal monad and the dyad that is from it are showing themselves simultaneously, and the whole is simultaneously monad and triad … However, what is from the monad is not a dyad, because the two are from it not in the same way, but each of the two in a specific way — one being born and another one proceeded.²⁸

The position, from the fourth century to the twelfth, is that the “three” of the Trinity is not the three of arithmetic, and therefore that no arithmetical argument bears on it in either direction. The tradition’s answer to “why not four?” has never been a derivation. It has always been a refusal of the question’s terms.

Two things follow, and they cut in opposite directions.

In the tradition’s favour: this is not an ad hoc retreat invented when the derivation failed. It is datable, citable, and older than the derivation it is being contrasted with. It also has a modern peer-reviewed formalisation — Basil Lourié has reconstructed the patristic notion of number as a paraconsistent, non-arithmetical “quasi-ordinal,” which is a serious attempt to say precisely what the refusal amounts to.

Against it: a refusal is only as good as the account of what “three” does mean if not the numeral, and that account is not yet in hand. Refusing the question also does not remove the residual on its own terms. If the answer is “prove the whole system, and the Trinity comes with it,” then the question simply relocates: why does the whole say three rather than four? That is untouched.

Retiring the derivation costs Orthodoxy remarkably little of what it actually holds, which is worth saying plainly in fairness. Eustratius was condemned in 1117 on unrelated Christological grounds, his triadology was rejected by the mainstream of his own century, and it was later taken up chiefly by his opponents’ Latinising successors. What is lost is not Orthodox Triadology. It is a candidate derivation that the tradition’s mainstream never relied on — and, with it, the popular apologetic moves that rest on it.

What can be proved, and it is not a number

Everything so far has been negative. Something positive does come out of this material, and it deserves to be stated on its own, because it is the one thing on this page that is settled.

Take any two-place condition — begets, say — and any two things a and b. Suppose the condition holds one way round and fails the other: a begets b, and b does not beget a. Then a and b are not the same thing. The proof is one line. If they were the same thing, then “a begets b” and “b begets a” would be two ways of writing down the very same fact, because swapping a thing for itself cannot change whether a sentence is true. But one of them is true and the other is false. So they are not the same thing.

Call that the asymmetry bridge, and notice what it is not. It is not a claim about God, or about relations, or about metaphysics at all. It is a consequence of what identity means, and it holds in every model and every subject matter. Applied here: if the Father begets the Son and the Son does not beget the Father, then the Father is not the Son, and there is nothing further to establish.

Three things follow, and the third is the honest limit.

It steps around the objection that has dogged this whole area. The usual route to distinctness is the two-spheres manoeuvre described above, where a relation nothing can bear to itself forces separate occupants. That route has a serious published objection: to know the relation really is irreflexive, you seem to have to know already that its relata are two, so the distinctness being derived was presupposed all along.²⁹ The asymmetry route never incurs that debt. Write the discerning relation out — a stands to b in a way that b does not stand to a — and it is self-contradictory when a and b are the same, as a matter of pure logic, with no substantive fact about irreflexivity waiting to be grounded in anything. There is nothing for the objection to bite on.

It makes the collapse reading contradictory rather than merely unattractive. The modalist’s move throughout this page has been to redescribe the three as one thing under three descriptions. Against an asymmetric relation of origin, that redescription does not merely lose something. It asserts that one and the same person both does and does not beget himself.

And it does not give a number. The bridge yields at least two. Between two and seventeen it is completely silent.

So, stated exactly: if the relations of origin are real — not two ways of describing one undifferentiated thing — and asymmetric — the Son from the Father, the Father from no one — then there is more than one divine person, and no redescription can undo it. Both of those conditions come from revelation, and this page does not pretend otherwise. What changes is where the burden now sits. Someone who wants a solitary God can no longer reinterpret the datum. He has to deny it, against the creed and the text, in the open, where the denial can be scored.

What the monarchy of the Father bounds, and what it does not

If a ceiling exists anywhere in Eastern theology it should be here. The monarchy of the Father is Orthodoxy’s strongest structural doctrine: the Father alone is unoriginate and is the single source of the other two, neither of whom originates anything. It is what Eastern writers reach for when the question is pressed. So model it and see what it delivers.

Draw the persons as points, and put an arrow from each person to anyone he is the source of. The monarchy says exactly one point has no arrow coming in, and every other point has exactly one, coming from that same source. In the language of graphs, the structure is a tree of depth one: a root, direct children, no chains.

That is a real constraint, and it does precisely two things.

It forbids iteration, absolutely. Since no originated person originates anything, nothing can proceed from a procession. This is why the sharpest patristic argument about a fourth divine person is fatal to the Latin addition to the creed and harmless to the East. Photius, in the ninth century, puts it as a regress:

Furthermore, if the Son is begotten from the Father and the Spirit — according to this innovation — proceeds from the Father and the Son, then likewise another hypostasis should proceed from the Spirit, and so we should have not three but four hypostases! And if the fourth procession is possible, then another procession is possible from that, and so on to an infinite number of processions and hypostases, until at last this doctrine is transformed into a [pagan] Greek polytheism!³⁰

On the monarchy that regress cannot start. This is a genuine upper bound — on the depth of the structure.

It says nothing whatever about the number of children. A depth-one tree with one root may have two leaves, or three, or seventeen, and nothing in the doctrine chooses between them.

Those two facts are easy to run together and they are not the same fact. Eastern theology has an exceptionally clean answer to why does it not go on forever and no answer at all to why not one more alongside. The doctrine that settles the first question is silent on the second, and the two look like one question only until they are drawn.

Read Photius again with that in view and something else surfaces. His fourth person is the absurdity a reductio ends at. He is not deriving “there is no fourth”; he is using it, as a premise so uncontroversial that his Latin opponent will grant it without discussion. The most on-point patristic text about a fourth divine person presupposes the answer, in the one place one would most expect to find it defended.

Nor can the gap be closed by counting the processions. The natural attempt runs: there are exactly two ways of coming forth from the Father, generation and procession; that inventory is complete; so no third kind of person is possible. It is unavailable, and it is unavailable on Orthodoxy’s own authority. John of Damascus:

But the Son is derived from the Father after the manner of generation, and the Holy Spirit likewise is derived from the Father, yet not after the manner of generation, but after that of procession. And we have learned that there is a difference between generation and procession, but the nature of that difference we in no wise understand.³¹

An inventory whose two members are told apart by a difference the tradition confesses it does not understand cannot be shown to be exhaustive. To argue that there are exactly two modes and that no third is possible, you have to be able to say what a mode is and what makes two of them different. Damascene says, as dogma, that we cannot. The apophatic rule that protects the doctrine from rationalist over-reach is the same rule that forbids the closure argument — which is the tradition being consistent, not the tradition being evasive.

There is a further cost here, and it falls on the East. Take the tidiest structural argument for exactly three: a causal order admits exactly three standings — causing only, both caused and causing, caused only — so there are three persons. Its middle cell is the Son as co-source of the Spirit. On the monarchy of the Father there is no such cell, because no caused person causes. The argument a Western reader is most likely to reach for requires the filioque, and an Orthodox one cannot use it.

And the same doctrine cuts once more, harder. On the monarchy there are exactly two causal standings, uncaused and caused, and three persons. Eastern Orthodoxy is therefore committed, by its own confession, to the proposition that a count of kinds is not a count of individuals. That is this page’s central objection, and it is not an external one. Any proponent who argues “there are exactly n modes of X, therefore exactly n persons” is refuted by his own creed in the single case where his tradition has actually done the counting.

Every reason for three is a floor, and nobody has built a fence

Suppose the demand for a strict derivation is dropped and the question becomes the softer, more reasonable one: not why there must be three but why three makes sense. This is where most thoughtful believers actually live, and it deserves to be taken seriously rather than treated as a retreat.

It survives the move. But it survives in a shape that has to be stated carefully, because “fitting” is an excellent hiding place for the necessity claim that has just failed — the same argument with the modal word quietly deleted leaves no trace in the prose. Three tests, fixed before the candidates were assembled: would the reason have produced three for someone who did not already hold three; does it score rival accounts of God without automatically favouring the Trinity; and does it stay a claim about fittingness instead of sliding back into “and therefore exactly three”. Four reasons survive in some form.

Parsimony above a revealed floor. Revelation names three; do not multiply entities beyond need; so three is the count to hold, and a fourth is not merely unattested but unwarranted. This is the strongest of the four and the most modest, and it is the honest core of the whole case. It is also a rule about what to believe rather than a fact about what can exist: it explains why nobody should posit a fourth and it excludes nothing. Note, too, which way it pulls. Parsimony’s unaided pressure is downward, toward one. It supports three only because the floor is set somewhere else.

Three is the first number past a closed polarity. With exactly two, each is nothing but the other’s opposite pole — the Father is the one who begets this one, the Son is the one begotten by that one — and there is nothing further to say. Lossky, who states this as the specifically Eastern reason and is careful to distinguish it from the Western argument from love, calls the resulting diversity merely relative: “the dyad is always an opposition of two terms, and, in that sense, it cannot signify an absolute diversity… Where the Trinity is concerned, we are in the presence of the One or of the Three, but never of two.”³²

One correction belongs here, because the intuition is usually stated too strongly. It is not true that a pair is formally degenerate. The two-element structure with a single begetting arrow has no non-trivial symmetry, and its two members are distinguishable at the strongest grade logicians recognise: one of them begets something and the other does not. What is true, and it is weaker, is that with two members the relation and its converse use up every pair there is, so every fact about either member is a fact about that one ordered pair. Whether that polarity is a defect is a theological judgement and not a theorem.

Shared love. Richard of St Victor’s argument, quoted above, with the necessity claim that kills it removed. A love closed on a pair and a love that wills the beloved’s love to reach past the pair are different in kind, and the second is the richer; three is the first point at which the second is available. Two caveats, and they are large. The ranking of the two kinds is exactly what a rival with a different account of divine perfection will decline. And Lossky rules the whole argument-form out as a basis for the doctrine — he wants no “moral basis for the doctrine of the Trinity, e.g. the idea of love seeking to share its own plenitude with others”³³ — while popular Orthodox apologetics reaches for it constantly. That tension is real, and disclosing it is better than smoothing it over.

Three is the first number at which each person can be fully unfolded. Dumitru Stăniloae’s version, and the closest of the four to how the intuition is usually felt.³⁴

Now lay them side by side and read their logical form, which is the point of the exercise.

Four of the five are minimality claims in their own authors’ words — “at least”, “first”, “suffices”. The fifth is the only one that gestures at an upper bound, and it does so by asserting the boundary rather than arguing for it, which is why Lossky glosses it as showing “the insufficiency of any number other than three” instead of treating it as a demonstration.³⁵

So the shape of the evidence is this. The case for three is a stack of independent arguments that push upward and come to rest at three, and on the other side there is nothing. Not a weak ceiling: no ceiling. Across fifteen hundred years and six disciplines, this research has not located one argument, in any tradition, whose form is four is impossible. The ceiling is held by revelation and guarded against speculation by parsimony.

That is a real result for the intuition and a sharp constraint on how it may be stated. “Three makes a lot of sense” is true. What makes sense is reaching three. Nothing in the reasons makes sense of stopping there, and the reasons say so in their own words.

One thing should not be undersold in the other direction. Four independent partitions of relational standing — modes of love, causal standing, minimal irreducible arity, passage beyond polarity — all land on three cells, produced by four traditions with different vocabularies and different aims. That convergence shows three is not arbitrary, which is a genuine thing to have shown and worth a good deal against the charge that the number is a historical accident. It is also a completely different thing from showing three is necessary, and both halves of that sentence need saying.

What would close this

Because the gap is now precise, the specification for closing it can be written down — and two of its five items can now be marked off, which is worth doing openly, since the same specification was published here with all five open.

1. An irreflexivity theorem, not an irreflexivity axiom. Show that the relation constituting the divine inner life cannot be borne by an individual to itself, from something more basic than the assertion that it cannot. Largely met. The asymmetry bridge derives it, and derives it from logic rather than from theology. Augustine’s flat axiom is no longer needed and should not be used: it is the general claim, and the general claim is exactly what the circularity objection is built to attack. The weight moves instead onto the premise that the relations of origin are real and asymmetric — which is revealed, and which this page has never claimed otherwise.
2. A way out of the circle. The distinctness has to be establishable without presupposing that the relata are already distinct. Met, for asymmetric relations. Black’s spheres work as an objection because “two miles apart” is symmetric: identify the two spheres, read the space as a cylinder, and the symmetric relation survives intact while the irreflexivity evaporates. Against an asymmetric relation the same manoeuvre yields a contradiction rather than a cylinder. Every published objection in this literature is aimed at the symmetric case, because the symmetric case is where the physics examples live.
3. A cardinality step separate from the first two. Distinctness yields at least two. Getting to exactly three needs an independent argument. Not met, and the failure is now diagnosed rather than merely recorded: every candidate dies at the same step, reading a complete inventory of kinds off as a census of individuals. Richard’s shared love remains the best-shaped candidate and fails there too.
4. A genuine upper bound. “Adds no new kind” is not “cannot exist.” The upper bound must be a modal claim — a fourth is impossible — not a parsimony claim that a fourth is unnecessary, because parsimony cuts against the fecundity rationale used to get past one. Not met, and now known to be unattempted. No argument of that shape was located in any tradition or century searched.
5. Symmetric scoreability. Whatever criterion the argument uses must be stated in advance and applied to every rival — Islamic attribute-theism, Neoplatonic emanation, unitarianism, modalism — under the same standard. Met, and it is the reason this page hands nobody a victory.

Orthodoxy’s own answer: the derivation was never on offer

Items 3 and 4 have been open for eighteen centuries. The natural reading is that this is a hole in Christian theology which nobody has yet got round to filling. There is a more interesting possibility, and it turns out to be what Eastern Orthodoxy actually teaches: that the hole is exactly where its own dogmatics say it has to be.

Vladimir Lossky is the most widely read modern Orthodox systematic theologian, and he is not conceding under pressure. Writing on the very passage of Gregory of Nazianzus this page opened with, he says:

It is the Trinity, and this fact can be deduced from no principle nor explained by any sufficient reason, for there are neither principles nor causes anterior to the Trinity.³⁶

Of Gregory’s monad–dyad–triad sentence itself — the one every popular version produces as a proof:

St. Gregory Nazianzen is not seeking to vindicate the trinity of persons before the human reason: he simply shows the insufficiency of any number other than three.³⁵

And, drawing the conclusion:

Here apophaticism finds its fulfilment in the revelation of the Holy Trinity as primordial fact, ultimate reality, first datum which cannot be deduced, explained or discovered by way of any other truth; for there is nothing which is prior to it. … This is the reason why no philosophical speculation has ever succeeded in rising to the mystery of the Holy Trinity.³⁷

His reason has the shape of a transcendental argument turned back on itself. A derivation has to run from premises that are prior to its conclusion; nothing is prior to the ultimate; so no derivation of the ultimate can exist. He states it as flatly as it can be stated elsewhere: the Trinity is “a primordial truth, incapable of being based on any process of reasoning whatever, because all reasoning, all truth, and all thought prove to be posterior to the Trinity.”³⁸

Notice what that does to the dispute. It converts “Orthodoxy has an unfilled hole” into “Orthodoxy predicted this hole and explained why it is there.” This research spent five passes demolishing every available derivation and landed where Lossky’s dogmatics already stood. What the demolition refutes is not Orthodox teaching. It is a popular apologetic that promises a derivation Orthodox teaching says is unavailable.

Three qualifications, because this is a finding and not a trophy.

It is completely symmetric, and so it wins nothing comparatively. A unitarian, a Muslim, an Advaitin or a naturalist can say precisely the same of their own ultimate: nothing is prior to it, so demanding a derivation of it is incoherent. The principle neutralises the demand for everybody at once. It removes a supposed Christian-specific deficit and supplies no positive reason to prefer Christianity to anything.

It has a price, and presuppositional apologetics pays it. If nothing is prior to the Trinity, then no argument from the preconditions of thought can deliver the Trinity as its conclusion — which is exactly what Sorem said, unprompted, in the livestream at the top of this page, and exactly what the popular presentation of that argument claims. Lossky’s principle and that presentation are inconsistent, and Lossky is the one with dogmatic standing.

Lossky is not the whole tradition. Richard of St Victor is the standing counterexample to any claim that Christian theologians have always held the Trinity underivable: he denied it explicitly and set out to supply necessary reasons for it. He is also a Latin, Orthodoxy never received him, and Lossky repudiates his argument-form by name.

Where this leaves the question

Open — but characterised now, rather than merely open.

The whole of it in one paragraph. There is no argument, in any tradition or century this research located, that derives exactly three divine persons from anything more basic; and the failure is not a series of near-misses but one identifiable mistake made five times, a complete inventory of kinds — offices, argument-places, degrees of value, modes of love, modes of causing — read off as a census of individuals. That inference is invalid in general, and Eastern Orthodoxy is committed to its invalidity by its own doctrine, since on the monarchy of the Father there are two causal standings and three persons. What can be proved is distinctness: if the relations of origin are real and asymmetric, more than one person follows as a one-line consequence of what identity means, and the collapse into a single person is not merely unattractive but self-contradictory. What cannot be proved is the number, in either direction — and the sharper finding is that nobody has ever attempted a ceiling. Every reason ever given for three is a minimality claim in its own author’s words. The ceiling is held by revelation and guarded by parsimony, which is a rule about what to posit rather than a fact about what can exist. Nor is the revealed structure closed: the monarchy bounds the depth of the divine origination structure and says nothing about its breadth, and the completeness of the two-procession inventory is not merely unproven but unprovable within Orthodoxy’s own commitments. All of which lands exactly where Orthodoxy’s own dogmatics already stood.

Three things follow that are worth keeping apart.

This page hands nobody a victory. The verdict lands on the Trinitarian side’s argument because that is who was advancing it, and it applies with equal force in the other direction. A unitarian who argues “the principle of actualisation is one-place, therefore exactly one divine person,” or a modalist who argues that because the triad’s three moments are stages of one process the occupant must be one, commits the identical error. Structure alone underdetermines occupant-count in both directions, and whoever wants a count must bring a claim about what a particular relation is and defend it.

Not arbitrary is not the same as necessary, and the gap between them is where most of the argument actually lives. Four traditions with different vocabularies converging on three is a real datum and answers the charge that the number is a historical accident. It is worth a great deal rhetorically and nothing at all modally, and anyone who reports one half without the other is misreporting it.

And the two literatures still have not been introduced. The metaphysics of relations has a precise apparatus that settles the general case and has never been applied to the Trinity. Trinitarian theology has an unfinished individuation project and has never reached for that apparatus. That is a better place to be than a bare gap, because the tools exist and the remaining work is joinery rather than invention — and the asymmetry bridge above is one piece of that joinery, done.

References

References

1. Gregory of Nazianzus, Oration 29.2. Translation cross-checked against the Lossky/Burnett rendering. The Greek arche is the “beginning” of John 1:1. ↩

2. Maximus the Confessor, Ambiguum 1 — a commentary on the Gregory passage above. Translation from the Dormition Press edition. ↩

3. “Monday (07/27) PF Q&A Discussion,” Patristic Faith, streamed 27 July 2026, 2h43m. Quotations are from the automatic English captions at 01:12:02, 01:13:06, 01:15:34, 01:17:31 and 01:18:04, lightly corrected for speech-recognition errors that are noted in the underlying research file (“diad” for dyad, “ethical councils” for ecumenical councils). Speaker identity confirmed from the stream itself at 01:03:19 and 01:14:25. The chat question is quoted verbatim from the live-chat replay at 01:07:35. ↩ ↩² ↩³ ↩⁴ ↩⁵

4. Aristotle, De Caelo I.1, 268a6–15, Stocks translation (Oxford). ↩

5. Proclus, Elements of Theology, proposition 35, Dodds translation. ↩

6. Proclus, Elements of Theology, proposition 33, Taylor translation: “Every thing which proceeds from a certain thing and is converted to it, has a circular energy… it conjoins the end to the beginning, and the motion is one and continued.” ↩

7. Basil Lourié, “Numerology and Logical Schemes in Byzantine Triadology of the 12th Century,” Logico-Philosophical Studies 16, nos. 1–2 (2018), 22–24; and, in a fuller treatment with clean Greek and additional footnotes, “Eustratius of Nicaea, a Theologian: About the Recent Publications of Alexei Barmin,” Scrinium 16 (2020), 344–358. ↩

8. Eustratius of Nicaea, Sermon on the Holy Spirit, ed. A. V. Barmin (Moscow, 2006), 518–565; Greek and English as printed by Lourié, Scrinium 16 (2020), 353. Quoted here at one remove from Lourié; the critical edition was not independently consulted. ↩

9. Photius, Bibliotheca, codex 187, summarising Nicomachus of Gerasa’s Arithmetical Theologumena, on the tetrad. ↩

10. Robert Burch, A Peircean Reduction Thesis (Texas Tech University Press, 1991); Joachim Hereth Correia and Reinhard Pöschel, “Peircean Algebraic Logic and Peirce’s Reduction Thesis,” Semiotica (2011). Peirce’s own formulation is at Collected Papers 3.423–4. ↩

11. “Logical reduction of relations: from relational databases to Peirce’s reduction thesis,” Logic Journal of the IGPL (2023), theorem 16. The equality can fail on finite domains, and the authors prove it fails for almost all relations on large finite ones. ↩

12. “Is Peirce’s Reduction Thesis Gerrymandered?”, Transactions of the Charles S. Peirce Society (2022); preprint arXiv:2406.14058. ↩

13. Andrew Robinson, God and the World of Signs: Trinity, Evolution, and the Metaphysical Semiotics of C. S. Peirce (Brill, 2010), chapter 2; and Robinson and Southgate, “Semiotics as a Metaphysical Framework for Christian Theology,” Zygon 45 (2010). Characterised here from the chapter opening and secondary summaries rather than a full reading of the book. ↩

14. Cody Gilmore, “Slots in Universals,” Oxford Studies in Metaphysics 8 (2013), 187–233: “Slot theory… is the view that (i) there exist such entities as argument places, or ‘slots’, in universals, and that (ii) a universal u is n-adic if and only if there are n slots in u.” ↩

15. T. Scott Dixon, “Plural Slot Theory,” Oxford Studies in Metaphysics 11 (2018), 193–223, stating slot theory’s occupation axioms. The rival “pocket theory” drops the first axiom and retains the second. ↩

16. Jukka Keränen, “The Identity Problem for Realist Structuralism,” Philosophia Mathematica 9, no. 3 (2001), 308–330. See also Tim Button, “Realistic structuralism’s identity crisis: a hybrid solution,” Analysis 66, no. 3 (2006), 216–222, whose repair is — again — irreflexive relations. ↩

17. F. A. Muller and S. Saunders, “Discerning Fermions,” as quoted in Dieks and Versteegh, “Identical Quantum Particles and Weak Discernibility,” Foundations of Physics 38 (2008). The underlying principle is Quine’s, from “Grades of Discriminability” (1976). ↩

18. Augustine, De Trinitate I.1. Latin verified against the CSEL text: nulla enim omnino res est quae se ipsam gignat ut sit. English from the Haddan translation. ↩

19. Aquinas, Summa Theologiae Ia q.28 a.1, objection 2, Freddoso translation: “every real relation requires two really distinct extremes.” ↩

20. Aquinas, Summa Theologiae Ia q.28 a.1 ad 2, English Dominican translation. See also q.28 a.3: “The very nature of relative opposition includes distinction.” ↩

21. Aquinas, Summa Theologiae Ia q.36 a.2, English Dominican translation. The four-relations count is at q.28 a.4; the two-processions claim is grounded at q.27. ↩

22. Richard of St Victor, De Trinitate III.19, in the translation quoted in “Love for a Third,” First Things. Richard’s Latin was not obtained; the quotation is at one remove. ↩

23. Dennis Bray, “Richard of St. Victor’s Argument for the Necessity of the Trinity: An Exposition and Analysis.” Bray reconstructs the argument in four steps, the fourth being that the metaphysics of the divine processions and of love together rule out a fourth person. Cited here from the abstract; the full text has not yet been read, and its treatment of that fourth step is the single most on-point unread source for this case. ↩

24. Internet Encyclopedia of Philosophy, “Trinity,” section on Latin Trinitarianism. ↩

25. Michael Rea, “Relative Identity and the Doctrine of the Trinity,” Philosophia Christi 5, no. 2 (2003), 443. ↩

26. Evagrius Ponticus, Kephalaia Gnostica VI.10–13, reported in Lourié, Scrinium 16 (2020), 354, together with Gregory’s Oration 23.8. ↩

27. Nicholas Mouzalon, On the Proceeding of the Holy Spirit, chapter 47, ed. Zēsēs (1978), 325, reported and translated in Lourié, Scrinium 16 (2020), 355 and n. 35. ↩ ↩²

28. Nicholas of Methone, Refutation of Proclus’s Elements of Theology, ed. Angelou (1984), 135.24–31, Lourié’s translation. ↩

29. The objection is Steven French’s, in “Identity and Individuality in Quantum Theory,” Stanford Encyclopedia of Philosophy, and in French and Krause, Identity in Physics (Clarendon, 2006), 172; its sharpest form is Erica Shumener’s, in “The Metaphysics of Identity: Is Identity Fundamental?”, Philosophy Compass 12 (2017), which argues that the irreflexivity fact is itself grounded in the distinctness it was supposed to establish. Both are quoted verbatim, with page citations, in David Wörner, “On making a difference: towards a minimally non-trivial version of the identity of indiscernibles,” Philosophical Studies 178, no. 12 (2021), 4261–4278, which was read in full and which answers them for the symmetric case as well; the primary texts were consulted at one remove through it. No reply by French to Wörner was located. The point made here is narrower and does not depend on who wins that exchange: the asymmetry route never takes on the debt. ↩

30. Photius, Mystagogy of the Holy Spirit §37, in the English translation as hosted independently at two sites which agree verbatim. §38 presses the same point as a charge of arbitrary favouritism between the Son and the Spirit. The Greek was not obtained and no critical edition was consulted. ↩

31. John of Damascus, De Fide Orthodoxa I.8, in the standard English translation, verified as agreeing verbatim across three independently hosted editions; the Greek was not obtained. Lossky glosses the same passage: “Understood apophatically, the relation of origin describes the difference but nevertheless does not indicate the manner of the divine processions.” ↩

32. Vladimir Lossky, “The Procession of the Holy Spirit in Orthodox Trinitarian Doctrine,” §5, in In the Image and Likeness of God. He adds that the passage beyond the dyad “is not an infinite series of persons but the infinity of the procession of the Third Person: the Triad suffices to denote the Living God of revelation.” ↩

33. Same essay, §6, where he also rules out the Neoplatonist bonum diffusivum sui. ↩

34. Dumitru Stăniloae, Teologia Dogmatică Ortodoxă, reported in a Russian-language survey of his Trinitarian theology: the dyad is “a closedness of the two, and therefore imperfect,” and “three is the first number permitting each Person to unfold in fullness.” This is the weakest-sourced item on the page — read at one remove and in translation, from a secondary survey rather than from Stăniloae’s own Romanian or the English The Experience of God — and it is flagged for verification rather than relied on. ↩

35. Vladimir Lossky, The Mystical Theology of the Eastern Church (SVS Press, 1976), ch. 3, glossing Gregory of Nazianzus, Oration 45.4. Gregory is quoted here at one remove, in Lossky’s translation; the NPNF and Wickham renderings were not compared for this passage, and the oration numbering in Lossky’s edition differs by one from the modern numbering used elsewhere on this page. ↩ ↩²

36. Lossky, Mystical Theology, ch. 3, immediately after expounding Gregory’s monad–dyad–triad passage. The chapter was read in full, with its own footnote apparatus. ↩

37. Lossky, Mystical Theology, ch. 3, conclusion. ↩

38. Lossky, “The Procession of the Holy Spirit in Orthodox Trinitarian Doctrine,” §6. ↩