A Scholarly Essay in Seven Modes
On what is preserved and destroyed when a proof travels across representational systems
"The mathematician's patterns, like the painter's or the poet's, must be beautiful; the ideas, like the colours or the words, must fit together in a harmonious way."— G.H. Hardy, A Mathematician's Apology (1940)
Mode I
We begin where everything begins: the source. Euclid's Elements, composed around 300 BCE, is the foundational text of demonstrative mathematics. Book IX, Proposition 20 establishes that prime numbers are more numerous than any assigned multitude of prime numbers — which is to say, there is no largest finite collection of them. The proposition and its proof survive in Greek, transmitted through a manuscript tradition whose earliest complete witness is the Byzantine manuscript Vat. gr. 190, copied around 888 CE, more than a millennium after Euclid wrote.1
Prime numbers are more than any assigned multitude of prime numbers.
Let A, B, C be the assigned prime numbers. I say that there are more prime numbers than A, B, C.
For let the least number measured by A, B, C be taken, and let it be DE. Let the unit DF be added to DE. Then EF is either prime or not. First, let it be prime. Then the prime numbers A, B, C, EF have been found which are more than A, B, C.
Next, let EF not be prime. Then it is measured by some prime number. Let it be measured by the prime number G. I say that G is not the same as any of A, B, C. If possible, let it be so. Now A, B, C measure DE; therefore G also will measure DE. But it also measures EF; therefore G will measure the remainder DF, that is, the unit, which is absurd. Therefore G is not the same as any of A, B, C. And it was assumed to be prime. Therefore the prime numbers A, B, C, G have been found which are more than the assigned multitude of A, B, C.
Q.E.D.
"Euclid begins not with all primes but with any finite set of primes — a distinction that matters enormously."
Two features of Euclid's proof demand attention. First, he works with a concrete, named example — three primes A, B, C — and the generality is understood to hold for any such finite collection. This is typical of the Elements: Euclid proves a general theorem by demonstration on a representative instance.2
Second, and more importantly: Euclid does not form the product of all primes. He forms the least common multiple of the given finite set — which, since the set consists entirely of primes, is simply their product — and adds one. The resulting number N need not itself be prime; it suffices that N has some prime factor not in the original list. The proof does not claim N is prime, nor does it require N's prime factors to be discoverable. It suffices that they exist.3
The widespread modern formulation — "suppose p₁, …, pₙ are all the primes" — is a subtle but significant restatement. It turns a constructive demonstration into a proof by contradiction. Euclid's version is arguably more elegant: he proves a comparative claim (there exist more primes than any finite list) without the assumption that the list is complete.
Mode II
Theorem. There are infinitely many prime numbers.
Proof. Suppose, for contradiction, that there are finitely many primes. Enumerate them as a complete list:
Form the number
By the Fundamental Theorem of Arithmetic, N has at least one prime factor; call it q. There are two cases:
Case 1. q is one of the listed primes, say q = pⱼ for some j ∈ {1, …, n}. Then pⱼ ∣ N and pⱼ ∣ (p₁ × ··· × pₙ), so pⱼ ∣ N − (p₁ × ··· × pₙ) = 1. But no prime divides 1 — contradiction.
Case 2. q is not any of the listed primes. Then q is a prime not contained in our supposed complete list — contradiction.
Both cases yield a contradiction. Therefore the assumption that primes are finite is false. ∎
The modern version deploys the Fundamental Theorem of Arithmetic — that every integer greater than 1 has a unique prime factorization — as a lemma. This theorem, which Euclid proves separately in Book VII (Propositions 30–32), is implicit in Euclid's argument but not invoked by name. The algebraic rendering makes explicit what Euclid leaves to the reader's intuition about divisibility.
The shift to universal quantifiers and set-theoretic language ("let {p₁, …, pₙ} be all primes") accomplishes two things: it gains expressive clarity, and it loses constructive character. Where Euclid's proof is a procedure for extending any finite list, the modern proof is a demonstration that no finite list could be total. These are logically equivalent but phenomenologically distinct — a point to which we return in Mode VII.
Mode III
The Petrarchan (Italian) sonnet divides into an octave (ABBA ABBA) and a sestet (here CDC DCD), separated by a volta — a turning or counterstatement. The logical structure of the proof — assumption, construction, case analysis, contradiction — maps with uncomfortable precision onto the sonnet's binary architecture: assumption in the octave, collapse in the sestet. What the form cannot easily accommodate is the simultaneity of both cases; the sonnet, being sequential, must handle them serially. The gloss records where the form resists the argument.
The rhyme scheme is ABBA ABBA / CDC DCD: A = begun / undone / one / run (the "-un" sound, binding the assumption to the contradiction); B = whole / toll / goal / stroll (the "-ole/-oll" family, all of them words about completeness, payment, direction, movement — the octave's semantic field); C = found / unbound / unwound ("-ound," words of discovery and release); D = outright / right / sight ("-ight," words of clarity and vision, thematically apt for a proof that ends in seeing). One genuine imprecision: "arithmetic's goal" uses "goal" loosely — the Fundamental Theorem of Arithmetic is not a goal but a theorem about how integers are constituted. The argot of mathematics and the argot of poetry have different tolerances for teleology.
An honest note: the proof's simultaneous case analysis — "either Case 1 or Case 2, and both lead to contradiction" — is a logical structure that poetry handles awkwardly. The sonnet handles Cases 1 and 2 sequentially (lines 9–11 then 12–14) rather than simultaneously. This is not a failure of the poem but a constitutive difference between sequential art forms and logical proof, which can, in principle, assert "for all X in some set, P(X)" in a single step. Whether this means the sonnet does not prove the theorem — or proves it differently — is the question Mode VII and the critical essay address.
Mode IV
The scene is the agora of Athens. Arithmetikos, a geometer who has made a study of the odd numbers men call primes, has announced that he has at last completed his accounting of them. Socrates encounters him near the stoa.
Socrates: Arithmetikos! I have been looking for you. They tell me you have accomplished a great thing — that you have counted all the primes.
Arithmetikos: It is true, Socrates. After many years of labor, I have determined that there are exactly this many prime numbers, and I have their names, so to speak — their values — written here on this tablet.
Socrates: This is wonderful, if it is so. But I confess I am slow to understand such things. Help me: what is a prime number?
Arithmetikos: A number measured by no number except unity and itself. Two, three, five, seven, eleven — these and their fellows.
Socrates: And you say you have all of them here on your tablet?
Arithmetikos: I do.
Socrates: Forgive me — but what do you mean by "all"? When I say I have all the figs in this basket, I mean there is no fig outside the basket. Is that what you mean?
Arithmetikos: Precisely. There is no prime number that does not appear on my tablet.
Socrates: Very good. Now, is your tablet finite? Does it have an end?
Arithmetikos: Naturally. It is a tablet, not an infinite scroll. It has a last entry.
Socrates: Then you can, in principle, multiply all the numbers on your tablet together?
Arithmetikos: In principle, yes. The number would be very large, but finite.
Socrates: Let us call that product Π. And could we add one to Π?
Arithmetikos: Of course. Call it N. N is Π plus one.
Socrates: Now — every number greater than one is either prime itself or is measured by some prime number. Is this not so? We established something like this in our conversation last week with the geometer from Megara.
Arithmetikos: Yes, I accept this. Every number has some prime divisor.
Socrates: Then N has some prime divisor. Let us call it G. Now, is G on your tablet?
Arithmetikos: It must be. We said all primes are on the tablet.
Socrates: Then G is one of the numbers you multiplied together to make Π?
Arithmetikos: It must be.
Socrates: So G measures Π — divides it without remainder?
Arithmetikos: Yes, since G is one of the factors of Π.
Socrates: And G measures N, since G is a prime divisor of N?
Arithmetikos: Also yes.
Socrates: Then G measures the difference between N and Π. What is that difference?
Arithmetikos: N minus Π is... one. G divides one.
Socrates: And can a prime number divide one?
Arithmetikos: No. A prime is greater than one by definition, and no number greater than one divides one — for that would make the quotient a fraction, and we are speaking of whole numbers. [A pause.] I see the difficulty.
Socrates: Tell me the difficulty as you see it.
Arithmetikos: If G is on my tablet, we reach an absurdity — it would have to divide one, which no prime can do. So G is not on my tablet. But I said G is prime, and I said all primes are on my tablet. I have contradicted myself.
Socrates: Yes. So what must we give up?
Arithmetikos: Either — either G is not prime, but we know it is. Or — G is not on my tablet, which means the tablet is not complete. Or: there is no such thing as a complete tablet of primes. The primes are without end.
Socrates: I think you have discovered something true. But notice: you did not need to examine the primes beyond those on your tablet, nor to find G explicitly. The mere existence of the argument was sufficient. Does it trouble you that we have proven there is a prime you cannot name?
Arithmetikos: It troubles me greatly. I feel as though I reached for the wall of a room and my hand passed through it.
Socrates: Perhaps that is what infinity feels like, when you first touch it. Shall we go in and have some wine and speak of other impossibilities?
Mode V
Below is the logical skeleton of the proof rendered as a natural deduction tree, reading from premises (top) to conclusion (bottom). Horizontal lines represent inference steps, labeled at the right margin. The tree uses standard connectives: ¬ (not), ∧ (and), → (implies), ⊥ (contradiction/absurdity).
The tree reveals something that prose obscures: the proof's structure is a disjunction elimination. Once we know that N has some prime factor q, we do not need to determine which case obtains. The argument works for both branches simultaneously. This is why the proof is non-constructive in a precise sense: it does not identify q.5
Mode VI
The following renders the proof in the voice, syntax, and chapter-and-verse organization of the Authorized Version of the Bible (1611). The logical steps are preserved in full.
The Book of Numbers Eternal — Chapter the First
Chapter the Second — Concerning the Division of N
Chapter the Third — The Refutation and the Wonder
Mode VII
This section proceeds differently from the others. It does not render the proof in another code; it attempts to describe what it is like to understand the proof — attending to the structure of the experience rather than the structure of the argument. The tradition being drawn on is phenomenological: Husserl's account of mathematical ideality, Merleau-Ponty's analysis of bodily knowing, and the more practically-minded accounts of mathematical discovery in Hardy, Poincaré, and Hadamard.6
Begin with what is there before understanding begins. You read the proposition: there are infinitely many primes. The sentence arrives as a string of words, each legible, none individually obscure. And yet the sentence as a whole does not yet illuminate; it sits on the page as a claim to be verified or refuted, not as a thing seen. Hardy distinguishes "seeing" a mathematical truth from merely knowing it, and the distinction is phenomenologically serious.7
Then you read the proof. The first move — assuming finitude — requires a kind of voluntary imaginative act: you construct, in something like mental space, a closed box containing all the primes. This is not a visual image exactly, but it has a spatial quality. The box has edges. It contains something. You are aware of yourself performing this construction, bringing a hypothetical world into being for the duration of the argument.
The second move — the product, plus one — requires no special imagination; it is mechanical, almost. You do not need to compute; you need only to follow. There is the product (large, indeterminate, but there), and there is the product-plus-one. The plus-one is what Wittgenstein might call a grammatical move: it creates a number with a definite relationship to everything in the box — it cannot be divided by any of them without leaving a remainder of one.
Then comes the pivot. The proof asks: where does the prime factor of N live? And here something happens that Poincaré described as characteristic of genuine mathematical insight: the two cases — it's on the list and it's not on the list — are felt as genuinely exhaustive, genuinely mutually exclusive. The exhaustiveness is not proved at this moment; it is grasped as a property of the structure that has been constructed.8 You hold both cases in mind simultaneously, and you see — before you have followed either branch to its conclusion — that both will end in the same place.
This is the moment of insight. It is not preceded by conscious reasoning; the reasoning comes after, to verify what has already been seen. Hadamard, surveying mathematicians including Poincaré, found that genuine mathematical discovery characteristically involves this structure: unconscious preparation, sudden illumination, then conscious verification.9 Whether this phenomenology is peculiar to mathematics or is shared by other kinds of problem-solving remains debated, but the felt quality — the "aha" as a distinct experiential event — appears to be reliably reported.
What one sees, in seeing the proof, is something like a topology: the space of primes is such that any finite closure of it generates a point outside it. The argument has spatial-structural quality, not merely linguistic or propositional quality. This is what Hardy means, plausibly, when he writes that mathematics is not about the manipulation of symbols but about the discovery of patterns — patterns in a space whose objects have the peculiar quality of necessity, of not being able to be otherwise.
After the insight, something persists. The proof is not merely added to one's stock of known propositions; it changes how the primes themselves look. They are the same objects they were before, and yet they are now recognizably infinite — not as a property one has learned, but as something one can see in them directly. Husserl called this modification of intuition by conceptual apprehension "Kategoriale Anschauung" — categorical intuition — the seeing of abstract structure in concrete instances.10 Whether or not the Husserlian framework is fully adequate to describe mathematical experience, it points to something real: understanding the proof changes what you see when you look at the primes, not merely what you believe about them.
What cannot be conveyed in any of the seven modes — what exists strictly in the first person — is the phenomenal character of this change: the specific texture of this understanding, for this person, at this moment. The proof is public; the understanding of it is, in its deepest dimension, private. This is not a skeptical point about communication, but a structural observation about the difference between a proof and its comprehension.
Critical Essay
Seven modes of the same proof. We can now ask systematically: what do they share, what does each uniquely contribute, and what is, perhaps, incommunicable in any mode?
The logical structure — the entailment relations — survives every translation. In every mode, there is a finite assumption, a construction, a case split, a contradiction in each branch, and a conclusion. Even the KJV narration and the Petrarchan sonnet, the two modes most distant from mathematical notation, contain all the logical steps. The Greek text, the algebraic rendering, the proof tree, the dialogue, and the phenomenological description all agree on the inference pattern: assume finite, construct N, derive contradiction. This invariant is the proof itself, stripped of its mode of presentation.
Lakatos argues, against a formalist reading, that what makes something a proof is not its form but its persuasive force — its capacity to produce conviction by a process that is itself surveyable and criticizable.11 On this view, the seven modes above are all genuine proofs (or near-proofs), because they all deploy the same surveyable pattern of reasoning. Whether the sonnet "proves" anything is a question Lakatos would reformulate: does it produce the right kind of conviction by the right kind of means?
The Greek original reveals the historical contingency of mathematical notation. Euclid's proof uses named line-segments (A, B, C, DE, EF) rather than algebraic variables, because Greek mathematics is essentially geometric: numbers are lengths. The proof is structurally identical to the modern version, but its idiom shows that the same argument can be conducted within radically different ontological frameworks.
The algebraic rendering reveals the inferential skeleton with maximum efficiency. Variables and symbols eliminate all rhetorical content, leaving only the logical bones. But efficiency has a cost: the algebraic proof conceals the constructive character of Euclid's argument and replaces it with a cleaner reductio. The readability gained is, simultaneously, an erasure of the genetic story.
The sonnet reveals the rhythmic and aesthetic dimension of mathematical argument. A proof has a shape — it builds, turns, and resolves — and the sonnet, whose architecture is defined by exactly this kind of tripartite movement (octave / volta / sestet), maps the proof's shape onto a form that embodies it in sound and meter. What the sonnet reveals is that the "beauty" Hardy attributes to proofs is not metaphorical: it shares structural properties with aesthetic experience in other domains. The sonnet also reveals, by the friction it creates (the compressed sestet, the metrical stresses landing on "prime" and "one"), exactly where the proof's argument resists poetic paraphrase — and that resistance is itself a finding.
The Socratic dialogue reveals the social and dialogic structure of mathematical persuasion. Proofs, in practice, are not inscriptions but conversations: a mathematician leads another to see what they have seen. Socrates' characteristic move — not asserting but questioning — shows that the structure of Euclid's proof is elicitory: each step can be extracted from the interlocutor by the right question. The dialogue also dramatizes the experience of aporia — the moment when Arithmetikos discovers that his tablet must be incomplete — which is part of the proof's phenomenological reality but is suppressed in formal presentations.
The proof tree reveals the tree structure — the fact that the proof is a directed acyclic graph of inferences, not a linear sequence of sentences. The tree makes visible the simultaneous-case-elimination structure that prose linearizes. It also reveals what the proof requires: the Fundamental Theorem of Arithmetic and the divisibility lemma appear as distinct, explicitly labeled premises. In prose, these premises are often so obvious that they go unstated; the proof tree forces them into visibility.
The KJV rendering reveals the communal and received character of mathematical knowledge. The biblical register — with its "and it was said unto him" and "it is written in the law of number" — locates mathematical truth in a tradition of transmission, in a community of interpreters and teachers. Mathematical knowledge does not exist only in formal systems; it exists in the practice of a community that transmits it, glosses it, disputes it, and hands it on. The KJV mode makes visible what Wittgenstein calls the "form of life" within which mathematical assertions have their sense.12
The phenomenological description reveals the temporal and subjective structure of mathematical understanding — the process by which a static proof becomes dynamic comprehension. Poincaré's insight, confirmed by Hadamard's surveys, is that mathematical understanding is not simply decoding: it involves imagination, felt structure, and the particular experience of seeing one thing in terms of another. The phenomenological mode reveals that proofs are objects with a life in time, not merely arrays of symbols on a page.
Euclid's experience of discovering the proof — the cognitive event that occurred in Alexandria (or wherever he worked) around 300 BCE — is inaccessible to all seven modes and to any mode whatsoever. This is a trivial point in one sense (we cannot recover any ancient person's inner life) but philosophically significant in another: it suggests that a proof is not identical to the act of discovery. Hardy is right that there is mathematical beauty, but the experiencing of that beauty is strictly first-personal. Mathematical objects may be necessary and eternal, as Plato believed; the act of apprehending them is historically contingent and phenomenally particular.
There is also a limit within each mode that no mode transcends: the reader's understanding. A proof, however perfectly rendered, does not force comprehension. The proof tree requires that you know natural deduction; the Greek requires that you read Greek; the sonnet requires that you hold the argument in mind through the compression of meter. Each mode imposes its own barriers to understanding, which are different from, but no less real than, the barriers of the formal proof.
Jakobson distinguished three types of translation: intralingual (within a language), interlingual (between natural languages), and intersemiotic (between sign systems).13 The other modes are all, on this scheme, intersemiotic translations of the Greek mathematical text into different sign systems: algebra, verse, dialogue, ASCII, Biblical English, phenomenological prose. Jakobson's framework raises the question: can intersemiotic translation preserve propositional content? Can the sonnet carry the same assertion as the proof?
The Wittgensteinian answer would be cautious. For Wittgenstein, a mathematical proof does not primarily convey a proposition; it changes the grammar — it creates a new rule, a new pattern of inference, a new way of going on.14 If this is right, then the question is not whether the sonnet asserts the same thing as the proof, but whether it does the same work: whether it produces the same rule-change in the reader's mathematical practice. Almost certainly it does not. The sonnet may illuminate, may produce aesthetic conviction, may even produce something like intellectual conviction — but it does not install the proof as a new rule of grammar in the way that actually working through the proof does.
And yet the sonnet is not merely decorative. It shows — by showing — that the proof's argumentative structure maps onto a recognizable aesthetic form, and that this mapping is non-trivial. The fact that the argument fits the sonnet form is itself a mathematical observation about the structure of the proof. The fit is not perfect (the sestet is overloaded, as we acknowledged), and the imperfections are informative: they show where the logical structure exceeds the metrical form's capacity to articulate it. Both the fit and the friction constitute, together, a kind of meta-proof about the proof.
"The imperfections of the sonnet are not failures of craft but revelations of structure — they show exactly where the proof's logic exceeds what verse can carry."
Hardy wrote that a mathematical proof should have "unexpectedness combined with inevitability" — the final step should feel both surprising and, in retrospect, the only possible step.15 This is also a description of the best sonnet volta. The mapping between these aesthetic categories suggests that Hardy's notion of mathematical beauty is not merely analogical: the structures that produce aesthetic pleasure in proofs and in sonnets may be instances of a single, more general structural property — call it compressed necessity.
That the same structure appears across mathematical proof, lyric poetry, Socratic dialogue, and biblical narrative is not a coincidence. It suggests that these representational modes are not merely parallel but are co-evolved strategies for encoding, transmitting, and inducing insight in the minds of other people. What Euclid discovered, around 300 BCE, was an entailment relation in the structure of the integers. What seven modes reveal is that this entailment relation can be made manifest — partially, distinctly, with different costs and benefits — in seven different cognitive and cultural media. The medium shapes what can be seen; it does not determine what is there.