Two people agree something, out loud, sincerely, with nobody else in the room.
Then the circumstances change. One of them goes home to a family with a view, or a partner with a view, or a set of obligations that were always there and were not being weighed at the moment of agreeing. And the agreement quietly becomes the sort of thing that is remembered differently by each side.
There is no remedy for this and everybody knows it. The remedy people reach for instead is a witness — somebody present, somebody who will still be there later, somebody whose memory is not owned by either party. That is what witnesses are for, and it is why the institution long predates any court that could compel one to appear.
Madhya-līlā 5 takes that arrangement and does something extraordinary with it: it makes a stone statue the witness, and then has to work out what happens when the statue is called.
An elderly brāhmaṇa on pilgrimage is served, at length and with real devotion, by a young one. In gratitude he promises the young man his daughter in marriage.1
The young man's response is not gratitude. It is procedural.
"If this brāhmaṇa later hesitates to give me his daughter, my dear Lord, I shall call on You as a witness."2
He is standing in front of the Gopāla Deity at Vṛndāvana, and he names it as the witness to the promise — in advance, out loud, at the moment the promise is made rather than afterwards when it is in dispute.
And he turns out to have been right to. The elder goes home and his family objects: the young man is of lower standing, and the match is refused. The elder's son goes further and lodges a counter-accusation — that the young man drugged his father with dhuturā and robbed him, which is the standard charge of the period and the same one levelled at another young brāhmaṇa elsewhere in the corpus.3
The village assembles. The assembly acts as mediators. And the dispute is resolved in a way nobody reading a devotional text expects.
The village writes it down.
The terms are recorded "in black and white," and the signatures of both parties are taken.4
Stop on that. This is a chapter about a walking Deity, in a hagiography, and the mechanism by which the dispute is actually managed is a written instrument with executed signatures, produced by a lay assembly acting as mediators, in a village, with no court involved.
The corpus does not remark on it. It reports the paperwork the way it reports the nine hundred pots at Govardhana — as the obvious thing that was done.5
Which tells you what this community actually trusted. The theology says the god is present, sees everything and is the guarantor of truth. The practice, at the point where a real dispute between two families had to be settled, reached for a document and two signatures — and then used the theology as the content of the document rather than as a substitute for it.
A written contract whose named term is a divine appearance is still a written contract. The form is legal and only the consideration is miraculous.
Now the part that makes the episode worth a page.
The son agrees to the terms precisely because he is certain the witness cannot appear.6
He is described as an atheist, and he behaves like one, consistently and rationally. He is being asked to accept a settlement conditional on a stone Deity walking several hundred miles to give evidence. From where he stands that condition is not a risk; it is a guarantee, and signing is free.
So he signs.
And this is a general mechanism worth naming, because it does not require anybody to be religious. A person who is confident a condition cannot be met will accept it, and in accepting it they convert their confidence into a binding obligation. The condition costs them nothing to grant and everything to be wrong about.
Which means certainty is the thing being harvested, not belief. The believer would have negotiated — would have hedged, or refused the term, or asked what counts as an appearance. The sceptic gives it away entire, because to a sceptic an impossible clause is a dead clause.
The corpus tells this as a story about a god who shows up. It is at least as much a story about a man whose confidence in his own worldview was the instrument used to bind him, and the text is not naive about this: it makes the son's certainty explicit and makes it the reason the settlement was reachable at all.
Then the young man goes to Vṛndāvana and asks the Deity to come, and the Deity objects.
"I've never heard of a Deity's walking from one place to another."7
That is a reasonable objection and it is an argument from category: statues do not walk, this is a statue, therefore it does not walk. It is the same argument the atheist son made, offered from inside.
And the reply is the sharpest single sentence in the chapter:
*"That is true, but how is it that You are speaking to me, although You are a Deity?"*8
The god is out-argued from his own precedent. The move is not you are omnipotent and not scripture says you may — either of which would have been available and neither of which would have worked, because both are contestable and both are external. The move is: you have already broken this category once, in this conversation, on your own initiative, and I am only asking for the second instance of something you have just done.
That is a complete and portable technique and it deserves stating in general form. An exception already granted is the cheapest possible argument for the next one, because it cannot be denied without withdrawing the first — and the first has usually already been enjoyed, relied on, or in this case is literally in progress while the objection is being made.
Notice what makes it irrefutable here. The objection and the counter-example occupy the same moment. The Deity is not being reminded of something he did years ago and might dispute; he is speaking, now, in the act of saying that statues do not do things. The evidence is the objection.
Which is why the corpus can let the god lose the argument without the god losing standing. He does not concede that he was wrong; he concedes that the young man is right, and then sets his own terms — walk behind, do not look back or I stop where I stand, you will know Me by the sound of My ankle bells, and cook one kilogram of rice daily and offer it.9
An Orpheus taboo with a catering clause. The corpus supplies both in the same breath and finds neither strange.
The last element is easy to skim and it changes the register of the whole thing.
The young man says outright that the marriage is not what he is after.
"I am simply thinking that the brāhmaṇa has broken his promise, and that is giving me great pain."10
He is not litigating for a wife. He is litigating because a promise was broken, and the breach is the injury. That is a claim about what he thinks was damaged — not his interest, but the standing of a spoken word — and it is the reason he is willing to walk to Vṛndāvana and back for a settlement he could have abandoned at any point.
And then he does something almost nobody in a dispute does. He defends the man who broke the promise.
"He had no desire to break his promise, but fearing that his kinsmen would commit suicide, he deviated from the truth."11
Read the structure. He concedes the breach, supplies the motive, and rates the motive as sufficient to explain the conduct without excusing it. The elder is not a liar; he is a man under a coercion the young man can see and name, and the naming is done by the injured party rather than by the defence.
Which puts the young man in an unusual position: he is simultaneously the only person insisting the promise be kept and the only person who has publicly understood why it was not. The corpus records both and connects them to nothing.
Layer A supplies the promise, the invocation of the witness, the family's objection, the dhuturā accusation, the assembly, the written settlement with signatures, the objection and reply at Vṛndāvana, and the conditions of the walk.12
Layer B expands the episode into an account of the Deity's non-different status — that the form is not a symbol of the Lord but the Lord — and of the fate of the atheist who is converted by the appearance.13
Worth marking the divergence, because it is a clean case. Layer A's interest is procedural throughout: who promised, who objected, what was written, who signed, what conditions were set. Layer B's interest is ontological, and it treats the procedure as scaffolding around the miracle.
Both are honest to their own purpose. But the reader who arrives through the commentary will learn that a Deity is fully the Lord and will not learn that a village assembly resolved a marriage dispute with a signed document, which is the single most historically interesting sentence in the chapter.
You are making an agreement that nothing can enforce. Name the witness at the moment of agreement, not later. The young man's first act on receiving the promise was to specify how it would be tested, out loud, in front of the other party — and a witness named in advance is a term of the agreement, while a witness produced afterwards is an accusation.
Somebody accepts a condition you thought they would fight. Ask what they think is impossible about it. Free acceptance almost always means the person has priced the condition at zero, and the thing you have obtained is not their agreement but their confidence — which is a much weaker asset than it looks and a much stronger one to hold.
You want an exception. Ask for the second one, not the first. Find the thing the other party has already done — ideally something they are doing while refusing you — and ask only that it be repeated. Refusing the second instance requires disowning the first, which almost nobody will do.
You are the injured party. Say what was actually damaged. You broke a promise and that is what hurts is a different and much harder claim to deflect than I want what I was promised, because it cannot be settled by delivery and it does not put you in the position of a person pursuing an interest.
The evidence is a continuous narrative with an unusual density of procedure: a named condition, an accusation, an assembly, a written instrument, executed signatures, an argument with a counter-example, and a set of stated conditions.14
The tension the source doesn't close: the community's theology says the god is the guarantor and its practice reached for a document. Both are on the page, adjacent, and the corpus treats the paperwork as unremarkable rather than as a comment on the theology.
A second one. The settlement was reachable because one party was certain the divine term was inert. The episode's happy outcome therefore depends on the son's disbelief — the corpus needs the atheist to be a good atheist, and never notices that its miracle was underwritten by scepticism.
Open: what the daughter thought. She is the consideration in a written contract between two families and an assembly, she is named nowhere, and no part of the narrative turns on anything she said.
This page and The Excavation of Gopāla are the same Deity in two operating modes. There he issues instructions by dream and a region builds him a temple; here he is summoned as a witness and argued into walking. In both he is an agent with preferences, and in both the human work — the pots, the paperwork — is what the chapter actually spends its lines on.
Against The Deity as Mobile Asset it is the third of the three relocations and the only one narrated as a favour rather than a rescue. The god walks here because he was out-argued; elsewhere he is buried, transferred and carried off, and in every case the movement is attributed to his own choosing.
And it belongs beside The Epilepsy Cover Story, where the same dhuturā charge is the soldiers' inference on the road to Prayāga. Twice in this corpus somebody is accused of drugging and robbing a traveller, and both times the accusation is the competent reading of the available evidence and both times it is false — which is the corpus's most repeated observation about how people get things wrong, made twice and named neither time.
Self-Control and Commitment Devices — that page's mechanism is that the reliable way to secure future conduct is not to strengthen resolve but to remove the option: structure the situation now so that the version of you who will be tempted later has no move available. The willpower problem is sidestepped rather than won.
The witness invoked at Vṛndāvana is that device aimed outward instead of inward, and the pairing exposes something the self-control literature does not have to face.
A commitment device works on your future self because you control the environment. You can throw out the cookies. You can lock the account. The binding is real because the binding is yours to install.
A witness is a commitment device you install on somebody else, and it has no enforcement at all — a witness cannot compel, only testify. What it does is convert a private promise into a public fact, so that breaking it costs standing rather than merely conscience.
Which gives the finding neither page reaches. The strongest binding available between two parties is not a penalty but an audience, and the whole art is in specifying the audience at the moment of agreement, when naming it is a formality, rather than afterwards, when naming it is an attack.
And this chapter is candid about the failure mode. The witness the young man named could not, on any ordinary account, appear — so the arrangement had exactly the weakness a witness always has, which is that it binds only those who expect it to be checked. The atheist son's certainty that it would not be checked is the reason he signed, and the corpus's resolution is a miracle, which is precisely what an arrangement in that condition would require.
Sharpest implication. An exception already granted is the cheapest argument for the next one, and it is irrefutable when the first exception is still in progress — how is it that You are speaking to me, although You are a Deity? Meanwhile the binding force of a witness comes not from enforcement but from audience, and it therefore holds only over people who expect to be checked. Both halves of this chapter are about obligations that nothing can compel, and both are settled by somebody's confidence being wrong.
Generative questions.