Somebody you respect looks at a thing you own, and then looks away, and then looks at it again.
They say nothing. They are perfectly pleasant. And you spend the rest of the afternoon with the object in your peripheral vision, and by the evening you have worked out what the looking was about, and by the next week the object is gone.
Nobody instructed you. If you asked them about it afterwards they could truthfully say they had not raised it. And the correction was more complete than any conversation would have produced, because you arrived at it yourself and there was nothing to argue with.
That is the most efficient corrective mechanism there is and it is almost entirely undocumented, because it leaves no record. Madhya-līlā 20 records one.
Sanātana Gosvāmī has taken up the life of a mendicant, and he is wearing a valuable woollen blanket — the one substantial possession left from a career as a Nawab's minister.1
Caitanya says nothing about it.
He glances at it. Repeatedly.
And Sanātana reads the glance, and goes and trades the blanket to a Bengali mendicant for a torn quilt.2
Only then does Caitanya speak. And what he says is two things, both after the fact.
"After vanquishing a disease, a good physician does not allow any of the disease to remain."
And: *"It is contradictory to practise mādhukarī and at the same time wear a valuable blanket… one will also become an object for jokes."*3
Everything about this depends on the sequence, and the sequence is deliberate.
The correction is delivered by attention. Not by instruction, not by hint, not by a question. A repeated glance, which carries no proposition and cannot be denied or disputed.
The subject does the work. Sanātana has to notice the glance, infer what it is about, decide whether he agrees, and act — and every one of those steps is his.
And the explanation arrives only after compliance. Which is the part that separates this from ordinary manipulation. The reasons are given, in full, in plain language — after the blanket is gone, when they can no longer function as pressure.
That ordering does something specific. A reason given before an action is an argument, and arguments have opponents. The same reason given afterwards is a confirmation: it tells the person their inference was correct, and it converts an act of compliance into an act of judgement they made themselves.
Which is why they remember it, and why nobody in this scene has anything to resent.
The two explanations are not equivalent, and the second is the interesting one.
The physician analogy is about completeness — vanquish the disease and leave none of it. It is the standard renunciate argument, applies to any residual attachment, and could have been made about anything.
The second is about being laughed at. One will also become an object for jokes.
That is not a spiritual argument at all. It is a claim about how the man will look to other people — a mendicant begging house to house in a rich blanket is a figure with an obvious inconsistency, and the inconsistency will be noticed and will be funny.
And it is the reason with teeth. Sanātana had already accepted the renunciate premise; what he had not thought about was the optics, and the optics are what the corpus's protagonist actually raises.
This corpus is repeatedly and unsentimentally interested in how things read to observers. It stops an ecstasy mid-street because it frightens people; it conceals ecstasy in crowds; it renames a divine state as epilepsy to soldiers.4 The blanket belongs to that family: a concern with the impression, argued frankly, by somebody who does not treat impression-management as beneath him.
And there is a reason ridicule in particular is the lever, which the corpus never states because it never had to.
This order has no coercive instrument at all. It cannot fine anybody, cannot imprison anybody and cannot compel attendance; its members are volunteers who arrived on their own feet and can leave the same way. What it has instead is a reputation, held collectively, which is the only asset a voluntary body owns and the only one every member draws on.
Which makes being laughed at both the last sanction and the first. A mendicant in a rich blanket does not damage himself. He damages the credibility of the practice for everybody currently doing it, and everybody currently doing it will notice.
That is why the second reason bites and the first does not. The physician analogy asks Sanātana to be consistent with a principle; the jokes argument tells him what his inconsistency costs other people — and only one of the two has a constituency standing behind it.
The obvious objection is that a sentence would have been quicker, and it is worth answering because the answer generalises.
A sentence creates a rule. Do not wear valuable blankets is a policy, applies to everybody, invites edge cases, and has to be enforced. It also puts the person's compliance permanently on somebody else's authority.
A glance creates nothing. No rule exists afterwards. Nobody else is bound. And the person who acted has not obeyed anything — they have applied a principle they already held to a case they had not examined.
Which is why the method scales badly and works completely. It requires a relationship, proximity and a shared premise, and where those exist it produces a correction with no residue: no rule to police, no resentment, no record, and no possibility of the person feeling managed.
Where those do not exist it produces nothing at all, or worse — a person who notices they are being looked at, cannot infer why, and concludes something arbitrary.
The corpus supplies both halves of that. It works here because Sanātana and Caitanya share every relevant premise. Nowhere in the book does anybody attempt it on an outsider.
And the method cannot survive the person. A glance requires a body in a room. The moment the man is dead, or merely elsewhere, there is nothing left to read — and a community that corrected by attention for a generation is left with no corrective at all unless it converts the glances into rules.
Which is what commentary is for, and the exchange is worth stating plainly. A rule is what a signal becomes when the sender is no longer available, and the conversion is a real loss: the rule binds everybody, invites edge cases, and puts compliance back on somebody else's authority — every cost the glance was built to avoid.
But the alternative is not the glance. The alternative is nothing. The purport layer's instruction about renunciation and retained comforts is this scene surviving in the only form that travels — a worse instrument, and the reason anybody outside that room ever learned anything from it.
There is a cost and the corpus does not name it, though it demonstrates it constantly.
A person who corrects by attention can never be held to it. Caitanya said nothing; if Sanātana had traded the blanket for the wrong reason, or had not noticed, or had noticed and been hurt, there would have been no instruction to point at and no exchange to revisit.
That is a genuine asymmetry and it runs one way. The subject bears all the interpretive risk.
And it produces the household's characteristic condition, which this build has now documented at length: an entire community reading a man for signals, engineering around his silences, and inferring his wishes from his attention.5 The blanket is that arrangement working perfectly. The two-year obstruction of the Vṛndāvana journey is the same arrangement producing a coalition that overrides him.
Correction by attention trains people to watch you, and once they are watching they will act on readings you did not intend and cannot correct — because you established, by using this method, that your attention is instruction.
Layer A gives the blanket, the glances, the trade, and the two explanations in their order.6
Layer B expands the passage into instruction on the incompatibility of renunciation with retained comforts, and on the necessity of complete rather than partial detachment.7
Worth marking what the commentary loses. Layer B keeps the physician analogy and drops the jokes. The completeness argument is doctrinally portable and generalises to any reader; the observation about how a man begging in a valuable blanket will look to a village is specific, worldly, and slightly undignified.
But it is the one that actually moved the blanket. A reader working from the purports gets a lesson about detachment. A reader working from the verse gets a man being told, kindly, that he looks ridiculous — and only the second explains why the scene ends the way it does.
You want somebody to change something and you have a relationship with them. Look at it and say nothing. The correction they arrive at themselves is more complete than one you deliver, and it leaves no rule, no record and no resentment. Then explain afterwards, in full — the explanation confirms their judgement instead of overriding it.
You are being looked at and you cannot tell why. Ask. The method fails silently in exactly this case, and the failure mode is that you invent a reading and act on it, which is worse for both parties than a question would have been.
You are giving reasons. Notice whether they come before or after. Before is an argument and has an opponent; after is a confirmation and has none — and the same sentence does completely different work depending on which side of the action it sits.
You correct by attention and people have started watching you. They will now act on readings you did not intend. That is not their error; it is the mechanism you installed, and the only way out is to start saying things out loud again, which is slower and which you will resent.
The evidence is a short, fully-ordered scene: an object, repeated glances, an unprompted trade, and two explanations delivered afterwards.8
The tension the source doesn't close: correction by attention is used on an intimate and never on anybody else. The corpus's method with outsiders is argument, and with insiders is signal — and it never asks what happens to somebody who is neither.
A second one. The method's success requires the community to read him constantly, and the corpus separately documents that community overriding his stated wishes on the strength of such readings. The blanket and the two-year obstruction are the same capability.
Open: whether Sanātana knew he was being corrected or thought he had noticed something himself. The corpus records the trade and not the reasoning, and the difference is the whole question of whether this is teaching or influence.
This page and Care as Containment are the same capability pointed in opposite directions. Here he reads a devotee and the devotee acts; there the household reads him and acts on his behalf. The corpus approves of both and never notices it has described one skill.
Against The Bengali Who Drank the Footwater it is the corpus's gentlest correction beside its most theatrical. One is a glance and a private trade; the other is performed anger, physical ejection and a scripted pardon — and the difference is entirely whether there was an audience whose expectations needed setting.
And it belongs beside The Muslim Mendicant at the Door, which is the same man days earlier being told to shave. Two corrections of Sanātana's appearance in one chapter-range, one delivered as instruction and one as attention, and both about how he will be received.
Gestures and Unconscious Signals — that page's subject is the information leaking out of a person continuously below the level of speech: where the eyes go, what the hands do, what the body orients toward, and the fact that observers read it accurately without being able to say what they read.
The blanket is that channel used deliberately, and the pairing isolates the thing that makes it work.
Signal-reading literature is written from the observer's side. Its interest is in extracting information from somebody who is not choosing to send it, and its ethical texture comes from the asymmetry: you know more about them than they meant you to.
Here the sender is choosing. The glance is not a leak; it is a message, sent on the channel usually reserved for leaks — and it derives its whole force from that. A message on the leak channel is received as an observation rather than as a demand, and the recipient credits themselves with having noticed it.
Which gives the finding neither page reaches. A deliberate signal sent on an involuntary channel gets the credibility of a leak and the precision of a statement, and that combination is the most powerful corrective instrument available between two people who know each other.
It is also unfalsifiable and unaccountable, for exactly the same reason. The sender can always say they were only looking — and in most cases they will believe it, because the line between noticing something and letting somebody see you notice it is not one anybody can locate from inside.
Sharpest implication. A correction delivered by attention and explained only after compliance is the most complete one available: it leaves no rule, no resentment and no record, and the person credits themselves with the judgement. Its price is that it trains everybody near you to read your attention as instruction — and they will then act on readings you did not intend, which is not their mistake but the mechanism working as installed.
Generative questions.