Every other passage in Law 47 warns that continuing past your mark risks losing what you gained — new enemies, a reversal, a fall. This one describes a cost that arrives even when everything continues to go well.
Greene: "the moment when you stop has great dramatic import. What comes last sticks in the mind as a kind of exclamation point. There is no better time to stop and walk away than after a victory. Keep going and you risk lessening the effect, even ending up defeated."1
The claim is about composition, not danger. You caught the witness in a contradiction. You then asked eleven more competent, unremarkable questions, and lost none of them. Nothing bad happened. And the thing you achieved is gone anyway, because it was never a fact in the record — it was a shape in someone's memory, and you kept adding material after the shape was finished.
That is a different failure from the rest of the chapter and it needs different handling, because none of the usual warning signs fire. There is no resistance, no counterattack, no moment where it turns. It just quietly stops having been a triumph.
The source is Francis L. Wellman's The Art of Cross-Examination (1913), and it is one of the very few places in the book where Greene quotes a real, named, dated work rather than an unattributed anecdote:
"So many lawyers succeed in catching a witness in a serious contradiction; but, not satisfied with this, go on asking questions, and taper off their examination until the effect upon the jury of their former advantage is lost altogether."2
The effect upon the jury. Wellman is exact about where the loss occurs. The contradiction is still in the transcript. It happened, it is recorded, it can be read back. What has been lost is its position in the only place that decides the case — twelve people's impression of what they watched.
So the currency here is not the achievement. It is the achievement's salience, which is a separate quantity, subject to separate rules, and destroyed by exactly the behaviour that feels most conscientious.
The phrase to sit on is Wellman's "not satisfied with this."
What is the dissatisfaction? Not greed for a bigger win — the contradiction is already fatal. It is doubt. The lawyer is not sure the jury saw it. So he asks another question that circles the same ground, to make it unmistakable, and then another.
Which means the over-questioning is verification behaviour, and this is the mechanism worth carrying out of the passage. The impulse is not aggressive. It is anxious. You are checking that your point landed, and the checking is what unlands it — because every subsequent question is evidence to the jury that the previous one was not enough.
This generalises well past courtrooms. Explaining your own joke. Restating the argument you just won, in different words, in case. Sending the follow-up message that clarifies. Each of those is the same move: an attempt to confirm receipt that instead broadcasts uncertainty about the thing being confirmed.
The confirmation is the retraction. You cannot check whether a point landed without telling the room you are not sure it did.
Slot this against the law's headline instruction — set a goal, and when you reach it, stop3 — and the two do not fit.
You cannot decide beforehand which question will break the witness. The mark here is not a target you named in advance; it is an event you must detect while performing. Wellman says so directly: "be ever on the alert for a good place to stop."2
Ever on the alert, throughout, concurrently. That is a monitoring process running alongside the performance, using the same attention the performance is consuming, and it must produce a decision within seconds of the moment arriving.
So Law 47 contains two remedies with almost nothing in common. One is pre-commitment: name the mark while you are calm, be bound by it when you are not. The other is live detection: recognise, in real time, that the peak just happened, and act on the recognition before the next thing you were going to say comes out of your mouth.
They are not substitutes. Pre-commitment protects against the campaign that runs three months too long. Live detection protects against the eleven extra questions. Most people are far better at one than the other, and Greene never separates them.
"What comes last sticks in the mind as a kind of exclamation point."1
[PLAUSIBLE — needs corroboration] — this converges with well-replicated findings on how episodes are evaluated retrospectively, in which the ending carries weight out of all proportion to its duration and the total quantity of good or bad experience matters remarkably little. Greene cites nothing and offers no mechanism, but the claim is not merely rhetorical.
If that is right, the practical implication is stronger than stop after a win. It is that the ending is not the last part of the thing — it is the part that determines what the whole thing was. A meeting that ran well and closed on an unresolved irritation was a bad meeting. A negotiation that was brutal for six hours and closed with an agreement everyone could live with was a good negotiation.
Which converts the instruction from a caution into a design question. Not when should I stop but what do I want this to have been, and then arranging for that to be the last thing in the room.
Wellman's technique assumes a bounded performance with a curtain: the examination ends, the jury deliberates, the impression is fixed. Almost nothing else works like that.
In an ongoing relationship there is no curtain. You stop at the high point and see the person again on Thursday, and Thursday is now the ending. The salience you preserved is not banked — it decays, and the next interaction overwrites it.
That is the limit of the play and Greene does not mark it. Stopping with a victory is a technique for performances, not for relationships. In an ongoing arrangement the equivalent instruction is nearly opposite: do not engineer high points to withdraw on, because the withdrawal is itself read, and a pattern of leaving on a high becomes a pattern of leaving.
Where it does transfer cleanly: presentations, negotiations with a close, interviews, arguments in front of an audience, any session with a defined end and an evaluating third party. The presence of the third party is the tell — Wellman's mechanism is about the jury, and without a jury there is nothing for the exclamation point to land on.
Of everything in Law 47, this is the piece with the best cost-to-benefit ratio, and it is worth saying plainly.
Not going past your mark militarily requires you to forgo real gains. Measuring your victories in service requires you to under-produce. Switching from force to cunning requires a second instrument you may not have.
Stopping talking after you have made your point costs nothing. There is no forgone gain, no strategic sacrifice, no unbought asset. The eleven extra questions were never going to yield anything; they were the tax on your own uncertainty.
Which suggests the ordering for anyone trying to actually use this chapter: start here. It is the only instruction in Law 47 that is free, applies daily, and can be practised deliberately without anyone else's cooperation.
You are in the meeting, or the call, or the examination, and something just landed. You can feel it — the room adjusted.
The next thing that will happen is that your mouth will keep going. It is already loaded. This is the entire skill: notice the landing and let the loaded sentence die.
Then say nothing, and let the silence sit longer than is comfortable. The discomfort is the price and it is the whole price. Everyone in the room is currently doing your work for you — assembling the meaning of what just happened — and any further contribution from you interrupts it.
If you find yourself wanting to restate the point in different words, name what that is. It is a check on whether they got it, and it will announce that you do not think they did. They got it. If they did not, the second version will not fix it either.
Then close. Not a summary, not a recap — those are the taper. Something short that ends the session while the landing is the most recent thing: That's what I wanted to cover. Stand up.
And if this is a relationship rather than a performance, do the opposite: stay in the room past the high point and let it become ordinary. Engineered exits work once and are read the second time.
Evidence quality: [POPULAR SOURCE] for Greene's framing; the Wellman quotation is unusually well attributed by this book's standards — named author, named work, date (1913) — and the reasoning is a practitioner's, from a still-cited manual. The recency claim is [PLAUSIBLE — needs corroboration] per the convergence noted above; Greene supplies no citation.
Tension: the law's headline instruction assumes a mark set in advance; this passage requires detecting the mark in real time. Both are called stopping at your mark and they are different capabilities. Preserved rather than reconciled — see above.
Open questions: does the technique survive repetition with the same audience? A jury sees you once. A colleague who has watched you land a point and immediately close, four times, may start reading the close itself as the move. Nothing in Wellman or Greene addresses a repeated audience.
This passage sits oddly in the corpus because it is the only place Greene treats power as fundamentally a matter of composition — arrangement, sequence, where the emphasis falls — rather than of position or leverage. Elsewhere the game is about what you hold. Here it is about the order in which things were experienced by someone watching.
That aligns closely with the theatrical material in The 48 Laws — the chapters on court presence, on dramatic effect, on making your accomplishments seem effortless — and the connection is never drawn. Law 47's closing paragraph is a theatre note filed inside a strategy chapter.
The tension is with the corpus's usual view of thoroughness. Several laws recommend completeness, leaving nothing unfinished, crushing entirely. This says the thorough version of a triumph is worse than the abrupt one, and that the desire to be thorough is anxiety in a respectable costume. Greene does not notice that he has just undercut Law 15's argument on a different axis — not finish your enemy versus don't, but the finishing is legible as insecurity.
The Finale as Paragraph Reward — the craft material treats the ending as a structural position with its own function rather than as the place where the material happens to run out, and it names the thing Wellman is protecting: the reader's or listener's sense of having been delivered somewhere. What that adds to the strategic account is the reason the taper is so destructive relative to its size. Eleven mediocre questions after a devastating one do not average out with it; they replace the position the devastating one occupied, because the final slot is not one slot among many but the one that assigns meaning to everything before it. The insight neither frame produces alone: Greene and Wellman both frame the error as dilution — the effect gets weaker, spread thinner. The craft frame says it is not dilution but displacement, which predicts something different and checkable. Under dilution, a few extra questions would cost a little and many would cost a lot, proportionally. Under displacement, the cost is incurred almost entirely at the first extra question, because that is where the last position changes hands, and everything after it is already free. Which is why the practical instruction has to be about a single moment rather than about restraint in general, and why practitioners describe it as a knife-edge rather than as a matter of degree.
Emotional Versus Narrative Memory — the two-track account explains why Wellman's jury behaves as it does, and it corrects the folk model buried in Greene's advice. Narrative memory holds the sequence: the witness was asked X, answered Y, contradicted himself. Emotional memory holds the charge: that was the moment. They are encoded differently, decay differently, and are retrieved differently, and a deliberating jury is running mostly on the second while believing it is running on the first. Wellman's effect upon the jury is a claim about the emotional track specifically — the contradiction is still fully available narratively and has simply stopped carrying charge. That yields the insight neither reaches alone: the eleven extra questions do not erase the memory, they decouple the charge from it, and this predicts a specific and testable asymmetry. A juror asked to recall the examination would still report the contradiction accurately; asked what mattered, they would name something else. Which also explains why the lawyer's verification impulse is exactly backwards. Repeating the point strengthens the narrative track — the part that was never in doubt — while spending the charge that was the only thing at stake.
The testable claim is the displacement model from the craft handshake: that the cost of continuing past a peak is incurred almost entirely at the first extra move rather than accumulating with each one. That predicts a step function rather than a slope, and it is checkable in any recorded performance with an evaluating audience.
The open edge is the repeated-audience question. Wellman's mechanism assumes a jury that sees the technique once. Whether stopping-on-a-victory survives being recognised as a technique by people who watch you regularly is untested, and the ongoing-relationship limit above suggests it does not.