UNITED STATES of America, Plaintiff-Appellee, v. Anthony D. EDWARDS, Defendant-Appellant, 581 F.3d 604

POSNER, Circuit Judge. The defendant was convicted by a jury of distributing 5 grams or more of crack and was sentenced to 108 months in prison. His appeal raises several issues. After being arrested and jailed, he was given the Miranda warnings, and after agreeing to waive his Miranda rights was questioned for a quarter of an hour or so and then returned to his cell. Thirty to forty minutes after the waiver — which is to say roughly fifteen to twenty-five minutes after completion of the questioning — he was returned to the interview room for further questioning by another agent.

Before beginning, the agent showed the defendant the waiver form he had signed before the first round of questioning and asked him whether he understood his rights. He replied that he did. The form made clear that he could stop the questioning at any time. But he argues that the admissions he made during the second round of questioning should not have been placed in evidence at the trial because the Miranda warnings had not been recited to him before the second round began.

The defendant asks us to adopt a doctrine of “staleness” that would require re-administering the Miranda warnings after any break in an interrogation — even, as in this case, a very short one — if there is any reason to think that the person questioned may have forgotten or misunderstood the warnings or thought they had lapsed or been unable to claim them because of new pressures brought to bear on him after the break, though if his statement was coerced this would be an independent ground for suppression — coerced confessions were inadmissible long before the Miranda case.

The defendant points out that the second interrogation was conducted by different officers from the first one and that he made inculpatory statements only at the second one. And he argues that he was in a frightened, emotional state throughout the entire period of the interrogations. The form that he was read, and signed, included the statement that “if you decide to answer questions now without a lawyer present, you will still have the right to stop answering at any time.

You also have the right to stop answering at any time until you talk to a lawyer.” So if the defendant had not wanted to be questioned the second time, he had only to refuse. Of course during the 30 or 40 minutes that elapsed between his signing the waiver form and the second interrogation, he might have forgotten that he had the right to clam up even though he had answered questions at the first one. But he might, for that matter, have forgotten that he had that right if the questioning had lasted for 30 or 40 minutes after he was informed of his rights.

The logic of his argument is that the Miranda warnings should be repeated periodically in the course of protracted questioning. But such reiteration would convey to the defendant a suggestion that he not waive his Miranda rights; it would be like saying “Are you really sure you want the questioning to continue?” The cases do not require that the warnings be repeated after an interruption in the questioning, e.g., United States ex rel. Patton v. Thieret, 791 F.2d 543, 547-48 (7th Cir.1986); United States v. Ferrer-Montoya, 483 F.3d 565, 569 (8th Cir.2007) (per curiam); United States v.

Rodriguez-Preciado, 399 F.3d 1118, 1128-29 (9th Cir. 2005); see also Wyrick v. Fields, 459 U.S. 42, 48-49, 103 S.Ct. 394, 74 L.Ed.2d 214 (1982) (per curiam), even if the interruption is much longer than it was in this case. See, e.g., United, States v. Diaz, 814 F.2d 454, 460 and n. 6 (7th Cir.1987) (several hours); United States ex rel. Henne v. Fike, 563 F.2d 809, 813-14 (7th Cir.1977) (per curiam) (nine hours); People v. Dela Pena, 72 F.3d 767, 769-70 (9th Cir.1995) (nearly fifteen hours); Stumes v. Solem, 752 F.2d 317, 320 (8th Cir.1985) (nearly five hours); Jarrell v.

Balkcom, 735 F.2d 1242, 1253-54 (11th Cir.1984) (three hours). In Thieret the suspect was placed in a holding cell between the warnings and the defendant’s waiver of his Miranda rights. In Diaz the warnings were given at the hotel where the suspect was arrested and his inculpatory statements came during the subsequent booking. In Fike the warnings were given in the evening and the statements the following morning, and the warnings and the interrogation were by different officers, as they were in Jarrell and in the present case. In United States v.

Pruden, 398 F.3d 241, 247-48 (3d Cir.2005), roughly 20 hours and a change of location intervened between warnings and statement and the defendant was merely reminded before he made the statement that he had received the warnings the previous afternoon. The defendant points to a pair of state court cases and one district court case as contradicting the decisions we have cited, but the intervals in those cases were much longer than in the present case. Commonwealth v. Wideman, 460 Pa. 699, 334 A.2d 594, 598-99 (1975) (12 hours); Commonwealth v.

Riggins, 451 Pa. 519, 304 A.2d 473, 477-78 (1973) (17 hours); United States v. Jones, 147 F.Supp.2d 752, 761-62 (E.D.Mich.2001) (18 hours). Vagueness is the bane of legal reasoning. This case presents several examples, beginning with “staleness,” a word with no proper application to a statement. What is a “stale statement”? Interrogators might try to negate the Miranda warnings; had the second interrogator in this case told the defendant that he must answer his questions because a Miranda waiver is forever the answers could not have been used in evidence. See Hart v.

Attorney General, 323 F.3d 884, 894-95 (11th Cir.2003); United States v. Beale, 921 F.2d 1412, 1435 (11th Cir.1991); United States v. San Juan-Cruz, 314 F.3d 384, 387-89 (9th Cir.2002). And likewise if, as in Ex parte J.D.H., 797 So.2d 1130 (Ala. 2001), so much time had elapsed between the rounds of questioning — 16 days in that case — that the agent should have realized that the defendant might well have forgotten the warnings, and specifically the paragraph tucked into the Miranda form that entitles a suspect to interrupt the questioning at any time and summon a lawyer.

Yet even in J.D.H. the court emphasized circumstances beyond the long delay between interrogations in deciding that the inculpatory statement should have been suppressed. The practical question is not whether Miranda warnings given to a defendant became “stale,” or, though the courts love the phrase, whether the “totality of the circumstances” indicates that the inculpatory statement was made knowingly. It is whether the defendant when he gave the statement didn’t realize he had a right to remain silent.

The Miranda form told him he had that right, and the presumption should be that he would remember this even if some time had elapsed between his receiving the warnings and undergoing the questioning that elicited the inculpatory statement. The cases do not speak in terms of a presumption but that is the practical effect of their reluctance to attach dispositive weight to a break in questioning, even when the break is protracted and other circumstances might have made it less likely that the defendant would remember that he could stop the questioning at any time.

The presumption can be rebutted but was not in this case. The defendant makes the unrelated argument that evidence of prior criminal activity by him should not have been admissible at the trial. Rule 404(

b) of the evidence rules forbids the use of such evidence to establish a person’s propensity to commit crimes but permits it “for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.” The government’s principal witness was a drug dealer named Beagle who testified that, working as a government informant, he had arranged to make a controlled purchase of drugs from the defendant.

There was conflicting testimony about whether the defendant had drugs with him when arrested upon arriving at Beagle’s house, where the purchase was to take place. But Beagle testified in detail about the procedures used when the defendant had sold drugs to him on previous occasions- — how the sale would be set up, where it would take place, and so forth— and that he had followed the same procedures in the transaction for which the defendant was being prosecuted, though the defendant was arrested before the sale took place.

The earlier sales were of course criminal too, but they were not charged in the present case. The parties duel over whether the evidence of these sales was nevertheless admissible under the “intricately related” or “inextricably intertwined” gloss on Rule 404(b), despite our criticism of these tongue-twisting formulas in United States v. Taylor, 522 F.3d 731 (7th Cir.2008).

The rule lists purposes for which evidence of prior crimes may be presented, but the list is illustrative rather than exhaustive because the rule forbids only the use of prior-crimes evidence to show that since the defendant committed crimes in the past, probably he committed the crime of which he is now accused, or that since he’s a criminal the jury might as well resolve any doubts about his guilt against him. To satisfy Rule 404(b), all the government need show is a purpose other than to establish the defendant’s propensity to commit crimes. Id. at 735-36.

The fact that prior-crimes evidence is “inextricably intertwined” with or “intricately related” to (and are these the same tests or different?- — who knows?) the charge in the case at hand is neither here nor there, if indeed any meaning can be assigned to such terms. Some cases restate the test for admissibility as whether the prior-crimes evidence is needed to “complete the story” or “tell a complete story” of the crime with which the defendant is charged, e.g., United States v. Gilmer, 534 F.3d 696, 705 (7th Cir.2008); United States v. Ramirez, 45 F.3d 1096, 1102 (7th Cir.1995); United States v.

Quinones, 511 F.3d 289, 309 (2d Cir.2007), or to fill a “conceptual void.” E. g., United States v. Gilmer, supra, 534 F.3d at 705; United States v. Wantuch, 525 F.3d 505, 517 (7th Cir.2008); United States v. Gougis, 432 F.3d 735, 742 (7th Cir.2005). But these formulas, too, lack clarity. We recall Holmes’s admonition to think things not words, by which he meant that the words in which judges state a legal doctrine should be transparent to the goals or policies or concerns that animate the doctrine.

So in this case the focus of inquiry should be on whether the prior-crimes evidence is relevant (other than to show propensity, which may be relevant to guilt, but is impermissible as evidence) to an issue in the case, and, if so, whether the probative weight of the evidence is nevertheless substantially outweighed by its prejudicial effect or by its propensity to confuse or mislead the jury. Fed.R.Evid. 403; see Old Chief v. United States, 519 U.S. 172, 182, 117 S.Ct. 644, 136 L.Ed.2d 574 (1997); United States v.

Dunkin, 438 F.3d 778, 780 (7th Cir.2006); Advisory Committee Note to 1972 Proposed Rules, Rule 404(b). So did Beagle’s previous drug purchases from the defendant bolster the government’s case that it had arrested the defendant in the course of a drug sale? They did. There was disagreement over whether the defendant had had drugs in his possession when he was arrested. The previous purchases substantiated Beagle’s testimony that he had arranged the meeting with the defendant to buy drugs from him.

His past transactions with the defendant had followed a familiar pattern: He would call the defendant to set up the meeting. The defendant would usually be late, prompting a second phone call by Beagle. The defendant would park in the alley behind Beagle’s house, and if Beagle didn’t approach the car immediately upon his arrival the defendant would drive away and Beagle would have to call the defendant on the latter’s cell phone to summon him back. When the defendant returned, he would give Beagle the drugs.

This scenario unfolded as usual in the run up to the defendant’s arrest except for Beagle’s not coming out of his house to take possession of the drugs; the defendant was arrested before that final step. All prior-crimes evidence is prejudicial; otherwise there would be no need for Rule 404(b). But the judge did not abuse his discretion in ruling that the admission of the evidence in this case passed muster, for without it the jury might have thought that Beagle had fabricated a planned drug sale in order to curry favor with the government.

Another unfortunate bit of legal jargon has insinuated itself into the appeal. The defendant accuses the government of having “vouched for” the truthfulness of two of its witnesses — Beagle and an officer who testified that he saw a bag containing an off-white rock-like substance fly off the defendant’s person during the arrest and that when the defendant was later removed from the police car the officer saw additional bags containing a similar-looking substance lying at the defendant’s feet.

The prosecutor in his closing argument asked the jury rhetorically “what possible reason does he [the officer] have to risk his career?” by testifying falsely. And he asked Beagle on direct examination what would happen to his plea agreement if he didn’t testify truthfully and he replied that “they could throw it out,” in which event he would “be looking at more time” in prison. What “vouching for” means in this context is telling or hinting to the jury that the prosecutor has reasons unknown to it for believing that a government witness is telling the truth. E.g., United States v.

Young, 470 U.S. 1, 18-19, 105 S.Ct. 1038, 84 L.Ed.2d 1 (1985); United States v. Morris, 498 F.3d 634, 642 (7th Cir.2007); United States v. Brown, 508 F.3d 1066, 1075-76 (D.C.Cir.2007). A number of cases suggest that there is another form of vouching as well — the prosecutor’s expressing his personal belief in the witness’s truthfulness, thus “placing] the prestige of the government behind the witness.” E.g., United States v. Anderson, 303 F.3d 847, 855 (7th Cir.2002). It is unclear how different that is from the first form of vouching.

The jurors know that the prosecutor wants a conviction; otherwise the charges would have been dismissed. Unless jurors revere prosecutors, the only reason for a juror to accept the prosecutor’s expression of a personal belief in the witness’s truthfulness is that the juror thinks the belief is based on something the prosecutor knows and the jurors do not — -there is no improper “vouching” if the prosecutor merely reminds the jury of evidence presented at the trial that tends to show that a witness was telling the truth.

Improper vouching is trying to bolster a witness’s believability with “evidence” that was not presented and may well not exist. The question the prosecutor asked Beagle about the possible consequences of his lying was innocent. It merely probed Beagle’s understanding of the consequences of lying on the stand. The prosecutor was not implying that he had secret information that Beagle would have been afraid to lie because of the consequences if he did. The case law allows the government to present evidence that plea deals are conditioned on truthful testimony. E. g., United States v.

Morris, supra, 498 F. 3d at 642-43; United States v. Renteria, 106 F.3d 765, 767 (7th Cir.1997); United States v. Hansen, 434 F.3d 92, 101 (1st Cir.2006). One case breaks from this pattern, United States v. Brooks, 508 F.3d 1205, 1210 (9th Cir.2007), but the other cases on which the defendant relies are distinguishable from the present one because in them the prosecutor went beyond just asking the defendant what he thought the consequences of lying would be. See United States v. Carroll, 26 F.3d 1380, 1389 (6th Cir.1994); United States v. Kerr, 981 F.2d 1050, 1053 (9th Cir.1992); United States v.

Francis, 170 F.3d 546, 550-51 (6th Cir.1999). The prosecutor’s rhetorical question about the agent’s jeopardizing his career by lying about the drugs he found in order to frame the defendant presents a more difficult issue. It could be thought just an appeal to the jurors’ common sense. Jurors know that a witness takes an oath to tell the truth, and they doubtless have heard of the crime of perjury, so they might wonder on their own what motive a police officer would have for lying under oath, since it could get him into trouble.

But when the prosecutor explicitly invites such a speculation, jurors may infer that the government fires officers who lie under oath (even though they are lying to help the government’s case), or prosecutes them, so that an officer who lies is indeed jeopardizing his career or his liberty even if he thinks he is helping the prosecution- — and no evidence to that effect was presented. We do not want to encourage lawyers to bring in a parade of witnesses to testify to how often police officers lose social or professional standing because of dishonest testimony.

Were such evidence required, testimony that Beagle stood to lose the benefit of his plea bargain could also be thought impermissible vouching because no evidence was presented on how frequently plea bargains are renegotiated or withdrawn after the government’s informant testifies. An entire category of argument — that the jury should ask what a witness can lose by lying, and should believe those witnesses with a lot to lose and disbelieve those (such as defendants) who can get off the hook by perjury — might be ruled out.

For there is never evidence about how often these gains and losses occur and whether this understanding of human motivation is supported by social-science research. Still, there is a difference between the two questions — the actual to Beagle and the rhetorical to the officer. The question put to Beagle was merely intended to elicit the fact that his plea agreement had been conditioned on his testifying truthfully. Not even a glancing reference was made to the probability that Beagle would be punished if he lied; an estimate of that probability was left to the common sense of the jury.

The question about the officer — a question put not to him but to the jury and unmistakably rhetorical in character (the jury was not being invited to answer it)- — • “what possible reason does he have to risk his career?” — implied that the prosecutor had an undisclosed reason to believe that the consequences if the officer lied would be sufficiently grave to guarantee that he would testify truthfully. Perhaps impelled by cases like United States v. Johnson-Dix, 54 F.3d 1295, 1304-05 (7th Cir.1995); United States v. Badger, 983 F.2d 1443, 1451 (7th Cir.1993), and United States v.

Boyd, 54 F.3d 868, 871-72 (D.C.Cir.1995), which disapprove of comments that make such an insinuation, the government confesses that the prosecutor’s comment about the officer’s motives was improper, but points out that it was harmless. The critical evidence of the defendant’s having drugs with him in his encounter with Beagle was the latter’s testimony. It is true that Beagle, huddled in his basement while the defendant was being arrested, did not see the drugs.

But his testimony that he had arranged to buy drugs from the defendant was strongly supported not only by his testimony about his previous dealing with the defendant but also by records of more than a hundred phone calls from Beagle’s home telephone to the defendant’s cell phone and the defendant’s confession at the police station, which he now denies having made, that it was indeed a drug deal that the police interrupted. The last issue concerns the sentence. When arrested, the defendant had $765 in cash on his person. The judge inferred that he had received the money in a previous sale of crack cocaine.

On this assumption, the judge had solid grounds for adding 12.75 grams to the amount of crack that other evidence (including evidence of 10 grams in prior sales to Beagle) showed that the defendant either had sold or had possessed with intent to sell. The result of the addition (since the supposed earlier sale was, if it really took place, “relevant conduct” within the meaning of the sentencing guidelines) was a higher guidelines sentencing range (108 to 135 months). The defendant testified that he had received the money from the sale of a minivan.

But he also testified that the minivan had no license plates, that title had never been transferred to him, and that the vehicle had never been registered in his name; and this made it impossible to verify his having sold, or for that matter ever owned or possessed, a minivan. And so the prosecutor asked the judge to disbelieve the defendant’s story, and the judge obliged. That was fine, as far as it went. But the defendant argues that even if his testimony was false, the prosecution should have been required to present evidence of what the true source of the money was.

We think that’s true in this case, though not because it is always improper to satisfy a burden of proof by discrediting an opposing party’s evidence. Suppose it were certain that the $765 was the proceeds of the sale either of a Luna moth or of a minivan. If the seller testified that it was a minivan, and his testimony was discredited, the trier of fact could infer that the object that had been sold was the moth, for there would be no alternative hypothesis. United States v.

Hyde, 448 F.2d 815, 831-32 (5th Cir.1971), discussed a situation in which it was known that one of two criminal defendants was guilty of a crime, so proof of one defendant’s guilt exonerated the other. And Clausen v. M/V New Carissa, 339 F.3d 1049, 1057 (9th Cir.2003), allowed evidence of “differential diagnosis,” a technique for identifying the cause of a medical problem by eliminating all the alternative possible causes. The falsity of the defendant’s testimony makes reasonably clear that the $765 was proceeds of an illegal transaction of some sort, but does not show that it was proceeds from selling crack.

For all one knows, the defendant sold other illegal drugs (he had been convicted in the past of possession of marijuana) or other contraband, such as guns, but did not want to acknowledge other illegal behavior, which he might have thought would get him into even worse trouble than he was in. In this case as generally, the fact that a witness lies about one thing doesn’t automatically invalidate all his testimony. E.g., United States v. Reed, 297 F.3d 787, 789 (8th Cir.2002); United States v. Urban, 404 F.3d 754, 782 (3d Cir.2005); Janigan v. Taylor, 344 F.2d 781, 784 (1st Cir.1965).

The maxim falsus in uno, falsus in omnibus is no longer followed, when understood as a rule that a trier of fact may or must disbelieve the entirety of a witness’s testimony if he disbelieves any part of it. Kadia v. Gonzales, 501 F.3d 817, 821 (7th Cir.2007); Allen v. Chicago Transit Authority, 317 F.3d 696, 703 (7th Cir.2003); Piraino v. International Orientation Resources, Inc., 137 F.3d 987, 991 n. 2 (7th Cir.1998); Lambert v. Blackwell, 387 F.3d 210, 256 (3d Cir.2004); United States v. Weinstein, 452 F.2d 704, 713-14 (2d Cir.1971).

As we explained in the Kadia case, “anyone who has ever tried a case or presided as a judge at a trial knows that witnesses are prone to fudge, to fumble, to misspeak, to misstate, to exaggerate. If any such pratfall warranted disbelieving a witness’s entire testimony, few trials would get all the way to judgment.” 501 F.3d at 821. Rather, the trier of fact must consider whether, as in United States v.

Connolly, 504 F.3d 206, 215-16 (1st Cir.2007), particular falsehoods in a witness’s testimony so undermine his credibility as to warrant disbelieving the rest of his testimony — or a critical part, such as, in the present case, the defendant’s denial that the cash found on him when he was arrested was the proceeds of a sale of crack. It thus is only the automatic inference from disbelief in one part of a witness’s testimony to disbelief in the rest that the modern cases reject, with the occasional exception, such as United States v.

Jackson, 3 F.3d 506, 510-11 (1st Cir.1993), which upheld the district judge’s inferring that cash was proceeds of a sale of cocaine because he disbelieved evidence that it had a legal source. But Jackson is inconsistent with United States v. Sandridge, 385 F.3d 1032, 1037-38 (6th Cir.2004), which rejected this form of automatic inference. The district judge gave no reason for his belief that the $765 had to be proceeds of selling crack. It’s not as if $765 were the “list price” of some standard quantity of crack. There was no evidence of that. Neither, as in United States v.

Sepulveda, 102 F.3d 1313, 1317-19 (1st Cir.1996), were marked bills from a known drug transaction found. Nor, as in United States v. Keszthelyi, 308 F.3d 557, 577-78 (6th Cir. 2002), was the amount of money found so great that, given the defendant’s financial circumstances, it could have derived only from selling crack. There is a further problem.

The $765 may have been money that the defendant had been paid by Beagle for previous sales, in which event there was double counting, because the entire estimated quantity of the crack that he had sold Beagle previously was separately counted in figuring his sentence (the 10 grams we mentioned).

The government argues that since Beagle had dealt with the defendant for only five months, but the defendant admitted having dealt crack for at least eight months, his “history of selling crack cocaine includes at least three months without any sales to Beagle — thus ensuring that the $765 did not represent the proceeds of crack sales to Beagle” (emphasis in original). But $765 in cash is unlikely to have been sitting in the defendant’s pocket for months on end.

Now that the sentencing guidelines are advisory rather than mandatory, the judge might have given the defendant the same sentence irrespective of the source of the $765. The judge knew that the defendant had been a dealer for at least eight months, that he had customers other than Beagle, and that only the quantities sold to Beagle had been accounted for. Because “in arriving at its factual findings, the district court may rely on any evidence bearing sufficient indicia of reliability,” United States v.

James, 487 F.3d 518, 529 (7th Cir.2007), he could infer that the defendant’s sales to those customers during that period had amounted to at least 12.75 grams. And when the defendant was arrested, he had 17 grams with him. Although this was the quantity Beagle had ordered, it greatly exceeded the usual quantity he would sell to Beagle. But it shows that the defendant had access to considerable quantities of crack, and if anything it is unlikely that he would have sold a total of only 12.75 grams to multiple customers over a three-month period.

So the judge could have assumed that the 12.75 grams derived from earlier sales of crack, without basing the assumption on the $765; and even if he had excluded 12.75 grams in calculating the guidelines sentencing range, he might have decided to sentence the defendant as severely as he did. So if we were certain or nearly so that the judge would not have imposed a lower sentence even if he had drawn no inference from the $765, there would be no point in remanding the case for a new sentencing hearing. But we lack that confidence.

The judge must have thought the 12.75 grams material; why else say as he did that he was crediting the presentence investigation report’s calculation? And he derived the figure from the $765. Without that quantity of grams as relevant conduct, the defendant’s sentencing range would have dropped to 87 to 108 months, though this assumes that the judge in figuring the sentence used the estimate of drug quantity in the presentence investigation report (39.75 grams, which included the 12.75 grams in question).

Although the judge said that he “could easily and conservatively estimate that the amount of crack was between 35 and 50 grams,” the sentence that he imposed — a sentence at the very bottom of what he thought the guidelines sentencing range was — suggests that he thought the lower end of the 35-to-50-gram range a better estimate of the quantity of crack sold by the defendant. His decision to impose the minimum guidelines sentence suggests a lean toward lenity, making it difficult to predict the outcome of a new sentencing hearing. In Gall v.

United States, 552 U.S. 38, 128 S.Ct. 586, 597, 169 L.Ed.2d 445 (2007), the Supreme Court instructed that in reviewing a sentence the court of appeals “must first ensure that the district court committed no significant procedural error, such as failing to calculate (or improperly calculating) the Guidelines range.” We cannot be confident that the judge did not commit a significant error of that kind. We therefore order the sentence vacated and the case returned to the district court for a further sentencing hearing. The conviction, however, is affirmed. Affirmed in Part, Vacated in Part, and Remanded.