United States v. Samuel B. Cerro

POSNER, Circuit Judge. A jury convicted Samuel Cerro of five counts of conspiracy to distribute cocaine and three counts of filing false income tax returns. The judge gave him concurrent 15-year sentences on the first three conspiracy counts and consecutive 15-year sentences on the fourth and fifth conspiracy counts, for a total of 45 years. (Sentences of 3 years on each of the tax counts were made concurrent with the conspiracy sentences.) Cerro had also been charged with being a drug “kingpin,” see 21 U.S.C. § 848, a charge that allows the most severe punishment possible under current federal law — life in prison without possibility of parole, see 21 U.S.C. §§ 848(a), (c), but this charge was dropped before trial on the authority of a dictum in United States v. Jefferson, 714 F.2d 689, 702 n. 27 (7th Cir.1983), that a drug conspiracy is not one of the felonies on which conviction of being a kingpin can be based, see 21 U.S.C. § 848(b)(1). The dictum was rejected in *910 United States v. Young, 745 F.2d 733, 750-52 (2d Cir.1984). See also United States v. Schuster, 769 F.2d 337, 345 (6th Cir.1985), and cases cited there. Its validity was left open in our recent decision in United States v. Markowski, 772 F.2d 358, 361 n. 1 (7th Cir.1985), and need not be decided in this case either; the government did not appeal the dismissal of the kingpin count, as it could have done under 18 U.S.C. § 3731. The principal issue raised by Cerro’s appeal is whether the five conspiracies were indeed separate or were instead aspects of a single conspiracy; if the latter, the maximum sentence (apart from the tax counts) is 15 years and Cerro must be resentenced. Each of the alleged conspiracies was between Cerro and one other person, to whom Cerro consigned (“fronted”) up to one ounce of cocaine each week for resale to users of cocaine. Each dealer would remit all or part of his retail sales revenues to Cerro, either retaining some of the revenues as compensation for his services or retaining some of the cocaine in lieu of money compensation. The five dealers were the principal witnesses against Cerro, testifying under grants of immunity. In Count I the dealer was Amato, and the evidence showed that the conspiracy ran from April 1980 to December 1980. The dealer and dates for the other conspiracies were as follows. Count II: Corti, summer or fall of 1978 to November 1980; Count III: Gaulke, March 1977 to the spring of 1979; Count IV: Schnidt, August 31, 1977 to December 8, 1980; Count V: Phillips, fall of 1979 to spring of 1981. Why the district judge imposed concurrent sentences for the first three conspiracies and consecutive ones for the last two is unclear. The fifth conspiracy did end later than the others, but the fourth (August 1977 to December 1980) fell within the overall span of the first three (March 1977 to December 1980). There was no evidence that any of the dealers knew that Cerro was selling cocaine through other dealers, beyond what might be inferred from the nature and timing of the conspiracies and the facts that all the transactions took place in Madison, Wisconsin, that Amato and Gaulke were brothers-in-law, that Corti “fronted” cocaine to Amato as well as selling cocaine at retail for Cerro, and that Phillips knew that Corti sold cocaine, though there is no indication that he knew that Cerro was Corti’s supplier. More often it is the government that is arguing for a single overarching conspiracy and the defendant for multiple conspiracies, the most recent examples in this court being United States v. Andrus, 775 F.2d 825, 840-41 (7th Cir.1985), and United States v. Towers, 775 F.2d 184, 189-90 (7th Cir.1985). The reasons, in descending order of importance, are that the statements made by a conspirator are admissible against all coconspirators, that multiple conspiracies involving different parties can raise serious problems of misjoinder under Rule 8(b) of the Federal Rules of Criminal Procedure, see United States v. Velasquez, 772 F.2d 1348, 1352-53 (7th Cir.1985), and that if a defendant is acquitted on one count of conspiracy and his convictions on other counts are reversed because of an error at the trial, retrial on those counts will be barred on the ground of double jeopardy if they actually charge the same offense as the one of which the defendant was acquitted. Although it might seem that the government would want to charge multiple conspiracies in the hope that the defendants) would receive heavier sentences, this is rarely the case. The indictment of a major drug distributor will usually contain multiple substantive counts, for which consecutive sentences can be imposed, as well as conspiracy counts. And consecutive sentences for a substantive drug offense and a conspiracy to commit such an offense are also permissible, see, e.g., United States v. Wylie, 625 F.2d 1371, 1381-82 (9th Cir.1980); United States v. Cardi, 519 F.2d 309, 315 (7th Cir.1975); United States v. Espinosa, 771 F.2d 1382, 1402-03 n. 27 (10th Cir.1985), though infrequent, see Marcus, Con *911 spiracy: The Criminal Agreement in Theory and in Practice, 65 Geo.L.J. 925, 938 (1977). By a fluke, none of the disadvantages of multiple conspiracies to the government is present here (there is no misjoinder problem because there is only one defendant, so that Rule 8(a), which is more liberal than 8(b), see United States v. Velasquez, supra, 772 F.2d at 1352-53, applies). And since the government unaccountably failed to charge any substantive drug offenses and the “kingpin” count was dismissed, the only possibility for imposing a very long prison term on the defendant was to prove multiple conspiracies. Whether this should be important when the defendant is 59 years old and in exceedingly poor health is not for us to say. A conspiracy is an agreement; and to be a party to an agreement you must know something of its general scope and objective though not necessarily its details. See id. at 1351; Blumenthal v. United States, 332 U.S. 539, 557, 68 S.Ct. 248, 256, 92 L.Ed. 154 (1947); United States v. Williams, 737 F.2d 594, 614 (7th Cir.1984); United States v. Andolschek, 142 F.2d 503, 507 (2d Cir.1944) (L. Hand, J.); cf. Raffles v. Wichelhaus, 2 Hurl. & C. 906, 159 Eng.Rep. 375 (Ex.1864); Smith v. North Am. Co. for Life & Health Ins., 775 F.2d 777 (7th Cir.1985). You must, of course, at the very least know there is an agreement. So for any of Cerro’s dealers to be a party to an overarching conspiracy to distribute cocaine through more than one dealer, the dealer would have to know that Cerro was selling through other dealers. See, e.g., Kotteakos v. United States, 328 U.S. 750, 755, 66 S.Ct. 1239, 1243, 90 L.Ed. 1557 (1946); United States v. Elam, 678 F.2d 1234, 1245-47 (5th Cir.1982); United States v. Barnes, 604 F.2d 121, 155 (2d Cir.1979); United States v. Borelli, 336 F.2d 376, 383 n. 2 (2d Cir.1964) (Friendly, J.). The cases sometimes say must know {Borelli, Elam, many others), less frequently must have reason to know (e.g., Barnes). Taken literally the latter formulation would imply something very curious indeed, that a conviction can be upheld without proof beyond a reasonable doubt of an element of the crime (knowledge of the conspiracy). But it should not be taken literally. Although usually in the law to say that someone has “reason to know” something means that he would be negligent in not knowing it, in the present context it means only that knowledge can be inferred from circumstantial evidence. If the facts indicate that the defendant must have known something (the formulation in Borelli), then a jury may be able to find beyond a reasonable doubt that he did know it, especially since the requirement of knowledge is satisfied by proof that the defendant willfully shut his eyes for fear of what he might see if he opened them, United States v. Josefik, 753 F.2d 585, 589 (7th Cir.1985). This interpretation of “reason to know” is supported by the full text of the passage in the case from which Barnes got the phrase: “if each [defendant retailer] knew, or had reason to know, that other retailers were involved with the Hernandez organization in a broad project for the smuggling, distribution and retail sale of narcotics, and had reason to believe that their own benefits derived from the operation were probably dependent upon the success of the entire venture, the jury could find that each had, in effect, agreed to participate in the over-all scheme.” United States v. Baxter, 492 F.2d 150, 158 (9th Cir.1973) (emphasis added). But the jury would have to find participation beyond a reasonable doubt. A jury could have found that Cerro’s dealers knew there were other dealers if for example Cerro had told the dealers this, if other dealers had told them, if they had observed cocaine dealings between Cerro and other dealers, or if they had seen Cerro with a quantity of cocaine too large to be distributed through one dealer. See, e.g., Blumenthal v. United States, supra, 332 U.S. at 549-59, 68 S.Ct. at 252-57; Note, Federal Treatment of Multiple Conspiracies, 57 Colum.L.Rev. 387, 388-89 (1957). Considering that the dealers had a common *912 supplier, that most of the conspiracies overlapped in time, sometimes by years, that the conspiracies had the same purpose and one common participant, and that Cerro apparently made no effort to insulate any dealer from knowledge of the others, it is highly likely that the dealers did know they were part of a conspiracy involving other dealers. Nevertheless there is no direct evidence of this, nor evidence from which such knowledge could be inferred with the degree of certitude required for a criminal conviction. Even if the relationship among Amato, Corti, and Gaulke could be thought to warrant an inference that they knew of each other’s involvement in Cerro’s distribution of cocaine — a proposition doubtful on its face but supported by language in some of the cases cited in the Columbia Law Review Note, supra, at 388 n. 12— these were the dealers involved in the first three conspiracy counts, and Cerro received concurrent sentences on those counts. The only link forged between those dealers and Schnidt and Phillips, the dealers involved in the other two counts, was Phillips’ testimony that he knew that Corti was a cocaine dealer, and he did not testify that he knew that Corti was a dealer for Cerro. The failure to prove a single conspiracy is not surprising. It was no part of the government’s purpose to show an overarching conspiracy — quite the contrary — or to show that the prosecution witnesses were members of such a conspiracy. And while Cerro might have wanted to establish a single conspiracy in order to put a cap on his sentence if he was convicted, he would not have improved his chances for acquittal by exploring the extent of the witnesses’ knowledge of each other’s dealings on his behalf. Testimony of their knowledge would have strengthened the case against him; testimony of their lack of knowledge would not have weakened it. But lack of proof that the dealers knew about each other need not be decisive on the question whether Cerro could be punished for multiple conspiracies. It is never right to try to answer a legal question, such as (in this case) whether there was one conspiracy or five conspiracies, without considering the purpose of the question. The purpose here is to decide whether Cer-ro was given an illegal sentence. For other purposes it would be clear that there were multiple conspiracies. It would be clear for example if any of the dealers had been charged with participating in an overarching conspiracy and the issue was whether the dealer could be convicted; lack of proof of a dealer’s knowledge of a conspiracy broader than his own dealings with Cerro would doom any attempt to make such a charge stick. If on the other hand Cerro had been charged with being the hub of a single conspiracy embracing all five dealers, proof of the dealers’ knowledge, though no doubt easy to come by if the government had wanted to explore that knowledge, might conceivably not have been necessary. Suppose, as has become familiar from the literature on espionage, revolution, and terrorism, that the leading members of a clandestine organization seek to minimize the chances of detection by adopting a system of “cut-outs” under which subordinate members know nothing about their superiors, their coordinate members, conceivably even the objects of the organization. See, e.g., Copeland, Without Cloak or Dagger: The Truth About the New Espionage 111-12, 140 (1974); Seth, Anatomy of Spying 97 (1963). It might be possible to argue that the leaders’ ability to conduct their operations through completely insulated cells, thereby making those operations more rather than less effective, should not protect the leaders from charges of participation in one large conspiracy rather than a series of petty two-man conspiracies, with all the obstacles to effective prosecution in the run of cases that multiple-conspiracy charges create; that the “hub” of an illegal enterprise can be guilty of conspiracy even if the “spokes” do not have a common rim. Otherwise there is the paradox that “a central figure will be guilty of one conspiracy or of many, depending entirely upon whether he informed each individual with whom he plotted of the others’ existence.” *913 Developments in the Law, Criminal Conspiracy, 72 Harv.L.Rev. 920, 928 (1959). Whether the objection that a conspiracy of many requires a multiplicity of knowing conspirators could be surmounted is fortunately not an issue we need decide in this case. Maybe it will never have to be decided. In the type of case we have put each spoke is bound to know, in fact, that he is “participating in a collective venture” of some sort, United States v. Martino, 664 F.2d 860, 876 (2d Cir.1981), though the government didn’t bother to prove that here. It will be rare for none of the spokes to know even the general character of the venture. And one possible solution to the problem of cut-outs (though limited to drug cases), if there really is a problem, is the “kingpin” statute, cited earlier. True, it requires that the kingpin’s substantive violations have been committed “in concert with five or more other persons with respect to whom [the kingpin] occupies a position of organizer, a supervisory position, or any other position of management,” 21 U.S.C. § 848(b)(2)(A); and the Supreme Court hinted strongly in Jeffers v. United States, 432 U.S. 137, 147-50, 97 S.Ct. 2207, 2214-15, 53 L.Ed.2d 168 (1977), that the words “in concert” imply an agreement. But this requirement might conceivably be satisfied by proof of a separate agreement between the kingpin and each of his underlings. Cf. United States v. Bec-ton, 751 F.2d 250, 254-55 (8th Cir.1984); United States v. Dickey, 736 F.2d 571, 587 (10th Cir.1984); United States v. Phillips, 664 F.2d 971, 1012-13 (5th Cir.1981). In this case, where the only issue is length of sentence and no one suggests that the defendant was attempting to minimize his fellow conspirators’ knowledge of the larger conspiracy, the argument against finding multiple conspiracies is compelling. The government cannot, just by choosing to prove less than it could easily have proved by asking its own witnesses another question — namely, whether each of the dealers realized he was not alone — get a defendant punished more severely; cannot, simply by not proving a single conspiracy, obtain punishment more severe than if it had proved one; cannot shift, to a defendant fearful of more severe punishment, the burden of proving how much his coconspirators knew about the conspiracy. The balance of advantages in a conspiracy case is adequately in the prosecutor’s favor without his being allowed to gain a sentencing bonus by atomizing a single venture. Maybe for purposes of trial the government can charge as many conspiracies as it wants unless the effect is to confuse the jury. But we do not think the government can get the judge to pile on consecutive sentences by arbitrarily, as in this case, declining to present minimum evidence of what is pretty apparent without evidence, that there really was one conspiracy. The issue has arisen more commonly when the government, rather than charging multiple conspiracies in the same case (for, as we have said, the government’s interests are ordinarily best served by charging a single overarching conspiracy), reindicts a defendant acquitted (or convicted) on another conspiracy charge, and the defendant argues that the conspiracies are one and the same for purposes of double jeopardy. Courts in such cases do not ask just whether the evidence required to establish the defendant’s guilt is the same in both cases; that would encourage the prosecution to withhold some of its evidence in the first case. They also ask whether the “conspiracy has been subdivided arbitrarily, resulting in multiple indictments for a single illegal agreement.” United States v. Castro, 629 F.2d 456, 461 (7th Cir.1980); see also United States v. Sinito, 723 F.2d 1250, 1256 (6th Cir.1983); United States v. Papa, 533 F.2d 815, 820 (2d Cir.1976). The inference of arbitrary subdivision is particularly strong in this case and persuades us that the government should not be allowed to obtain a sentencing advantage from having failed at trial to explore the dealers’ knowledge of the scope of the conspiracy in which each was involved. Contrary to its usual position in conspiracy cases, the government argues, how *914 ever, that it could not have charged a single conspiracy, because the dealings between Cerro and each dealer were not “mutually dependent.” Many cases do say that mutual dependence must be shown for a course of dealings to be a single conspiracy. See, e.g., United States v. Percival, 756 F.2d 600, 607 (7th Cir.1985); United States v. Dickey, supra, 736 F.2d at 582; United States v. Abushi, 682 F.2d 1289, 1295 (9th Cir.1982); United States v. Ber-tolotti, 529 F.2d 149, 154 (2d Cir.1975). But as the government well knows, this language is not taken literally. If it were, single conspiracies could not be charged in most cases (which are extraordinarily common) where a wholesaler of illegal drugs deals through more than one dealer, no matter how much each dealer knew about the scope of the operation. As is true of wholesale distribution generally, the viability of wholesale drug trafficking does not depend on the adherence of any single dealer. Lop off one dealer, and the wholesaler can hire another in his place or shift the dealer’s allotment of drugs to one or more of his remaining dealers or simply reduce the scale of the drug operation. The dealers are not dependent on each other as a bank robber is dependent on the driver of the getaway car and vice versa. Just as Exxon would not go out of business if it lost one of its retail dealers, so a substantial drug operation would not go out of business because one of its retail dealers left the business. The expression “mutual dependence” can be taken literally only in cases (and not all of them, either) where the conspirators are different links in the chain of distribution. If the conspiracy is to distribute imported drugs and the sole importer defects, an essential link has been severed. But when the conspirators (all but one) sell in parallel, each is a strand rather than a link, and normally an inessential one; yet conspiracies of this sort are prosecuted all the time, and successfully, too. You can be part of a common enterprise without being indispensable to that enterprise; and a common enterprise for illegal ends is a conspiracy whose inessential as well as essential members are conspirators. “Mutual support” would be a more accurate description of the cases, which require at most that the various arrangements and transactions alleged to constitute or manifest a single conspiracy contribute to the success of the overall undertaking and in that sense reinforce each other. At any rate, by choosing not to inquire into the dealers’ knowledge of Cerro’s operation, the government is barred from obtaining a sentencing advantage, since as we have said the facts show that Cerro was the hub of a conspiracy the “spokes” of which probably knew of each other’s existence, and which was no more effective or dangerous a conspiracy if, contrary to all the probabilities, they in fact were in the dark and thought, each one, that he was Cerro’s only dealer. The only other question that requires discussion (Cerro’s other grounds of appeal being plainly without merit) is whether his conviction should be reversed because the district court improperly curtailed his efforts to impeach the credibility of the dealers, the principal witnesses against him. The pretrial order in this case, which apparently is used routinely by this district judge in criminal cases, requires written notice before the final pretrial conference if either counsel intends to try to present at trial evidence of, among other things, “prior similar acts or any person ([Fed.R.Evid.] 404(b)).” Cerro’s counsel did not object to the order and did not give the required notice, and at trial the judge forbade him to present some of the evidence regarding the prior crimes of the witnesses with which he hoped to impeach them. The two most important items of evidence that were excluded were that witness Gaulke’s insurance-brokerage license had been revoked for fraud and that witness Corti was an extortionist. Although the district judge must know better than we what his pretrial order means, we find it very hard to understand how it could be read to require that defense counsel give the prosecution writ *915 ten notice of the evidence with which he hopes to undermine the jury’s belief in the truth of the testimony of the prosecution witnesses (and thus Cerro’s counsel cannot be faulted for having failed to object to the order when it was entered). Rule 404(b), which provides that evidence of other crimes, wrongs, or acts illuminating a person’s character is not admissible in order to prove that he acted in conformity with that character (but allows it to be admitted for other purposes), applies to evidence in chief, not to impeachment evidence. This is apparent from a comparison with Rules 608 and 609, which deal specifically with using evidence of character and of prior crimes to impeach witnesses. See also the Advisory Committee’s Note to Rule 608(a). Whatever the pretrial order means, there is a serious question whether a district judge is empowered to require discovery of impeachment evidence. Pretrial discovery of federal criminal defendants is regulated in detail by Rules 12.1, 12.2, and 16 of the Federal Rules of Criminal Procedure, and nothing in any of them suggests that impeachment evidence is discoverable. Rule 16(b)(1)(A) limits compelled disclosure of defense documents to those “which the defendant intends to introduce as evidence in chief at the trial.” Rule 16(d), on which the government relies here, regulates the manner rather than objects of discovery from a criminal defendant and is thus irrelevant to our issue. Whether the provision for pretrial conferences made by Rule 17.1 is intended to override the limitation in Rule 16(b)(1)(A) is most doubtful. Nothing in the language or history of the rule suggests that it is, an omission not surprising when we consider that a defendant’s interest in being able to conduct a vigorous and effective cross-examination — an interest central to the right of a criminal defendant under the Sixth Amendment “to be confronted with the witnesses against him,” see, e.g., Davis v. Alaska, 415 U.S. 308, 315-16, 94 S.Ct. 1105, 1109-10, 39 L.Ed.2d 347 (1974)—would be impaired if he had to give a précis of his cross-examination to the prosecution before trial. The effect of the pretrial order, as interpreted by the district judge, is to make the defendant do just that, whenever he plans to use evidence to buttress his cross-examination — as often he will. For not only does effective cross-examination require that counsel be prepared to prove the falsity of replies to questions intended to undermine the witness’s credibility, but independent evidence is usually a more fertile source of impeachment in cross-examination than replies to questions. See McCormick on Evidence 66 and n. 3 (Cleary 3d ed. 1984). But any error here was harmless beyond a reasonable doubt. Keeping the revocation of Gaulke’s broker’s license from the jury could do little harm to the defendant. Given Gaulke’s own testimony that he had stolen funds from his employer to pay for cocaine that he bought from Cerro, evidence that his insurance broker’s license had been revoked would have been anticlimactic and its exclusion was surely harmless. And similarly with Corti. Evidence that he was an-extortionist was excluded, it is true, but Corti admitted having committed perjury (though he said it had been done under duress), and perjury is more relevant to credibility than extortion would be. And remember that the whole premise of Gaulke’s and Corti’s testimony, as of the testimony of the other three dealers, was that they had sold cocaine illegally. Thus it is not as if the jury could have thought that Gaulke and Corti were anything but what they were: crooks singing for their freedom. Finally, the evidence of Cerro’s guilt would have been overwhelming even if Gaulke and Corti had been thoroughly discredited by a mountain of evidence of their prior crimes — especially once the three conspiracies are recharacterized as one, as we have held they must be. It is always perilous to speculate on what the effect of evidence improperly admitted was on a jury, or what the effect of evidence improperly excluded would have been. See Teitelbaum, Sutton-Barbere & Johnson, Evaluating the Prejudicial Effect of Evidence: Can Judges Identify the Impact of Improper Evidence *916 on Juries?, 1983 Wis.L.Rev. 1147. The lay mind evaluates evidence differently from the legal mind, and while many appellate judges have substantial experience with juries and perhaps great insight into the thinking process of juries, others do not. This is a reason to be wary about invoking the doctrine of harmless error (on which see, e.g., United States v. Hasting, 461 U.S. 499, 507-12, 103 S.Ct. 1974, 1979-82, 76 L.Ed.2d 96 (1983); 3 LaFave & Israel, Criminal Procedure § 26.6, at pp. 278-81 (1984)) with regard to evidentiary rulings in jury cases. Most cases of harmless error involve the erroneous admission of evidence favorable to the prosecution. As a practical matter, if the evidence of guilt that was neither erroneously admitted nor tainted by the erroneously admitted evidence is overpowering, and the tainted evidence neither probative nor inflammatory (inflammatory evidence might overbear the judgment of a lay trier of fact), so that the appellate court can safely conclude that the likelihood that the defendant would have been acquitted is very small — too small to warrant the delays and other costs entailed by ordering a new trial — the error will be pronounced harmless (harmless beyond a reasonable doubt, if the appellate court is strongly convinced of its conclusion). Despite some contrary language in Davis v. Alaska, supra, 415 U.S. at 318, 94 S.Ct. at 1111, the doctrine of harmless error is available not only when the error consists of erroneously admitting evidence but also when it consists of excluding defense evidence in violation of the defendant’s constitutional right to confront the witnesses against him. See, e.g., United States ex rel. Nance v. Fairman, 707 F.2d 936, 941-43 (7th Cir.1983); United States ex rel. Scarpelli v. George, 687 F.2d 1012, 1016 (7th Cir.1982), and the very full discussion of this issue by Judge Rubin in Carrillo v. Perkins, 723 F.2d 1165, 1170-72 (5th Cir.1984). The test cannot be identical for the two types of error, however, and this for two reasons. First, the fact that improperly excluded evidence was inflammatory, in the sense of likely to distract and confuse the jury, would hardly be an argument against finding the error harmless; what would be, however, would be that the evidence was of the sort that might properly have carried great weight with the jury, even though the evidence put in by the prosecution was very strong. Second, there is an asymmetry between the prosecutor’s being allowed to put in more evidence of guilt than he should have and the defendant’s being prevented from putting on a defense. If the defendant were utterly precluded from defending himself, it would be clear that his conviction had to be reversed even if the evidence of guilt was overwhelming and could not have been offset by the evidence that the defendant would have introduced if allowed to do so. The Constitution assures the criminal defendant who wants it of some minimum amount of adversary procedure even if the outcome of the contest is a foregone conclusion. Otherwise the right to trial by jury could be denied on the ground that the outcome of the trial was foreordained and therefore the trial would be a waste of money. See Walberg v. Israel, 766 F.2d 1071, 1074 (7th Cir.1985). But we are far from that extreme point in this case. Cerro’s defense was impeded, probably improperly, but the impediment was a minor one; the error in keeping from the jury impeachment evidence of a marginal and cumulative nature must be deemed harmless beyond a reasonable doubt in a ease such as this where the evidence of guilt was overwhelming and the defendant was allowed to put on a defense, if not quite so complete a defense as he might reasonably have desired. Cases in which error in the curtailment of a criminal defendant’s right of effective cross-examination was found not harmless, such as Hoover v. Maryland, 714 F.2d 301, 306-07 (4th Cir.1983), and United States v. Bates, 617 F.2d 585, 588 (10th Cir.1980), are readily distinguishable. So Cerro’s convictions must stand, along with the sentences for the tax offenses. But the case must be returned for resen- *917 tencing on the conspiracy counts in accord-anee with our conclusion that the defendant cannot be given consecutive sentences on those counts. Affirmed in Part, Vacated in Part, and Remanded with Directions.