The Government appeals the district court's decision granting, in part, Defendant -Appellee Timothy Willis, Jr.'s ("Willis" or "defendant") motion for a new trial. We AFFIRM the district court's decision.

In October 1997, following complaints about drug sales and gang-related activity, detectives in the Narcotics Unit of the Cleveland Police Department began conducting surveillance at 2267 East 83rd Street, a two-family residence in Cleveland, Ohio. Throughout that month, Detectives Mazur and Kooser conducted approximately fifty stationary and moving surveillances of the residence. The detectives noticed "a constant flow of traffic coming in and out of this house[,]" in which people would arrive at the house and then leave a short time later. Joint Appendix ("J.A.") at 143 (Mazur Test.). Throughout their surveillance of the residence, the detectives consistently noticed that the defendant, Timothy Willis, Jr., was present at the house. Detective Mazur testified that he never saw anyone other than the defendant answer the door at 2267 E. 83rd St. when someone knocked, nor did he ever see anyone else enter the residence without knocking.

On November 7, 1997, Detective Mazur arranged to have a confidential informant purchase heroin at 2267 E. 83rd St. Detective Mazur did not witness the transaction, which took place inside the residence. The informant who made the buy died before trial. In a hearing prior to trial, the district court held that any evidence of the informant's drug purchase would be inadmissible under Fed. R. Evid. 404(b).1

Following the successful controlled drug purchase, the Cleveland Police Department obtained a warrant to search the residence at 2267 E. 83rd St. Upon searching the home, the officers discovered in the bathroom an electronic gram scale that can be used to measure cocaine and heroin. Also, in a small, padlocked closet in the bathroom, officers discovered a jacket containing seven bricks (10.5 grams) of heroin; a Nike shoe box containing just over a thousand dollars in cash; a pair of jeans containing 93.52 grams of crack-cocaine and $6,000 in cash; and a Taurus gun loaded with ten rounds of ammunition. The search of the closet also revealed several receipts with the name "Tim Willis" on them, with some of them also containing the "2267 E. 83rd St." address.

In addition to the items found in the bathroom, the officers retrieved many papers addressed to or intended for "Timothy Willis" in the residence's one bedroom.2 These papers included bills, receipts, law firm correspondence, and personal letters all addressed to Timothy Willis or Timothy Willis, Jr.

Based on the evidence seized from the residence, an arrest warrant was issued for the defendant and he was thereafter indicted on two counts of possessing with the intent to distribute heroin and crack-cocaine in violation of 21 U.S.C. §841(a)(1), and one count of being a felon in possession of a firearm in violation of 18 U.S.C. §922(g)(1). Upon his arrest, officers obtained defendant's driver's license and car registration, which both listed 2267 E. 83rd St. as his home address.

At trial, the government relied in large part upon the evidence seized at 2267 E. 83rd St. coupled with the officers' testimony that the defendant was often seen in and around the house. The government also presented the testimony of Reginald Bryant, defendant's mother's boyfriend, who stated that he had given the Taurus firearm found by the police in the padlocked closet to the defendant in or around May 1996. The government also brought Virginia Profitt, defendant's probation officer, to the stand. Profitt testified that, during the course of defendant's parole and his monthly visits to her office, he consistently listed 2267 E. 83rd St. as his address. Furthermore, Profitt testified that Willis had a scheduled parole visit on November 19, 1997, the day the Cleveland Police executed the search of 2267 E. 83rd St. For that visit, Willis again listed 2267 E. 83rd St. as his address, and further noted on a form that he had not changed his address within the past month.

The defense presented several witnesses of its own, and also engaged in some effective cross-examination of the government's witnesses. On cross-examination, Detective Mazur admitted that Willis was not the only person whose papers were found at the residence. While Mazur did not present any of these papers in court, he recalled seeing documents with the names "William Warren" and "Tyano Montgomery" on them, also addressed to 2267 E. 83rd St.

As for its own witnesses, the defense called Tyano Montgomery, Timothy Willis, Sr.'s former fiance. Montgomery lived in the separate upstairs portion of the house at 2267 E. 83rd St. from some point in 1996 through January 1998. Montgomery testified that a lot of people lived in the downstairs quarters of the residence from January 1997 through January 1998, the year during which the police investigation was ongoing. Montgomery also testified that Timothy Willis, Jr. stopped living in the downstairs residence in August 1996 after he began receiving threats in school. Montgomery stated that the defendant had moved to the west side of Cleveland, though she was not sure where.

Montgomery testified that the defendant visited the residence occasionally in 1997, sometimes to pick up his mail. Montgomery stated that she retrieved the mail each day, and would leave mail specifically addressed to "Timothy Willis, Jr." on the table in the downstairs residence for him to pick up. Montgomery also explained that when mail was addressed generally to "Timothy Willis," she would open it to see if it was intended for Willis, Sr. If the mail was intended for Willis, Jr., she would again leave it for the defendant in the downstairs residence. Montgomery further testified that Charles, Richard, and William Warren, cousins of the defendant, all had keys to the downstairs residence at 2267 E. 83rd St.,3 and that both Charles and William Warren were living in the residence after May 1997.

The defense also called Betty Fitzgerald, the defendant's grandmother. Fitzgerald's testimony was consistent with Montgomery's in that she also testified that the defendant moved to the west side of Cleveland, although she did not know where, in August 1996 because "[h]e was having problems." J.A. at 493 (Fitzgerald Test.). Fitzgerald acknowledged that she owned the residence at 2267 E. 83rd St. and stated that after Willis left 2267 E. 83rd St., William and Charles Warren and others still lived in the downstairs residence. Fitzgerald testified that the locks were changed in the summer of 1997 after the house was burglarized, and that she did not believe the defendant had a key after this time. Fitzgerald stated, however, that she had seen Charles and William Warren using keys to enter the downstairs residence.

Following the presentation of the defense's witnesses, the government then informed the district judge that it wished to call Richard Warren, cousin of the defendant, for purposes of rebutting Montgomery's and Fitzgerald's testimony. The district court interviewed Richard Warren to determine if he had observed any of the trial proceedings or had contact with someone who had. Richard Warren informed the judge that his brother, William, had called him earlier that same day and told him that Richard would have to testify in court because "somebody was trying to put some dope and money against [Richard], talking about [Richard] ... selling drugs or something out of the house on 83rd, or something like that." J.A. at 530 (Richard Warren Voir Dire). William Warren also told Richard that if he did not come to the courthouse to testify, the government was "going to have a warrant out for [Richard's] arrest." J.A. at 531 (Richard Warren Voir Dire).

At sidebar, after interviewing Richard Warren, the district court noted its concern with having Richard Warren testify under these circumstances:

Let me tell you what my concern is. I don't have any problem with the prosecutors and police officers going out to get a person who is a potential witness and talk with them and to bring them in generally to prepare them, not about testimony, but about getting the facts and information, which they can then testify here to in court.

And I'm just having a problem understanding why the mechanism is used if a brother who came over here and is obviously scared out of his wits, but who heard some things in court, or told some things, to go out and talk to the other potential witness and tell him what those things are rather than to have the agents of the government themselves seek out the person and to make sure they are properly prepared without that kind of intervention."

J.A. at 536-37 (Richard Warren Voir Dire). Despite these concerns, the district court still permitted Richard Warren to testify, calling his testimony "appropriate rebuttal evidence." J.A. at 546 (Richard Warren Voir Dire).

In the course of direct examination, Richard Warren testified that the defendant had lived at 2267 E. 83rd St. for as long as long as Richard had known him, which, as Richard stated, was "[a]ll [his] life." J.A. at 551-53 (Richard Warren Voir Dire). Richard Warren stated that he never knew the defendant to have lived on the west side of Cleveland. Richard Warren further testified that only his brother, William, had ever lived at 2267 E. 83rd St., and that William had only lived in the upstairs residence at that address. Finally, Richard Warren identified the jacket in which heroin was found during the search of the residence as belonging to the defendant.

On cross-examination, Willis's attorney pointed out the suspicious circumstances under which Richard Warren was testifying, including the fact that a police car had picked him up and brought him to the courthouse, and that Richard's brother had told him that Richard would be arrested if he did not testify. Defense counsel was also able to impeach the witness with photographic evidence after the witness claimed that he had never congregated with others on the porch at 2267 E. 83rd St.

On May 7, 1998, after the testimony concluded, the jury found the defendant guilty of all three counts in the indictment. On August 24, 1998, following several motions to continue the sentencing date, Willis filed a motion for a new trial in light of an affidavit submitted by Richard Warren in which Richard recanted significant portions of his trial testimony. In his affidavit, Richard Warren stated that he testified because he was "extremely scared that" if he did not testify, he would be charged with some crime. J.A. at 28 (Richard Warren Aff.). Richard Warren stated that, during his ride to the courthouse in the police car, one of the officers told him he was not under arrest "'yet.'" J.A. at 28 (Richard Warren Aff.). According to Richard Warren's affidavit, the prosecutor told him that the government "wanted to get Timothy Willis and if [it] couldn't get Timothy Willis, [it] had to get somebody to pay for this crime." J.A. at 28 (Richard Warren Aff.).

Richard Warren further stated in his affidavit that he felt pressured by the prosecutor to testify that Willis lived on East 83rd St. and that the jacket in question belonged to Willis, but that those statements were "complete lie[s]." J.A. at 28-29 (Richard Warren Aff.). Richard Warren also stated that he lied in testifying that his brothers did not have access to the house at 2267 E. 83rd St. Richard Warren finally stated that he had come forward on his own free will because of "a guilty conscience[,]" and that no one had promised him anything or made any threats so that he would come forward. J.A. at 29 (Richard Warren Aff.).

The government filed a response in opposition to defendant's motion for a new trial on October 9, 1998. Just short of one year later, the government, because the district court still had not ruled on defendant's motion for a new trial, filed a motion to advance the case to sentencing. On October 28, 1999, the district court issued an order granting in part and denying in part the defendant's motion for a new trial based on the Richard Warren's recantation of his trial testimony. The court granted Willis's new trial motion with respect to his conviction for possessing with the intent to distribute crack-cocaine and heroin, but denied Willis's motion with respect to his conviction for being a felon in possession of a firearm.4 In granting Willis's motion for a new trial, the district court applied the test first adopted in this circuit in Gordon v. United States, 178 F.2d 896 (6th Cir. 1949), cert. denied, 339 U.S. 935 (1950). Applying the test adopted in Gordon, the district court held that a new trial should be granted based on Richard Warren's recantation of his trial testimony because 1) it was satisfied that Richard Warren's trial testimony was false, and 2) without the recanted testimony, it was "highly likely" that the jury would have reached a different verdict on Willis's drug possession charges. J.A. at 56 (Dist. Ct. Order Granting Def.'s New Trial Mot.). The government then appealed the district court's decision to this court.

A district court's decision to grant a defendant's motion for a new trial will not be overturned absent a clear abuse of the district court's discretion. United States v. Lewis, 338 F.2d 137, 139 (6th Cir. 1964). "A district court clearly abuses its discretion when it applies the wrong legal standard, misapplies the correct legal standard, or relies on clearly erroneous findings of fact." United States v. Turns, 198 F.3d 584, 586 (6th Cir. 2000) (quotation omitted). Whether the proper legal standard was applied, however, is a legal question that this court reviews de novo. In re Sorah, 163 F.3d 397, 400 (6th Cir. 1998) (citing In re Calhoun, 715 F.2d 1103, 1111 (6th Cir. 1983)).

B. Did the District Court Apply the Proper Legal Standard?

The government and Willis adamantly disagree over whether the proper legal standard for deciding motions for a new trial was used in this case. The government contends that the traditional, four-part test evaluating motions for a new trial based on newly discovered evidence should be applied in this case. This test was enunciated in United States v. Barlow, 693 F.2d 954, 966 (6th Cir. 1982), cert. denied, 461 U.S. 945 (1983), and has consistently been used in this circuit to scrutinize new trial motions based on newly discovered evidence. See, e.g., Turns, 198 F.3d at 586-87; United States v. Gaitan-Acevedo, 148 F.3d 577, 589 (6th Cir.), cert. denied, 525 U.S. 912 (1998); United States v. Frost, 125 F.3d 346, 382 (6th Cir. 1997), cert. denied, 525 U.S. 810 (1998). Pursuant to the Barlow test, a new trial motion will be granted based on newly discovered evidence only if the defendant can prove that the evidence was:

(1) discovered only after trial;

(2) could not have been discovered earlier with due diligence;

(3) is material and not merely cumulative or impeaching; and

(4) would likely produce an acquittal if the case were retried.

Barlow, 693 F.2d at 966.

Although the district court in its opinion recognized that the Barlow test is used to evaluate new trial motions based on newly discovered evidence, it held that a different standard is used when a new trial motion is based on a material government witness's recantation of his trial testimony. The district court then cited to the test first stated in this circuit's published opinion in Gordon as the appropriate standard in such cases. In Gordon, the defendant, following his conviction, filed a motion for a new trial based on the principal government witness's recantation of his trial testimony. In addressing the new trial motion, this court followed the U.S. Court of Appeals for the Seventh Circuit's decision in Larrison v. United States, 24 F.2d 82 (7th Cir. 1928), stating that a motion for a new trial based on the recantation of a material government witness should be granted only if:

(1)the court is reasonably well satisfied that the trial testimony given by the material witness is false; (2)without the false testimony, the jury might have reached a different conclusion; and

(3)the party seeking the new trial was taken by surprise when the false testimony was given, and was unable to meet it or did not know of its falsity until after the trial.

Gordon, 178 F.2d at 900.

In applying the test, the Gordon court first stated that, in cases such as these where all of the testimony given by a witness against the defendant is recanted, the requirement that the defendant be surprised when the false testimony is given and be unable to meet it or know of its falsity until after the trial is not "pertinent." Id. The district court in Gordon had denied the defendant's new trial motion based on the first factor of the Larrison test: that the court be reasonably well satisfied that the testimony given by the witness at trial was false. This circuit, noting the witness's history of crime and perjury, held that the district court's decision was not "unreasonable[,]" and affirmed the district court's denial of the defendant's new trial motion. Id. at 899-900.

After examining the history of this circuit's use of both the Barlow and Gordon tests, we agree with the district court in this case that the Gordon test, and not the Barlow standard, is the appropriate test to apply in these unique circumstances where a material witness testifying on behalf of the government later recants his trial testimony. No published opinion in this circuit has explicitly called the use of the Gordon test into question. Nor have any of our published opinions applied the Barlow test to facts like those at issue in Gordon and the current case. As the rules of this circuit require, a panel of this court is bound by the prior published opinions of this circuit "unless an inconsistent decision of the United States Supreme Court requires modification of the decision or this Court sitting en banc overrules the prior decision." Salmi v. Sec'y of Health & Human Servs., 774 F.2d 685, 689 (6th Cir. 1985); 6th Cir. R. 206(c).

On several occasions, this circuit has acknowledged that the Gordon test applies to these unique situations in which a material witness for the government recants his testimony after the trial. For example, in United States v. Lewis, 338 F.2d 137, 138-40 (6th Cir. 1964), cert. denied, 380 U.S. 978 (1965), we applied the Gordon test to a new trial motion based on the recantation of a prosecution witness. Furthermore, in more recent cases, we have recognized the Gordon test's continuing applicability in criminal cases, and we have applied the test to new trial motions in the civil context. See, e.g., Davis v. Jellico Cmty. Hosp., Inc., 912 F.2d 129, 134 (6th Cir. 1990); Abrahamsen v. Trans-State Express, Inc., 92 F.3d 425, 428 (6th Cir. 1996) (citing Davis and Gordon).

The only published opinion in this circuit that has not explicitly recognized the distinction between the Barlow andGordon tests is United States v. Chambers, 944 F.2d 1253, 1263-64 (6th Cir. 1991), cert. denied, 502 U.S. 1112 (1992), another case involving a new trial motion based on a recanting government witness. In Chambers, this court cited both to the Barlow test for new trial motions based on newly discovered evidence and to the Gordon test for new trial motions based on the recantation of a material government witness. Although the Chambers court cited to both Barlow and Gordon, the decision to deny the defendant's new trial motion was based on an application of the Gordon test.

The district court in Chambers, after examining the recanting witness's demeanor both during trial and at the evidentiary hearing, as well as the witness's motivation to lie after trial, found the witness's trial testimony to be credible and his subsequent recantation to be "'wholly unbelievable.'" Chambers, 944 F.2d at 1264. Based on these findings, the district court denied the defendant's motion for a new trial. This circuit, in affirming the district court's decision, recognized as "the primary ground for granting the new trial" motion the requirement emanating from Gordon that the trial judge be reasonably well satisfied that the recanting witness's trial testimony was false. Id. (quoting United States v. Kearney, 682 F.2d 214, 220 (D.C. Cir. 1982)). The Chambers court then held that, because the district court had observed the witness both at trial and at the evidentiary hearing, the district "court was uniquely qualified to pass on [the recanting witness's] credibility." Id. This court could find no abuse of discretion in the district court's decision denying the defendant's motion for a new trial.

Although the Chambers court cited to both the Barlow and Gordon tests in addressing the defendant's motion for a new trial, we applied only the Gordon test, and our decision was dictated by the defendant's failure to satisfy the Gordon test's first element. We do not believe that Chambers provides any basis for questioning Gordon's continuing applicability in this circuit.5 Even if it did, however, to the extent that Chambers calls into question a prior published opinion of this circuit, we remain bound by the holding of the earlier case. Sowards v. Loudon County, Tenn., 203 F.3d 426, 431 n.1 (6th Cir.),cert. denied, 121 S. Ct. 179 (2000); Brentwood Acad. v. Tenn. Secondary Sch. Athletic Ass'n, 180 F.3d 758, 765 (6th Cir. 1999), rev'd on other grounds, 531 U.S. 288 (2001).

Although we do not believe the Barlow test to be applicable to this case, it has been used by this circuit in cases similar to, though certainly not the same as, the current case. For example, in United States v. Pierce, 62 F.3d 818, 823-35 (6th Cir. 1995), cert. denied, 516 U.S. 1136 (1996), we applied Barlow to a new trial motion in which a character witness for the defendant came forward after the trial and confessed to committing the crime for which the defendant was convicted. In Turns, a similar case to Pierce, this court applied Barlow to a new trial motion in which the defendant's sister, who refused to testify at trial, later brought forth evidence that tended to exonerate the defendant. Turns, 198 F.3d at 586-87.

While this circuit has applied the Barlow test in cases similar to this one, it is clear that different concerns are at issue in cases like Pierce and Turns. In both Pierce and Turns, the defendants attempted to come forward after the trial with entirely new evidence never before heard by the jury in an attempt to exonerate themselves. In these cases, arguably the stricterBarlow standard should be applied in light of concerns that defendants might sandbag the prosecution, waiting to see if they are convicted before bringing forth new evidence in an attempt to get a second chance at acquittal. As this court stated in Turns:If the district court's decision [granting a new trial] was allowed to stand, then other defendants would be encouraged to file motions for new trials based solely upon the existence of previously uncalled witnesses who, after learning of the defendant's conviction, state for the first time that they are willing to testify truthfully on the defendant's behalf. Such a precedent would also encourage defendants to hold a witness or two in reserve, knowing that if they lost at trial, they might get another chance by producing sworn affidavits from their reserve witnesses.

Turns, 198 F.3d at 588.

In cases like Gordon and the current case, these sandbagging concerns are not at issue. Instead, the fear in a case like this one is that the defendant, his friends, or his family have reached the witness after trial and convinced him to recant his damaging testimony. While this surely is a legitimate concern, there are two "checks" on witness tampering in these cases that do not factor into the Barlow scenario.

The first check is the "'primary'" element of the Gordon test: that the district court be reasonably well satisfied that the original trial testimony given by the now-recanting government witness was false. Chambers, 944 F.2d at 1264 (quotingKearney, 682 F.2d at 220). Because the district court must be satisfied that this original trial testimony was false, this substantially allays concerns that posttrial witness tampering will serve as the basis for a successful new trial motion. Indeed, in Gordon, Lewis, and Chambers, the defendants' new trial motions were all denied because the primary element of the Gordon test had not been satisfied. Gordon, 178 F.2d at 900; Lewis, 338 F.2d at 139-40; Chambers, 944 F.2d at 1263-64.

The second check on coerced recantations is the threat of perjury charges being brought against the recanting witness. Whereas in Barlow cases a previously uncalled defense witness will not incur a substantial risk of perjury charges by coming forward with new testimonial evidence, in cases like Gordon and the one at bar, the recanting witness, by admitting under oath that the testimony he gave at trial was not true, clearly opens himself up to perjury charges being brought by the government.

With these two checks in place, arguably, new trial motions based on the recantation of a material government witness, unlike new trial motions based on newly discovered evidence, do not need the further "check" of a probability standard. Regardless of the policy justifications for having two separate tests, it is clear that Gordon is still good law in this circuit, and that we remain bound by it. The district court did not err in applying the Gordon test to Willis's motion for a new trial.

Before turning to the district court's application of the Gordon test, it is important to note that this court reviews the district court's decision whether to grant a defendant's motion for a new trial for an abuse of discretion. Lewis, 338 F.2d at 139. Factual findings of the district court that are clearly erroneous constitute an abuse of discretion. Turns, 198 F.3d at 586. While we apply a relatively lenient standard of review to a district court's decision regarding a new trial motion, weighing against this is the fact that new trial motions are disfavored and should be granted with caution. Id. Furthermore, as the district court recognized in its opinion, affidavits by witnesses recanting their trial testimony are to be looked upon "with extreme suspicion." J.A. at 52 (Dist. Ct. Order) (citing Chambers, 944 F.2d at 1264). The district court also acknowledged that the skepticism with which a court examines such an affidavit only heightens when "the recanting witness is a family member and the witness has feelings of guilt or the family members seek to influence the witness to change his story." J.A. at 52 (Dist. Ct. Order). Despite these inherent concerns with the credibility of Richard Warren's affidavit in this case, the district court granted, in part, Willis's motion for a new trial. We now examine whether the district court abused its discretion in doing so.

1.The District Court Must Be Reasonably Well Satisfied That the Trial Testimony Given by Richard Warren Was False

The first requirement of the Gordon test is that the district court must be reasonably well satisfied that the original trial testimony given by the government witness is false. It is essential to note that the district court, by presiding over the trial at which the recanting government witness first testified, is "uniquely qualified" to address the defendant's motion for a new trial based on the witness's posttrial recantation. Chambers, 944 F.2d at 1264. In this case, after having presided over the original trial and assessing the recanting witness's credibility, the district court held that it was satisfied that Richard Warren was not telling the truth at trial.

In reaching this determination, the district court noted that, during the trial, "the court itself had concerns about the veracity of Warren's trial testimony." J.A. at 53 (Dist. Ct. Order). At trial, the district court openly voiced its concerns about Richard Warren having been "picked up by a police car with [Richard's] brother telling him to get [to court] or [Richard] may be arrested[.]" J.A. at 547 (Richard Warren Voir Dire). The district court also noted in its order granting Willis's motion for a new trial that, "during the [s]idebar, it was clear that Warren was frightened and scared that he would either be accused of the crimes against Defendant or other crimes if he did not testify against Defendant." J.A. at 53.

Not only did the district court believe that Richard Warren's testimony at trial was of questionable veracity, but it also found the timing of his testimony to be suspicious. The district court stated that, after the government had presented its case, there still was no evidence linking Willis to the clothes in the padlocked closet in which the drugs were found. Furthermore, it was not until after the defense had presented its case, and it appeared possible that the drugs in the residence at 2267 E. 83rd St. may have belonged to one or several of the Warren brothers, that Richard Warren came forward to testify. Both Montgomery and Fitzgerald testified for the defense that several of the Warren brothers had been seen entering the residence, that the brothers had keys to the residence after May 1997, and that William and Charles Warren even lived in the residence for some period of time after May 1997. Only after the defense presented its evidence did Richard Warren agree to testify, and the district court's examination of Richard Warren showed that he took the stand having been told by his brother, William, that various people in court were trying to set him up on drug charges. Richard Warren was further informed by his brother, William, that if Richard did not protect himself by testifying in court, the government would have a warrant out for Richard's arrest. As the district court noted, all of these facts gave Richard Warren an "incentive to lie." J.A. at 53-54 (Dist. Ct. Order). In concluding its analysis of the Gordon test's first element, the district court explained that, unlike Richard Warren's self-exculpatory trial testimony, the trustworthiness of Richard's posttrial affidavit was bolstered by the fact that his recantation opened the way for perjury and possibly drug charges to be brought against him.

In light of the suspicious circumstances surrounding Richard Warren's testimony, we cannot hold that the district court abused its discretion in determining that it was satisfied that Richard Warren's trial testimony was false. Although the government is correct in noting that this court should be highly suspicious of exculpatory affidavits submitted by one of the defendant's family members, this suspicion does not outweigh the deference we must give to the factfinder's unique qualification to judge the credibility of the witnesses and the evidence presented at trial. Chambers, 944 F.2d at 1264.

2.Without the Evidence, the Jury Might Have Reached A Different Conclusion

The second part of the Gordon test requires the court to find that without the false testimony "the jury might have reached a different conclusion." Gordon, 178 F.2d at 900. Other circuits have criticized this element of theGordon/Larrison test as being too easy for the defendant to satisfy. United States v. Williams, 233 F.3d 592, 593-95 (D.C. Cir. 2000); Huddleston, 194 F.3d at 220; Sanders v. Sullivan, 863 F.2d 218, 225-26 (2d Cir. 1988) (noting disenchantment with the more liberal Larrison standard and applying it only in cases in which the government deliberately used perjured testimony); United States v. Krasny, 607 F.2d 840, 843-45 (9th Cir. 1979), cert. denied, 445 U.S. 942 (1980). Indeed, it is in this respect that the Barlow and Gordon tests are arguably the most different. Whereas the Barlow test holds that newly-discovered evidence can merit a new trial only if that evidence "would likely produce an acquittal if the case were retried," Barlow, 693 F.2d at 966 (emphasis added), the Gordon test permits a new trial motion to be granted if, without the recanting witness's testimony, "the jury might have reached a different conclusion[.]" Gordon, 178 F.2d at 900 (emphasis added).

While the Gordon test requires only the possibility of a different outcome without the now-recanted testimony, the district court in this case actually applied a probability standard when examining the chances of an acquittal had the jury not heard Richard Warren's testimony. See J.A. at 55 (Dist. Ct. Order) (stating the test as follows: "Without The Recanted Testimony, Is It Likely A Jury Would Reach A Verdict of Acquittal?"); J.A. at 56 (holding that it is "highly likely that the jury would have reached a different outcome without Warren's testimony"). Although we are bound by the possibility standard as stated in Gordon, it is significant that even under a more exacting standard the district court held that the second element of the Gordon test had, indeed, been satisfied.

In holding that it is "highly likely" that the jury would have reached a different outcome if not for Richard Warren's testimony, the district court stated that, while the government had offered evidence connecting Willis with the house at 2267 E. 83rd St., it had presented no evidence other than Richard Warren's testimony connecting Willis with the clothes recovered at the home in which the drugs and money were found. J.A. at 56 (Dist. Ct. Order). The court noted that the government had presented no evidence showing that any of the keys in Willis's possession at the time of his arrest could be used to unlock the 2267 E. 83rd St. residence or the padlocked closet in the home's bathroom.

The court also explained that, while personal papers apparently belonging to the defendant were found throughout the house, including in the padlocked closet, "such papers do not necessarily link him to the drugs and money found in the locked closet[,]" especially considering that the personal papers of others were also found throughout the home. J.A. at 55 (Dist. Ct. Order).

Finally, the court stressed the testimony of Montgomery and Fitzgerald, who both testified that Willis stopped living in the house as of August 1996. "Because a number of individuals had personal items at 2267 East 83rd Street and could have lived in the house," the district court held that it was "highly likely that the jury would have reached a different outcome without Warren's testimony, as there would have been a reasonable doubt as to whether the clothes (with the drugs and the money) belonged to the Defendant." J.A. at 56 (Dist. Ct. Order).

After a thorough review of the record, we are confident that the district court did not rely on any clearly erroneous findings of fact in reaching its determination regarding the recanted testimony's likely effect on the jury's verdict. We cannot say that the district court abused its discretion in holding that the second element of the Gordon test has been met in this case.

3.Was Willis Taken By Surprise When the False Testimony Was Given, and Was He Unable to Meet This Testimony or Learn of Its Falsity Until After the Trial Concluded?

Although the Gordon court announced this as a factor to be considered when determining whether to grant a motion for a new trial based on the recantation of a government witness, it quickly backpedaled on the necessity of satisfying this element of the test before a new trial motion could be granted. After announcing the test as it was stated in Larrison, theGordon court then held that, given the "special circumstances" of its case, in which all of the testimony given by a material government witness is recanted, the requirements of surprise and an inability to "meet" the testimony at trial were not "pertinent." Gordon, 178 F.2d 900.

The U.S. Court of Appeals for the Seventh Circuit, the circuit that first developed this test, has also openly questioned the importance and necessity of the surprise element of the test. United States v. Leibowitz, 919 F.2d 482, 484-85 (7th Cir. 1990). In Leibowitz, Judge Posner explained that while surprise is surely relevant in some cases, it should not be required in every case, especially where "the principal (though not the only) evidence of guilt is the testimony of an accomplice or eyewitness, [and] the only resource of the defendant in unmasking the falsity, even with all the advance warning in the world, may be cross-examination, which -- much mythology to the contrary notwithstanding -- is not an infallible lie detector." Id. at 484.

Just as in Gordon and Leibowitz, a crucial government witness in this case has recanted all of his damaging testimony. We follow both Gordon and Leibowitz in holding that the satisfaction of the final element of the Gordon/Larrison test is not a condition precedent to the defendant receiving a new trial. As the district court stated in this case, "[i]f Warren's testimony concerning Defendant's clothes and living residence was false, Defendant surely knew of it at the time of his testimony." J.A. at 57 (Dist. Ct. Order). Furthermore, while the defense counsel was able to impeach Richard Warren on the witness stand, there is only so much the attorney could do to refute Richard Warren's potentially false testimony stating that he had seen the defendant wearing the jacket which was later found in the padlocked closet at 2267 E. 83rd St.

The granting of a defendant's motion for a new trial is a substantial remedy that must be exercised with caution. Nevertheless, we review the trial court's decision with respect to these motions only for an abuse of discretion. Inherent in this standard of review is the notion that the district court, the trier of fact, is in a far superior position to judge the credibility of the witnesses and the evidence presented at trial, and to decide whether a new trial is needed to avoid the substantial risk of an injustice. The district court in this case, after presiding over the trial and examining the credibility of the witnesses and the evidence, believed, in light of Richard Warren's recantation, that an injustice was likely if Willis's drug possession convictions were allowed to stand in the face of the government's tainted testimonial evidence. From our review of the record, we find no basis upon which to state that the district court, in arriving at its conclusion, abused its discretion.

For the foregoing reasons, we AFFIRM the district court's decision granting the defendant's motion for a new trial.

ALICE M. BATCHELDER, Circuit Judge, concurring.

I agree with the majority that this case calls for application of the standard announced in Gordon v. United States, 178 F.2d 896 (6th Cir. 1949), and that under that standard the district court did not abuse its discretion. I write separately for two reasons.

First, in discussing the requirement of Gordon that "the party seeking the new trial was taken by surprise when the false testimony was given and was unable to meet it or did not know of its falsity until after the trial," 178 F.2d at 900, the majority holds that "the satisfaction of the final element of the Gordon/Larrison test is not a condition precedent to the defendant receiving a new trial." Maj. Op. at 20. This holding sweeps too broadly and misconstrues this circuit's opinion inGordon and the Seventh Circuit's decision in Larrison.

In Gordon, the court "backpedaled," maj. op. at 20, on the necessity for a defendant to satisfy this element of the standard based upon the peculiar facts presented there:

This last ground does not seem pertinent to our consideration of the special circumstances of this case, as, certainly, if all of the testimony given against a convicted man were recanted, it would not appear necessary to show that the party seeking a new trial was taken by surprise and was unable to meet such testimony or did not know of its falsity until after the trial.

178 F.2d at 900 (emphasis added). As indicated by this passage, "the special circumstances" presented in the Gordon case consisted of the recantation by the government's principal witness of his testimony implicating the defendant in the charged offense. Indeed, this witness presented virtually the only evidence against the defendant. 178 F.2d at 897-99. See also Gordon v. United States, 164 F.2d 855, 857 (6th Cir. 1947) (affirming the conviction and noting that "the conviction was based principally upon the testimony of Herman Frank Banning, a co-defendant under the indictment"). Accordingly, the court reasoned that since "all of the testimony given" against the defendant had been recanted, requiring satisfaction of the "surprise element" of the standard for a new trial was not necessary. In doing so, the Gordon court did not abandon the "surprise element" or hold that a defendant need not show surprise before succeeding on a motion for a new trial.

Similarly, the Seventh Circuit has preferred a case-by-case approach to the determination of whether a defendant must satisfy the "surprise element" of the Gordon/Larrison standard for granting a new trial to outright abandonment of the inquiry altogether. In United States v. Leibowitz, 919 F.2d 482, 484-85 (7th Cir. 1990), the Seventh Circuit rejected the government's contention that a defendant must always show surprise. Although the court questioned whether the Larrisoncourt erroneously adopted the requirement of surprise, Leibowitz makes clear that surprise remains a consideration except in the narrow class of cases in which the government's principal witness recants his testimony after trial:

Surprise is relevant, surely. If the defendant had every opportunity to meet the allegedly false testimony at trial, his failure to unmask its falsity at that time is some evidence that the testimony was true. But why the defendant should be required to demonstrate surprise in every case of recantation baffles us. In a case such as this in which the principal (though not the only) evidence of guilt is the testimony of an accomplice or eyewitness, the only resource of the defendant in unmasking the falsity, even with all the advance warning in the world, may be cross-examination, which--much mythology to the contrary notwithstanding--is not an infallible lie detector.

919 F.2d at 484. Subsequent decisions demonstrate that the Seventh Circuit does not regard Leibowitz as eliminating the need for a defendant to show surprise. See, e.g., Olson v. United States, 989 F.2d 229, 231-33 (7th Cir. 1993) (concluding that an inquiry into surprise was appropriate on the facts of the case).

For these reasons I think that a proper reading of Gordon and Leibowitz does not support the majority's sweeping statement that "the satisfaction of the final element of the Gordon/Larrison test is not a condition precedent to the defendant receiving a new trial." Maj. Op. at 20. To the extent that the majority intends to follow Gordon and Leibowitz in limiting the force of this holding to the uncommon situation in which a principal government witness recants his testimony, I do not think that the facts of this case justify reliance on Gordon or Leibowitz to excuse the defendant from showing surprise. As the majority's summation of the evidence adduced at trial shows, the government presented substantial evidence of the defendant's guilt other than the rebuttal testimony of Richard Warren. Therefore, this case is not one of those rare "special circumstances" in which the government's only or principal witness recants. Leibowitz, 919 F.2d at 484;Gordon, 178 F.2d at 900. In any event, I do not think the district court's conclusion, based upon the facts of this case, that the defendant was surprised by Warren's testimony constitutes an abuse of discretion.

More importantly, I am deeply troubled by the way in which the district court conducted the defendant's trial. After reviewing the record of the proceedings before the district court in its entirety, I am at a loss to explain many of the court's evidentiary and procedural rulings. For example, the record shows that the defendant placed Charles and William Warren, two of Richard Warren's brothers, on his witness list. Notwithstanding a separation order entered by the district court, defense counsel allowed Charles Warren to sit in the courtroom during the trial. When the prosecution then sought to call Charles Warren to the stand as a rebuttal witness, defense counsel objected on the ground that he had seen the proffered witness in the courtroom on two separate occasions in violation of the separation order. During his voir dire, Charles Warren informed the court that he had in fact observed part of the trial over two days. Based upon this admission the district court sustained the defense objection and precluded the prosecution from calling Charles Warren. In short, the district court penalized the United States for the violation of a separation order when defense counsel had listed Charles Warren as a potential witness, observed him in the courtroom over two days, and failed to take any action to notify the court of the violation or to remedy it. This and numerous other rulings, which it is not necessary to list here, lead me to conclude that the district court effectively prevented the prosecution from putting on its case. Since none of these rulings are before this court, however, I join the majority in concluding that the district court did not abuse its discretion by granting the defendant's motion for a new trial.