71 F.3d 408
D.C. Cir.
1996
SENTELLE, Circuit Judge:

Charles N. Lloyd, Jr., a tax preparer, was convicted on three counts of aiding and abetting in the preparation of false federal income tax returns and one count of first degree fraud in causing a false District of Columbia income tax return to be filed. On a previous appeal, we remanded the case for reconsideration of Lloyd’s claim that he was unfairly prejudiced by the trial court’s refusal to order the government to disclose the returns of the government’s taxpayer witnesses for the years immediately preceding the years of the returns upon which the indictment was based. United States v. Lloyd, 992 F.2d 348 (D.C.Cir.1993). On remand, the District Court concluded that the undisclosed tax returns probably would not have produced an аcquittal on any of the four counts. It accordingly denied Lloyd’s motion for a new trial, and Lloyd appeals that denial to this court. Because we conclude that the District Court applied the wrong legal standard to the new trial motion and erroneously determined that a new trial was not required, we reverse the convictions in Counts 7 and 11 and remand fоr new trial. We uphold the convictions in Counts 3 and 5.

I. Factual Background

In 1991, Charles N. Lloyd, Jr., was convicted under 26 U.S.C. § 7206(2) (1988) and 18 U.S.C. § 2 (1988) on three counts of aiding and abetting in the preparation of false federal income tax returns and under 22 D.C.Code ANN. §§ 3821(a), 8822(a)(1) and 105 (Repl.1989) on one count of first degree fraud in helping to file a false District of Columbia income tax return. Since the early 1980s, Lloyd had owned аnd operated a tax preparation business in Southeast Washington, D.C., called Delta Tax Service. Many of Lloyd’s clients were unable to pay him at the time he prepared their returns, so he arranged for their tax refund checks to be sent to his post office box. The client would pay Lloyd out of the refund proceeds. Following a lengthy investigаtion, Lloyd was indicted on numerous tax fraud counts stemming from tax returns he had prepared for some of his clients.

Lloyd sought pre-trial discovery of tax returns for his clients named in the indictment for the three years preceding the years of the returns which Lloyd had prepared. The government obtained some of the requested prior returns, but determined that 26 U.S.C. § 6103 (1988) prohibited disclosing the tax returns without a court order. Lloyd requested such an order, arguing that if the earlier tax returns revealed that the clients had filed fraudulent returns in the past, it would support his defense that the clients themselves, and not Lloyd, supplied the false information in the indictment returns. The court refused to issue the order, concluding that Lloyd had failed to meеt his burden of showing that the returns would be useful in his defense. After trial, the jury convicted Lloyd on four counts: Count 3, aiding and abetting in the preparation of a false federal income tax return for Phyllis Burton for the tax year 1984; Count 5, the same charge involving Juanita Pressley’s 1985 return; Count 7, the same charge involving Diane Caldwell’s 1985 return; and Count 11, first degree fraud in causing D.C. to be deprived of tax revenues. Count 11 involved the D.C. tax returns of Calvin Toler, Diane Caldwell, Michael Worthy, Thelma Davis and Donald Cooper.

On appeal, we examined Lloyd’s claim that the District Court had erred by inflating the materiality standard in Fed.R.CRIM. P. 16 because of the statutory prohibitions on disclosing the tax returns without a court order. We concluded that a materiality standard mоre demanding than Rule 16 might be appropriate in some cases involving the disclosure of tax returns for discovery purposes, but that Lloyd’s case was not one of them. We then remanded the case for consideration of a new trial motion, instructing the District Court to appraise the returns’ materiality under the normal standard of Rule 16. United States v. Lloyd, 992 F.2d 348, 352 (D.C.Cir.1993). On remand, the parties agreed to stipulate to the materiality of the returns for discovery purposes and allow the court to proceed directly to the merits of the new trial motion. The District Court applied Thompson v. United States, 188 F.2d 652 (D.C.Cir.1951), which set forth the test for whether newly discovered evidence requires a new trial. The court denied the new trial motion, concluding that the undisclosed returns (1) were irrelevant to the convictions in Counts 3 and 5, (2) were of only slight impeachment value as to the conviction in Count 7 and were unlikely to produce an acquittal, and (3) were either cumulative or useless for impeachment as to the conviction in Count 11.

II. Legal Analysis

A History and Standards

In denying Lloyd’s motion for a new trial, the District Court applied the rule of Thompson v. United States, 188 F.2d 652 (D.C.Cir.1951):

To obtain a new trial beсause of newly discovered evidence (1) the evidence must have been discovered since the trial; (2) the party seeking the new trial must show diligence in the attempt to procure the newly discovered evidence; (3) the evidence relied on must not be merely cumulative or impeaching; (4) it must be material to the issues involved; and (5) of such nature thаt in a new trial it would probably produce an acquittal.

Id. at 653. The District Court based its denial squarely on the grounds that the undisclosed returns ‍​​‌‌‌‌​‌​‌​​‌‌​‌‌​‌‌‌‌‌​​‌‌‌‌​‌​​‌​​‌​‌​‌​​‌​‌​​‍would probably not have produced an acquittal. However, we have already made it clear that the Thompson test does not apply where the evidence in question was not newly discovered, but was available to the prosecution at the time of trial, at least when the prosecution withheld that evidence in violation of due process standards under Brady v. Maryland, 373 U.S. 83, 86-88, 83 S.Ct. 1194, 1196-97, 10 L.Ed.2d 215 (1963). United States v. Kelly, 790 F.2d 130, 133 (D.C.Cir.1986). “[T]he fact that such evidence was available to the prosecutor and not submitted to the defense places it in a different category than if it had simply been discovered from a neutral source after trial.” United States v. Agurs, 427 U.S. 97, 111, 96 S.Ct. 2392, 2401, 49 L.Ed.2d 342 (1976).

The present facts present the distinct question of whеther the Thompson standard applicable to newly discovered evidence or the Brady standard applicable to evidence which existed and was known to the prosecution at the time of trial applies in a case in which the prosecution withheld the disputed evidence but without wrongdoing on the part of the government and with full disclosure to the court. We hold that it is the Brady analysis which must govern. The purpose in Brady is not to punish a wrongdoing prosecutor, but rather to аssure that the defendant was not convicted without due process of law. As the Supreme Court stated in Brady, “suppression by the prosecution of evidence favorable to an accused ... violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” Brady, 373 U.S. at 87, 83 S.Ct. at 1196-97 (emphasis added). Thus, the District Court should have analyzed the effect of the erroneous withholding of the tax returns under the analysis applicable to Brady evidence rather than the Thompson standard applicable to newly discovered evidence.

That leads us to the questions: What standard applies to Brady evidence, and was that standard met in the present case? In Kelly, we applied the Supreme Court’s test of whether the defendant’s lack of access to undisclosed evidence justifies a new trial:

[ T]he District Court must determine whether the undisclosed evidenсe was “material” to Kelly’s conviction_ The Supreme Court ‍​​‌‌‌‌​‌​‌​​‌‌​‌‌​‌‌‌‌‌​​‌‌‌‌​‌​​‌​​‌​‌​‌​​‌​‌​​‍has recently held that undisclosed information is material if “there is a reasonable probability that, had the evi- denee been disclosed to the defense, the result of the proceeding would have been different. A ‘reasonable probability’ is a probability sufficient to undermine confidеnce in the outcome.” United States v. Bagley, 473 U.S. 667, 682, 105 S.Ct. 3375, 3384, 87 L.Ed.2d 481 (1985).

Kelly, 790 F.2d at 135-36. If the undisclosed evidence is “material” in the Bagley/Kelly sense, then the defendant is entitled to a new trial. The Supreme Court recently elaborated on the meaning of materiality under Bag-ley, stressing that the relevant inquiry focuses on the fairness of the trial the defendant actually received rather than on whether a different result would probably have occurred had the undisclosed еvidence been revealed:

[ A] showing of materiality does not require demonstration by a preponderance that disclosure of the suppressed evidence would have resulted ultimately in the defendant’s acquittal Bagley’s touchstone of materiality is a “reasonable probability” of a different result, and the adjective is important. The quеstion is not whether the defendant would more likely than not have received a different verdict with the evidence, but whether in its absence he received a fair trial, understood as a trial resulting in a verdict worthy of confidence. A “reasonable probability” of a different result is accordingly shown when the Government’s evidentiary suppression “undermines cоnfidence in the outcome of the trial.” Bagley, 473 U.S. at 678, 105 S.Ct. at 3381.

Kyles v. Whitley, — U.S. -, -, 115 S.Ct. 1555, 1566, 131 L.Ed.2d 490 (1995). Kyles also made it clear that the Bagley analysis is a one-step process — “once a reviewing court applying Bagley has found constitutional error there is no need for further harmless-error review.” Kyles, — U.S. at -, 115 S.Ct. at 1566. No Bagley error could ever be harmless, the Court noted, because a reasonable probability of a different result “necessarily entails the conclusion that the suppression must have had substantial and injurious effect or influence in determining the jury’s verdict.” Id. (citations and internal quotation marks omitted). This court’s prior decisions are in accord.

The government argues in a footnote in its brief that the proper remedy for a District Court’s use of the wrong legal standard in ruling on a new trial motion is “to remand the case back to the district court ... where the appropriate fact-based determination can best be made” under the proper standard. Brief for Appellee at 21 n. 18. While it is true that the determination of materiality for Brady purposes “is inevitably fact-bound and so is committed to the trial judge in the first instance,” Kelly, 790 F.2d at 136, it is also true that the facts necessary tо the determination of this motion are clear and undisputed. The undisclosed returns are part of the record, and no further facts ‍​​‌‌‌‌​‌​‌​​‌‌​‌‌​‌‌‌‌‌​​‌‌‌‌​‌​​‌​​‌​‌​‌​​‌​‌​​‍need be determined for resolution of the new trial motion. The remaining question is one of law, not fact — did the nondisclosure of the returns violate Lloyd’s Brady rights as a matter of law? Such a determination was impossible the first time this eаse was before us, for the returns were not part of the record at that time. See United States v. Lloyd, 992 F.2d 348, 351-52 (D.C.Cir.1993) (remanding for reconsideration of the Rule 16 materiality of the undisclosed returns). But now that the returns are a part of the record, we can determine the proper resolution of Lloyd’s new trial motion without overstepping the bounds of appropriate apрellate jurisprudence.

When we last considered this case, we discussed different ways in which the undisclosed returns might be materially exculpatory. First, if a taxpayer who claimed that Lloyd was the source of the fabrication in the indictment return could be shown to have made similar claims in prior returns which Lloyd did not prepare, that evidence would helр Lloyd demonstrate that the taxpayer, not Lloyd, was the source of the fabrication. Second, if a taxpayer took the stand and claimed that Lloyd was the source of the false information in the indictment returns, these prior returns not prepared by Lloyd would “have the makings of a promising tool for impeachment.” Lloyd, 992 F.2d at 351. In this context, we note that the Supreme Court has specifically held that there is not “any difference between exculpatory and impeachment evidence for Brady purposes.” Kyles, — U.S. at -, 115 S.Ct. at 1565. With that background, we will proceed to examine each conviction and determine whether the returns are materially exculpatory under Bagley and Kyles.

B. Application

1. Count 7

Count 7 alleged that Lloyd aided and abetted in the preрaration and presentation of a false federal income tax return for the year 1985 for the taxpayer Diane Caldwell. The District Court did not order prosecution disclosure of any prior returns for Caldwell at trial, but the record now contains Caldwell’s 1984 federal return, which is under seal. (The government placed the 1984 return into the sealed record in the proceedings on the new trial motion in the court below after stipulating to the materiality of the return for discovery purposes.) Having reviewed this sealed 1984 return, we conclude that it contains materially exculpatory evidence of just the sort contemplated in our previous opinion in this case. This undisclosed evidence raises a reasonable probability of a different result had it been disclosed at trial. We therefore hold that the District Court erred in denying Lloyd’s new trial motion as to Count 7.

2. Counts 3 and 5

Count 3 mirrored Count 7 except it involved the tax return of Phyllis Burton for the tax year 1984. The record, however, contains no prior-year return for Burton. Apparently, the government was unable to produсe any prior returns for Burton. Lloyd, 992 F.2d at 350. When this case was previously before this court, Lloyd claimed that the government’s search was inadequate, but we held at that time that he had failed to raise that claim below and was therefore barred from raising it on appeal. Id. Count 5 also mirrored Count 7 except it involved the tax return of Juanita Pressley for the tax year 1985. The record also contains no prior returns for Pressley.

Lloyd challenges his convictions in both Counts 3 and 5 on the grounds that the nondisclosure of all the prior returns that were retrieved corrupts all four convictions, ‍​​‌‌‌‌​‌​‌​​‌‌​‌‌​‌‌‌‌‌​​‌‌‌‌​‌​​‌​​‌​‌​‌​​‌​‌​​‍not just those based on later returns of the individuals for whom prior returns were retrieved. To support this argument, he points to the statement in Kyles that Bagley materiality is defined “in terms of suppressed evidence considered collectively, not item by item.” Kyles, — U.S. at -, 115 S.Ct. at 1567. More specifically, “the effect of each nondisclosure must not only be considered alone, for the cumulative effect of the nondis-closures might require reversal even though, standing alone, each bit of omitted evidence may not bе sufficiently ‘material’ to justify a new trial.” United States ex rel. Marzeno v. Gengler, 574 F.2d 730, 736-37 (3d Cir.1978). Having considered the nondisclosed returns as a whole, however, we are not convinced that they would have provided any assistance to Lloyd in impeaching either Burton or Pressley or in buttressing his version of events as to their returns. We accordingly reject this “corruption” argument.

As to Count 5 alone, Lloyd points to a comparison the prosecution made between Lloyd’s testimony regarding the Caldwell indictment return and his testimony regarding the Juanita Pressley return, the basis of Count 5. According to Lloyd, this “linking” of the two returns shows that the nondis-closed Caldwell return could have helped him to boost his credibility on Count 7, thereby having a parallel effect as to Count 5. But examination of thе trial transcript shows the weakness of this argument. See Trial Transcript at 81-82. The prosecutor was attempting to show that Lloyd had falsified “original worksheets” for those returns which were being investigated by the IRS, including the Caldwell and Pressley returns. All that Lloyd could have shown with the undisclosed Caldwell return now in the sealed record was that he was not the source of the false information in the Caldwell indictment return itself. The prior Caldwell return offers no evidence to counter the prosecutor’s suggestion that Lloyd falsified “original worksheets” for all the returns under investigation. Even if Lloyd had not falsified the Caldwell indictment return, he had plenty of incentive to falsify the “original worksheets” for the returns being investigated by the IRS — he had failed to presеrve adequate records of the sources of the information in those returns. We therefore reject Lloyd’s invitation to grant a new trial as to Count 5.

3. Count 11

Count 11 alleged first degree fraud in violation of 22 D.C.Code Ann. §§ 3821(a), 3822(a)(1) and 105 (Repl.1989) by causing the District of Columbia to be deprived of more than $250.00 in tax revenues by helping to prepare false District of Columbia income tax returns. The returns supporting Count 11 were the 1986 tax year returns of Calvin Toler, Diane Caldwell, Michael Worthy, Thelma Davis and Donald Cooper. Lloyd had a copy of one of Davis’ prior returns in his files, so he was able to use it against her at trial. Lloyd, 992 F.2d at 351. The government was able to retrieve prior-year returns for Toler, Caldwell, Worthy and Cooper. We have already held that the Caldwell prior-year return contains materially exculpatory evidence under Bagley and Kyles. After reviewing the sealed returns for Toler, Worthy and Cooper, we have determined that they also contain materially exculpatory evidence of the sort we contemplated in our previous Lloyd opinion. As the government points оut, the jury only had to find a violation in one of the five returns to justify the conviction under Count 11. It is therefore possible that the jury relied solely on the Davis return and that the prior-year returns for the other four individuals would have had no effect on their decision to convict Lloyd under Count 11. But we must construe the undisclosed evidence, in this case prior-year tax returns for four of the five individuals named in the count, “collectively, not item-by-item.” Kyles, — U.S. at -, 115 S.Ct. at 1567. We therefore hold that the combined weight of the materially exculpatory evidence contained in ‍​​‌‌‌‌​‌​‌​​‌‌​‌‌​‌‌‌‌‌​​‌‌‌‌​‌​​‌​​‌​‌​‌​​‌​‌​​‍the undisclosed Caldwell, Toler, Worthy and Cooper returns merits a new trial on Count 11.

4. Other uses of the earlier returns

Lloyd suggests other theories under which the earlier returns now part of the sealed record сould have been used at trial to impeach the government’s witnesses as to all four counts on which convictions were obtained. We have already held that the undisclosed returns contain materially exculpatory evidence as to Counts 7 and 11, so it is unnecessary to consider Lloyd’s other theories for those counts. Lloyd is free, of course, to make use of these additional impeachment strategies against government witnesses in a new trial on these two counts. None of the undisclosed returns in the sealed record are related to Counts 3 or 5, however, so Lloyd’s theorized impeachment strategies could have had no effect on them. Lloyd’s theories give us no reason to order a new trial as to these two counts.

III. Conclusion

For the foregoing reasons, we hold that the relevant undisclosed returns constitute materially exculpatory evidence under Bag-ley, Kyles and Kelly as to the convictions in Counts 7 and 11. The District Court erred in denying Lloyd’s new trial motion as to these two counts. We affirm the District Court’s denial of the new trial motion as to the convictions in Counts 3 and 5, but we reverse the District Court’s denial of the motion as to Counts 7 and 11 and remand with instructions to grant the motion as to the latter two counts.