This appeal by the government comes from a directed verdict of acquittal following a jury verdict of guilty in a highly publicized criminal trial implicating Pennsylvania and Philadelphia public officials.1 Three questions are presented: whether there was a variance between the indictment and the proof; if so, whether it prejudiced substantial rights of the defendants; and if so, whether the proper remedy was a new trial or acquittal. Defendants were charged with participating, with unindicted co-schemers, in a single political patronage scheme, designed to defraud the Commonwealth of Pennsylvania, and which utilized the United States mails in violation of 18 U.S.C. § 1341. The trial judge, determining that there was a variance between the offenses charged in the indictment and the proof offered at trial and that substantial rights of the defendants were prejudiced thereby, entered a directed verdict of acquittal. We affirm.I.

Defendants, Peter J. Camiel, Vincent J. Fumo, Thomas M. Nolan, and Vincent F. Scarcelli were charged in a forty-four count indictment with violating the federal mail fraud statute, 18 U.S.C. § 1341, between December 1974 and December 1978. From 1974 until the end of May 1976 Camiel was Chairman of the Philadelphia Democratic County Executive Committee (City Committee) and Fumo was his assistant in charge of patronage. During the entire four-year period, Nolan was Majority Leader of the Pennsylvania Senate and in charge of the Senate Special Leadership Account (Senate payroll); Scarcelli, a state representative, was Chief Clerk of the Pennsylvania House of Representatives and in charge of House Per Diem and Staff Salary Accounts (House payroll). Each of these payroll accounts was intended to provide assistance to members of the respective legislative body, but were not intended to be used to pay salaries, wages, or expenses of persons who performed no services for state legislators.2

The indictment charged that the defendants, as well as unindicted co-schemers Henry J. Cianfrani, then a state senator from Philadelphia, and Martin Weinberg, Camiel's successor as City Committee Chairman, participated in a scheme of placing Democratic party loyalists on the House and Senate payrolls with the understanding that they were not required to perform any work for the Pennsylvania legislature. Because these payrolls were intended solely to assist state legislators, the hiring of these "no-show" employees constituted misuse of the payroll funds and worked a fraud on the Commonwealth of Pennsylvania. Moreover, the government alleged, because the "no-show" employees' payroll checks were mailed to them from Harrisburg, the scheme also violated the federal mail fraud statute, 18 U.S.C. § 1341. Under the government's theory of the case, there was a single fraudulent political patronage scheme with the City Committee at the center and with the various participants united by one common goal: increasing the power of the Philadelphia area Democratic party by securing "no-show" jobs with the Pennsylvania legislature for party loyalists.

From December 1974 through at least May 1976 there was an ongoing political battle within the Democratic party in Philadelphia for control of the City Committee between those members aligned with then Mayor Frank Rizzo and those opposed to him. Because Camiel was part of the anti-Rizzo faction, Rizzo denied city government patronage to the City Committee during Camiel's tenure as Chairman. Through his political affiliation with then Governor Milton J. Shapp, and his previous support for both Nolan and Scarcelli, however, Camiel was afforded access to the previously described House and Senate payroll accounts to fund his patronage requests. Nolan or Scarcelli would approve Camiel-recommended party loyalists for "no-show" jobs.

The power structure of the Democratic party in Philadelphia changed drastically in May 1976 when Weinberg, a Rizzo supporter, defeated Camiel in his bid for reelection as Chairman of the City Committee. Immediately upon taking office, Weinberg demonstrated his animosity toward the Camiel faction by firing six "no-show" patronage employees on the Senate payroll who had been recommended by Camiel during the December 1974 to May 1976 period. By June 30, 1976, both Camiel and Fumo had been removed from power on the City Committee. Thereafter, Fumo, who retained his position of ward leader, performed only a minor role in the alleged schemes, and Camiel, as the government concedes, no longer participated in any fraudulent patronage schemes.

According to trial testimony, Camiel and Fumo were aware that both the House and Senate payrolls were being used to fund "no-show" employees. Although this bicameral involvement continued even after the change in leadership of the City Committee, there is a substantial question whether Nolan in the Senate and Scarcelli in the House were aware of each other's activities. The only evidence from which the existence of such a common House-Senate scheme could be inferred involved Ms. Ann Moss who had been given a patronage position on the House payroll while Camiel was Chairman of the City Committee. Moss switched allegiances and supported Weinberg in his successful bid to oust Camiel from the City Committee Chairmanship. In return for her support, Weinberg first removed her from the House payroll and then obtained a similar "no-show" position for her on the Senate payroll, but as a contract rather than a salaried employee. Moss then approached Nolan inquiring if she now had the same employee benefits she had received while on the House payroll. He informed her that as a contract employee she was not eligible for any employee fringe benefits. The government adduced no other evidence from which it could be inferred that Nolan knew that Moss had been a Philadelphia "no-show" employee in the lower house.

Because of his ill health, Scarcelli was severed from the other defendants for a separate trial which has not yet taken place. A jury convicted Nolan on thirty-two counts, Fumo on fifteen, and Camiel on eleven. Camiel was also found innocent on two counts. After the jury returned its verdict, the district court judge granted the defendants' motions for a directed verdict of acquittal. The trial judge found that there was a variance between the offenses charged in the indictment and the proofs offered at trial, that the variance prejudiced substantial rights of each defendant, and that the proper remedy was acquittal of each defendant rather than a retrial. The government contests each of these rulings on appeal.

The parties agree that a multi-defendant mail fraud scheme may be analyzed as a conspiracy. United States v. Camiel, 519 F. Supp. 1238, 1241 n.7 (E.D.Pa.1981). Further, case law indicates that the precepts governing variance in conspiracy cases apply equally to multi-defendant mail fraud prosecutions. United States v. Rodgers, 624 F.2d 1303, 1307 (5th Cir. 1980), cert. denied, 450 U.S. 917, 101 S. Ct. 1360, 67 L. Ed. 2d 342 (1981). Finally, we note that in a conspiracy case, the determination of whether there is a variance sufficient to justify a trial judge's reversal of a jury conviction is controlled by the teachings of Kotteakos v. United States, 328 U.S. 750, 66 S. Ct. 1239, 90 L. Ed. 1557 (1946).

In Kotteakos, thirty-two defendants were charged with participating in a single general conspiracy to obtain fraudulent government loans. At trial, the government proved the existence of at least eight conspiracies rather than the unitary scheme charged in the indictment. A jury convicted the defendants, and the court of appeals affirmed. In its argument before the Supreme Court, the government conceded the existence of variance, 328 U.S. at 756, 66 S.Ct. at 1243, but claimed that no rights of the defendants had been prejudiced. The Court disagreed and reversed. It held that the defendants had a substantial right "not to be tried en masse for the conglomeration of distinct and separate offenses committed by others...." Id. at 775, 66 S.Ct. at 1252. Although Kotteakos does not expressly provide a methodology to gauge the correctness of a trial judge's reversal of a jury conviction in a mail fraud scheme case, we agree with the court below that Kotteakos impliedly establishes a two-pronged test which any such reversal must satisfy: (1) there must be a variance between the indictment and the proof, and (2) the variance must prejudice some substantial right of the defendant. 519 F. Supp. at 1242.

First, the government argues that there was sufficient evidence to support a jury verdict of guilty on the mail fraud charges and that the trial judge's ruling on variance was error. We disagree. Our standard of review in addressing this issue is whether "there is substantial evidence, viewed in the light most favorable to the Government, to uphold the jury's decision." United States v. Palmeri, 630 F.2d 192, 203 (3d Cir. 1980) (quoting Burks v. United States, 437 U.S. 1, 17, 98 S. Ct. 2141, 2150, 57 L. Ed. 2d 1 (1978) ). The trial judge found variance, similar to that present in Kotteakos, when he ruled that the government's evidence in the instant case disclosed multiple mail fraud schemes rather than the single scheme charged in the indictment. 519 F. Supp. at 1244.

It is important to remember that, as the Government notes in its brief, the indictment charged a mail fraud scheme and not a conspiracy: two related but distinct offenses. A conspiracy requires the existence of an agreement among the alleged co-conspirators, but the federal mail fraud statute requires only that the co-schemers participate in a common scheme. Thus, it is the existence of a common scheme, and not any agreement among the parties to participate in it, that is critical. See United States v. Read, 658 F.2d 1225 (7th Cir. 1981).

Several factors militate against finding a common scheme in this case. One concerns the Chairmanship of the City Committee. It is undisputed that the Chairman was the key person for recommending applicants for the "no-show" patronage positions. As the facts indicate, Camiel was Chairman until June 1976 when he was defeated in his bid for reelection by Martin Weinberg who served as Chairman throughout the remainder of the indictment period. While actual agreement between the two Chairmen as to the operation of the "no-show" patronage scheme is not required to prove a common scheme, lack of agreement places the burden on the government to produce other evidence proving its existence. Further, actual hostility between the Chairmen may raise the possibility that the requisite common scheme was non-existent. Here, the government concedes that during the indictment period, the Camiel and Rizzo forces were involved in an "ongoing battle ... for control of the Democratic City Committee." Brief for Appellant at 12. This antagonistic relationship between the two political factions represented by the warring Chairmen supports the trial judge's assertion that "political and personal differences existing among the alleged co-schemers made their participation in a single scheme unthinkable and impossible." 519 F. Supp. at 1240.

The government appears to recognize that the animosity between the City Committee Chairmen negates the possibility of using them as the basis for a common scheme and argues that the City Committee itself was the element that unified the patronage activities into a common scheme. That is, regardless of who was Chairman, the City Committee as an organization was the one constant element in the scheme; the City Committee was the beneficiary of the schemes and the common purpose was the advancement of the Democratic party in Philadelphia.

Although we will presume, without deciding, that an organization can be the element that welds disparate activities into a common scheme, such an analysis simply does not comport with the facts here. The reason Camiel started using state level patronage funds was not to increase the power of the Democratic party in Philadelphia but rather because a fellow member of that same party, Mayor Rizzo, cut him off from the normal reservoir of patronage, the city government. Out of necessity, Camiel turned to state legislators Nolan and Scarcelli for assistance, not simply because they were Democrats, but because they were Democrats who he had previously supported. Also, when Weinberg became Chairman, he immediately fired six of the Camiel-recommended, Senate payroll "no-show" employees. This was not an act taken to increase the power and influence of the Democratic party-the six were concededly Democrats-but an act indicative of severe intra-party antagonism. Therefore, we cannot infer from the evidence that the City Committee qua organization was the galvanizing element that unified the various patronage activities into a common scheme.

The government also offered considerable evidence to prove that the "no-show" patronage positions which the Chairman of the City Committee obtained, emanated from both houses of the Pennsylvania General Assembly. With the exception of the Ann Moss case, however, there was a dearth of evidence presented that could link the relevant activities of the two legislative bodies, and thus prove the requisite common, unified patronage scheme. Typically, the City Committee, with its Chairman's approval, would communicate directly with either Nolan or Scarcelli and thereby obtain the desired patronage position. Because Nolan and Scarcelli each exercised independent control over his respective payroll account, there was no need for a coordinated effort.

We think that the antagonism which existed between Camiel and Weinberg, coupled with the City Committee's inability to be viewed as a galvanizing force, and the lack of any evidence, save for Ann Moss, to show a House-Senate linkage, combine to negate any allowable inference of a common, singular, or unitary patronage scheme involving the named co-schemers, the City Committee, and the two houses of the Pennsylvania legislature. We hold, therefore, that, from the evidence presented, when viewed in the light most favorable to the government, a jury could infer the existence of either two (Camiel-era and Weinberg-era) or perhaps even four (because of the separate House and Senate involvements) distinct schemes to defraud. Following Kotteakos, however, the evidence is insufficient as a matter of law to support an inference of a single, unitary scheme involving all alleged co-schemers from December 1974 to December 1978 as found by the jury and as charged in the indictment.3

The government next argues that even if there was variance, no substantial rights of the defendants were prejudiced thereby. We disagree. Because the determination of the existence of prejudice to the defendants is a legal conclusion based on a factual predicate, our standard of review is mixed. "We must accept the trial court's findings of historical or narrative facts unless they are clearly erroneous, but we must exercise a plenary review of the trial court's choice and interpretation of legal precepts and its application of those precepts to the historical facts." Universal Minerals, Inc. v. C. A. Hughes & Co., 669 F.2d 98 (3d Cir. 1981) (footnote omitted). Therefore, we must identify both the historical facts as found by the court below and the legal precepts it applied to those facts in reaching its decision.

In directing a verdict of acquittal, the district judge agreed with the defendants "that the prejudice in this case derived from the 'spillover of evidence' from one scheme to another." 519 F. Supp. at 1244. The significant considerations in determining the existence of prejudice in this case are existence of multiple schemes and spillover of evidence. As we mentioned above, we are persuaded that there was evidence supporting an inference that at least two and possibly four schemes existed, involving several co-scheming groups during the indictment period, but that the evidence would not support an inference of a single, unitary scheme. We hold that the trial judge did not err in determining that the government presented evidence of multiple schemes in the context of the offense charged. The narrative and historical facts based on the testimony adduced at trial that undergirded this determination were not clearly erroneous. Moreover, if the single or multiple nature of the activities is to be regarded as a basic or inferred fact, we hold the finding not to be clearly erroneous; if regarded as an "ultimate fact"-a mixed question of fact and law-we find no error in finding the fact or in choosing, interpreting, or applying a legal precept.

Likewise, we conclude that the trial judge did not err in determining that there was a "spillover of evidence." This is more a question of law than a finding of fact, for whether evidence is relevant in the context of an indictment is governed by legal rules of evidence. The evidence was perforce presented by the prosecution in a manner that lent itself to jury confusion. Because many of the prosecution's witnesses, such as Ann Moss, testified concerning activities that involved several defendants and that occurred throughout the indictment period, the government did not and could not present its case in a chronological order. Also, because each of the multiple schemes involved some, but not all, of the defendants and no defendant was implicated in each of the schemes, the government did not and could not adequately insulate any one defendant from the harmful spillover effects of evidence introduced against another defendant. Accordingly, we conclude that the trial court did not err in determining that defendants Camiel and Fumo suffered spillover effects from evidence of post-June 1976 schemes involving Weinberg, and that Nolan suffered spillover effects from evidence of schemes involving House payrolls.

Having decided that the trial court did not err in determining the presence of multiple schemes and the resultant "spillover of evidence," the next inquiry is whether the substantial rights of the defendants were prejudiced thereby. Kotteakos teaches that the substantial right to be protected is "the right not to be tried en masse for the conglomeration of distinct and separate offenses committed by others...." 328 U.S. at 775, 66 S. Ct. at 1252. The trial judge held that the evidence of multiplicitous schemes worked a prejudice to the substantial rights of the defendants because, even if there had been sufficient evidence to convict any one of the defendants on a single scheme, the volume and manner of presentation of the evidence created the likelihood of spillover: i.e., that the jury might have been unable to separate offenders and offenses and easily could have transferred the guilt from one alleged co-schemer to another. We hold that the trial judge did not err in applying this legal precept, that the second prong of the two-part Kotteakos test is satisfied, and, therefore, that reversal of the jury by the district court judge was proper. We now consider the question of the proper remedy.

The government argues that if this court should hold that reversal of the jury verdict is justified under Kotteakos, the proper remedy is retrial rather than acquittal. It contends that even if variance existed, severance under Rule 14, F.R.Cr.P., would resolve any problems of prejudice. That is, that the three defendants could be retried separately on the same indictment without encountering the difficulties presented in the original trial. Again we disagree. Because this decision necessitates a choice between the competing considerations of either a new trial or acquittal the standard of our review is plenary. See Universal Minerals, Inc. v. C. A. Hughes & Co., 669 F.2d 98, 103 (3d Cir. 1981).

The indictment spans the period from December 1974 to December 1978 and involves the House, Senate, and City Committee. Even if defendants were to receive separate trials, we feel that the "spillover" and "guilt by association" effects-the heart of the Kotteakos problem-would still exist. Significantly, in its brief the government argues that "other than House evidence with regard to Nolan and Weinberg-era evidence with regard to Camiel, all of the evidence would have been admissible against each defendant in separate trials of each individual scheme." Brief for Appellant at 70-71. This concession illustrates why even separate retrials, under this indictment, of each of the defendants would not cure the legal defect. So long as "all of the evidence (with some minor deletions) would have been admissible against each defendant in separate trials of each individual scheme," there would be a reoccurrence of the problem outlawed by Kotteakos. Because that retrial would be under the indictment alleging a single conspiracy, then whether the defendants be tried jointly or severally, a cordon sanitaire cannot be constructed to prevent the evidence of one scheme from infecting the charge of another. The "spillover" and "guilt by association" possibility that would persist is the blot of ink in the glass of milk; it simply cannot be removed by retrial.

To understand this is also to understand that the issue at bar is not the propriety of accepting relevant evidence under the indictment, but rather a recognition that under the indictment the evidence is relevant. Because of the breadth of the current indictment, any evidence of a fraudulent "no-show" patronage scheme encompassing the City Committee during this time period would meet the threshold test of relevancy.4

Repeating the error it made in the district court, the government, in both its brief and oral argument, fails to appreciate the very nice, but very important, distinction between the relevancy of evidence allowable under the indictment generally and the prejudicial effect of its admission when that evidence, though relevant to the indictment, constitutes a fundamental variance to the indictment. Judge Clifford Scott Green recognized this problem in the proceedings below. On six separate occasions prior to trial, he warned the prosecutor of the danger of proving multiple schemes at trial when the grand jury charged only a single scheme in the indictment. He advised that "the results might be disastrous," App. at 136A, gave notice of the dangers of "a misuse of the evidence," id. at 141A, and alerted the government to the possibility that the "evidence will not support a one-scheme theory" and that "reading the Indictment, you have set forth one continuing-scheme theory." Id. at 143A. Judge Green pointedly stated to the prosecutor:

If I get into the middle of the trial and find that there is no one scheme, but that there are just a lot of people hooked together, in order to satisfy your desire to have cumulative evidence, then you have a problem.

Id. at 185A. Additionally, the trial judge warned about "sanctions ... which would not be ... to your liking," id. at 186A, and "if in fact a new scheme starts up with a new chairman, it may be very, very unfair to have evidence in 30 or 40 other counts against someone." Id. at 187A.

The defect in the indictment that the trial judge perceptively observed at the pre-trial stage-that the government could not try a case bottomed on multiple schemes on the basis of a single scheme indictment-is still present today. This defect cannot be cured by severance and separate trials. It is the indictment itself, therefore, that is the root cause of the prejudicial variance; it is the indictment itself whose breadth allows for, even encourages, the full range of evidence offered at the trial below; it is the indictment itself that is fatally flawed.

Once this is recognized the proper remedy is clear. We may identify the flaws in the current indictment but correcting those flaws is beyond our power. An indictment may not be amended, changed, or altered in any material respect except by resubmission to the grand jury. Russell v. United States, 369 U.S. 749, 82 S. Ct. 1038, 8 L. Ed. 2d 240 (1962); United States v. Goldstein, 502 F.2d 526 (3d Cir. 1974) (in banc). This court can neither act for a grand jury, nor speculate whether a grand jury would have indicted the named defendants had it realized that the indictment as written was overbroad.

Even if a court sought to limit a retrial in a highly compartmentalized fashion so that only evidence of a particular scheme could be introduced against the appropriate defendants-as suggested by the government's proposed severance remedy-neither this appellate court nor the district court possesses the authority to order it. Such a procedure would be based on an unwarranted assumption-that the grand jury would have returned indictments against the individual defendants based on discrete multiple schemes. We cannot make that assumption because no court may speak for the grand jury. The grand jury and not the court must fashion the precise charge against a defendant; the grand jury and not the court must determine whether an indictment will be returned on the theory of a single scheme or multiple schemes; and the grand jury and not the court must determine both the content and perimeters of any indictment.5

Therefore, a new trial with the possibility of severance is not the answer, and our only alternative is to affirm the judgment of acquittal. We offer no opinion as to the consequences of future activity, if any, in these proceedings.

The judgment of acquittal will be affirmed in all respects.