JAMES HUNTER, III, Circuit Judge:
This is an appeal by United States Gypsum Company (USG), National Gypsum Company (National), Georgia-Pacific Corporation (G-P), The Celotex Corporation (Cel-otex), Andrew J. Watt, Colon Brown, and J. P. Nicely, defendants in a complex case under Section 1 of the Sherman Act, 15 U.S.C. § l.1 Each corporate defendant manufactures, sells, and distributes gypsum board products, which are widely used in the construction industry. Watt is Executive Vice President of USG. Brown is Chairman of National’s Board of Directors, and Nicely recently retired as Vice President for Sales in National’s Building Products Division.2 For the reasons that follow, we reverse and remand.
I.
Gypsum board is produced in standard sizes, of which 4' X 8' X W is the most common. It consists of a layer of gypsum rock sandwiched between sheets of heavy paper. Since World War II, it has replaced gypsum plaster as the principal component of interior walls in all types of buildings.
The gypsum board industry is highly concentrated. Between 1960 and 1973, the number of producers varied from nine to fourteen, and the eight major producers accounted for more than ninety-four percent of national sales. The corporate appellants together account for more than seventy-five percent of national sales. There are no significant differences in quality or appearance among the standard types of board produced by the various manufacturers. Demand is determined by activity in the construction industry and is inelastic with respect to price. Because of the homogeneity of the different brands, a buyer’s decision to purchase one particular brand is generally based on price. Price discounts and changes in credit terms are the most important form of competition in the industry.
Between 1965 and 1970, the Department of Justice conducted several investigations of possible antitrust violations in the gypsum board industry. In 1971, a grand jury inquiry commenced, and another twenty-eight months of investigation ensued. On December 27, 1973, an indictment was returned in the Western District of Pennsylvania. It named the appellants and nine other defendants3 as participants in a conspiracy, starting before 1960 and lasting to the date of indictment, (1) to raise, fix, maintain, and stabilize the price of gypsum board; (2) to fix, maintain, and stabilize the terms and conditions of sale of gypsum board; and (3) to adopt and maintain uniform methods of packaging and handling, all in violation of Section 1 of the Sherman Act.
[118]*118All defendants moved for dismissal, alleging that unreasonable pre-indictment delay had worked a denial of their rights to due process of law as guaranteed by the fifth amendment to the United States Constitution. This motion was denied after a five-day evidentiary hearing. Subsequently, the nine defendants who are not parties to this appeal were convicted and sentenced on pleas of nolo contendere.4 Trial of the remaining defendants began on March 3, 1975, and the jury brought in its verdict on July 15,1975. The Government called thirty-five witnesses, appellants thirty. The trial record consists of more than 15,000 pages of transcript and some 6500 exhibits.
II.
At the outset, we are met with appellants’ contention that the due process clause of the fifth amendment requires dismissal of the indictment. They allege that the Government unduly delayed bringing the indictment and that this delay impaired appellants’ ability to present a defense. We affirm the district court’s refusal to dismiss the indictment.
In United States v. Marion, 404 U.S. 307, 92 S.Ct. 455, 30 L.Ed.2d 468 (1971), the Supreme Court indicated that “the Due Process Clause of the Fifth Amendment would require dismissal of the Indictment if it were shown at trial that the pre-indictment delay in this case caused substantial prejudice to appellee’s rights to a fair trial and that the delay was an intentional device to gain tactical advantage over the accused.” Id. at 324, 92 S.Ct. at 465. Because no prejudice was shown, however, the Marion Court upheld the indictment. Thus, there was no holding as to the specific requirements of a successful showing of prejudicial delay, and it was unclear whether both prejudice and governmental intent are necessary or whether prejudice alone would suffice. See, e. g., United States v. Barket, 530 F.2d 189, 194-195 & n.9 (8th Cir.), cert. denied, 429 U.S. 917, 97 S.Ct. 308, 50 L.Ed.2d 282 (1976). We need not resolve this question, because we find neither undue delay nor substantial prejudice.
We consider first the alleged undue delay. Appellants’ major argument is that the Government deliberately postponed empaneling the grand jury to await the outcome of a civil case, Wall Products Co. v. National Gypsum Co., 326 F.Supp. 295 (N.D.Cal.1971), involving charges of the same conspiracy. After a lengthy evidentiary hearing, however, the court below found that the interval between the initial investigations of the gypsum board industry and the bringing of the indictment was not unreasonable in light of the extraordinary magnitude of the case and the frequent dead ends encountered by the investigators. United States v. United States Gypsum Co., 383 F.Supp. 462, 464-70 (W.D.Pa.1974). As the district court noted, id. at 467, any delay in empaneling the grand jury resulted from the Government’s reluctance to frame criminal charges against appellants before learning whether plaintiffs in a civil action could carry their burden of proof. The district court’s findings are not clearly erroneous.
USG separately argues that the Government is collaterally estopped to deny intentional delay, relying on United States v. United States Gypsum Co., Civil No. 71-2467 (N.D.Cal., filed Dec. 30, 1971). In that suit against six gypsum companies, the Government sought to recover damages accrued before the start of the normal four-year period, claiming that defendants’ fraudulent concealment of their conspiracy had tolled the statute of limitations. The district court granted a motion for partial summary judgment, declaring that the Government did know, or should have known, of the alleged conspiracy prior to January 1, 1968. This finding, insists USG, estops the Government from denying delay in this case.
[119]*119In Scooper Dooper, Inc. v. Kraftco Corp., 494 F.2d 840, -844 (3d Cir. 1974), Judge Garth noted three requirements that must be satisfied before collateral estoppel will apply:
a) The issue decided in the prior litigation must be identical with the issue presented in the action in question;
b) The prior litigation must have resulted in a final judgent [sic] on the merits; and
c) The party against whom the estoppel is asserted must have been a party or in privity with a party, to the prior adjudication.
USG’s claim does not satisfy the first requirement. Governmental knowledge sufficient to preclude reliance on the doctrine of fraudulent concealment in a civil case is not identical with the availability of evidence sufficient to support the decision to bring criminal charges against the same defendants. Proof of the existence of the former does not entail proof of the existence of the latter. Therefore, there can be no collateral estoppel.
Having found no undue delay, we note that, in any event, appellants have not met their burden of proving substantial prejudice. In United States v. Dukow, 453 F.2d 1328 (3d Cir.), cert. denied, 406 U.S. 945, 92 S.Ct. 2042, 32 L.Ed.2d 331 (1972), this court, following Marion, refused to order the dismissal of an indictment where no substantial prejudice was shown. Accord, United States v. Benson, 487 F.2d 978, 984-85 (3d Cir. 1973). In Dukow, Judge Adams held the deaths of two potential witnesses not prejudicial, because there had been neither an offer of the testimony they would have given nor any indication that they possessed unique knowledge of the transaction in question. Dukow, supra, 453 F.2d at 1330. Here, as in Dukow, appellants argue that potential witnesses died in the lengthy interval between initial investigation and trial, but there is no showing that the deceased would have proffered testimony not merely cumulative of the evidence already at appellants’ disposal.
Of the thirty-six deceased potential witnesses listed, only three received particular attention from appellants. One was Melvin Quayle, who was depicted in other testimony as an emissary of the other alleged conspirators. Quayle’s denial before the grand jury of this role was read at trial. Appellants do not suggest that he would have testified any differently had he lived.
Appellants also insist that Donald Miller, late merchandising manager of Celotex, “would have made an excellent witness for the defense.” They vouchsafe no reason to believe, however, that Miller’s testimony would have been significantly more exculpatory than that offered by other witnesses.
Finally, appellants single out a USG memorandum of June 6, 1961, as a document the Government was able to exploit because its author, B. J. Nabors, was dead. Appellants, though, were free to offer alternative explanations of the document’s contents. Moreover, Nabors died in 1967, ten months before the date on which USG claims the Government should have known of the conspiracy and sought an indictment. Cases of this magnitude often last so long that documents are introduced when the authors are no longer available to dispel any inculpatory impressions the documents may create. To hold that this handicap, by itself, is a violation of due process would be to erect a practical barrier to suits of this sort.
Appellants argue that under United States v. Barket, supra, the Government bears the burden of proving that none of the unavailable witnesses possessed exculpatory evidence. We find this argument unpersuasive. Barket differs from the case sub judice in several important respects. First, the procedural posture of Barket was the reverse of that here: the district court had granted the motion to dismiss the indictment. Second, the Barket court expressed grave misgivings about the merits of the Government’s case. Third, the Eighth Circuit found the Government guilty of culpable negligence in delaying Barket’s indictment. Fourth, and most important, the Barket court explicitly found [120]*120that “Barket’s ability to defend himself on the crucial issue” had “undoubtedly” been impaired. Id. at 193.
None of these conditions obtains in the case before us. Because the court below refused to dismiss the indictment, we can order it dismissed only if we hold the district court’s findings clearly erroneous. We are unable to take the frankly skeptical view of the Government’s case on the merits, adopted by the Barket court. As we have already noted, the Government did not cause any undue delay in this case. Most importantly, we do not find that appellants’ ability to defend themselves was impaired. Because the Barket court explicitly reached the opposite conclusion, its statement concerning the burden of proof on the issue of missing witnesses’ possession of exculpatory evidence was dictum.5 And were the statement indeed necessary to the result in Barket, we would see no reason to depart from the clear teaching of Dukow and impose upon the Government the burden of proving that none of the unavailable witnesses possessed significant, exculpatory evidence. See, e. g., United States v. McGough, 510 F.2d 598, 604 (5th Cir. 1975).
III.
Thorough scrutiny of the voluminous record and careful consideration of the many issues raised on appeal convinces us that reversal is required.
A.
The Government’s case on price fixing centered on a practice appellants call “verification.” An officer of one gypsum board manufacturer would telephone a competing firm’s officer to determine the price at which the competitor was offering gypsum board to a specific customer. The scope, purpose, and duration of this activity are sharply disputed by the parties. The Government contends that the purpose of verification was to enable competitors to stabilize prices and “police” agreed-upon increases, that the calls involved broad discussions of present and future pricing policies, that the appellants verified daily, and that they continued to do so until 1973. Appellants insist that the only purpose of the calls was to ensure compliance with the Robinson-Patman Act (discussed below), that the conversations were limited in scope and number, and that the practice had been discontinued, except for a few isolated, unauthorized calls, before the start of the applicable limitations period, December 27, 1968. There is evidence to support each claim of both sides.
Appellants contend, however, that the court below erred in instructing the jury on the legal status of the practice. Although appellants assert that the evidence demonstrated that the purpose of certification was to comply with the Robinson-Patman Act, the trial court instructed the jury that any purpose was irrelevant so long as the jury [121]*121found that verification had a stabilizing effect on prices.6 The refusal to instruct the jury about the relevance of purpose was, for the following reasons, reversible error.
Section 2(a) of the Robinson-Patman Act, 15 U.S.C. § 13(a),7 prohibits sellers from discriminating in price between different purchasers of the same commodity. Section 2(b), 15 U.S.C. § 13(b),8 permits a seller to rebut a prima facie case under 2(a) by “showing that his lower price or the furnishing of services or facilities to any purchaser or purchasers was made in good faith to meet an equally low price of a competitor, or the services or facilities furnished by a competitor.” The Supreme Court has held that Section 2(b) places
on the seller the burden of showing that the price was made in good faith to meet a competitor’s. . . . [T]he statute at least requires the seller, who has knowingly discriminated in price, to show the existence of facts which would lead a reasonable and prudent person to believe that the granting of a lower price would in fact meet the equally low price of a competitor.
Federal Trade Commission v. A. E. Staley Mfg. Co., 324 U.S. 746, 759-60, 65 S.Ct. 971, 977, 89 L.Ed. 1338 (1945). This court declared in Viviano Macaroni Co. v. Federal Trade Commission, 411 F.2d 255, 259-60 (3d Cir. 1969), that the seller cannot meet this burden merely by accepting the purchaser’s word with respect to competitor’s offers; instead the seller must attempt to corroborate the report, as well as the reliability of the purchaser or other reporter.
Appellants claim that their verification practice was designed to comply with the dictates of Staley and Viviano. They assert that almost all discounting off list prices was not reflected in invoices, rendering that form of corroboration ineffective. In addition, they say, purchasers of the gypsum board were notoriously unreliable and often were discovered to have lied about a competitor’s offer in order to “whipsaw” a price cut. There is evidence to support these [122]*122assertions. Therefore, argue appellants, they had three choices: (1) offer the reduced price on the basis of the purchaser’s unconfirmed report and risk Robinson-Pat-man liability; (2) forego the price cut and risk losing the sale; or (3) call the competitor, verify his offer, and establish a section 2(b) defense to any Robinson-Patman charge concerning the price cut.
The countervailing policies of section 1 of the Sherman Act, however, complicate appellants’ exposition of the issue. The Supreme Court has dealt several times with the limits imposed by the Sherman Act on exchanges of information among competitors. In American Column & Lumber Co. v. United States, 257 U.S. 377, 42 S.Ct. 114, 66 L.Ed. 284 (1921), the Court held that section 1 of the Sherman Act condemned the exchange of specific price information with regard to specific customers, where the clear purpose was to stabilize prices. The Court reached the same result in United States v. American Linseed Oil Co., 262 U.S. 371, 43 S.Ct. 607, 67 L.Ed. 1035 (1923). Then in 1925, two cases upheld the exchange of information among competitors. In Maple Flooring Ass’n v. United States, 268 U.S. 563, 45 S.Ct. 578, 69 L.Ed. 1093 (1925), the Sherman Act was held to permit the exchange of average cost data relating only to closed transactions, despite an apparent stabilizing effect on price, where no intent to constrain individual competitive activity was found. And the Court in Cement Manufacturers Protective Ass’n v. United States, 268 U.S. 588, 45 S.Ct. 586, 69 L.Ed. 1104, approved the exchange of specific price information concerning specific customers, despite possible effects on prices, where the purpose of the dissemination scheme was to prevent buyers from defrauding sellers.
The Supreme Court has drawn a narrow line in deciding the data dissemination cases under section 1 of the Sherman Act. The difficulty derives from the fact that in competitive markets, information exchanges promote competition, while in oligopolistic markets, they depress competition. United States v. Container Corp. of America, 393 U.S. 333, 342-43, 89 S.Ct. 510, 21 L.Ed.2d 526 (1969) (Marshall, J., dissenting); P. Ar-eeda, Antitrust Analysis 13-16 (1974). Thus, the Court has attempted to fashion rules that will reach the latter, yet not amount to broad proscriptions embracing the former.
In Container, supra — the most recent data dissemination case — the facts were similar to those presented in the case sub judice. Eighteen manufacturers of corrugated containers accounted for 90% of the shipments of such products from the southeastern United States. The industry exhibited excess capacity and a downward trend in prices, yet twenty-one new competitors entered the industry during the period covered by the complaint — a clear indication that prices were above competitive levels. Containers were produced to customer specifications, but were substantially identical regardless of manufacturer. With a few exceptions, no manufacturer could command a higher price than another. From 1955 to 1963, defendants exchanged with varying frequency the prices most recently quoted to specific customers. No Robinson-Patman defense was raised, however.
The Supreme Court, 6-3, found a combination or conspiracy proscribed by the Sherman Act. The freedom to supply information only if each defendant so chose was described merely as the freedom to withdraw from the agreement. From the concentrated nature of the industry, the majority concluded that the information exchanges must have set a floor under prices. The Court distinguished Cement Manufacturers, supra, by pointing out that Container revealed no “controlling circumstance, viz., that cement manufacturers . exchanged price information as a means of protecting their legal rights from fraudulent inducements . . . .” Id. at 335, 89 S.Ct. at 511.
Mr. Justice Fortas concurred, declaring that he did not understand the majority to have established a per se ban of such exchanges. Instead, he concluded that the record supported a finding that the data dissemination had in fact affected prices. [123]*123Id. at 338-40, 89 S.Ct. 510 (Fortas, J., concurring). Justice Marshall, joined by Justices Harlan and Stewart, dissented. He agreed that exchanges of price information were not illegal per se, but he disagreed with Justice Fortas’ conclusion that the record showed any effect on prices. Id. at 340-47, 89 S.Ct. 510 (Marshall, J., dissenting).
Each of the Container opinions found an agreement in the reciprocal dissemination of price information. The discord arose over proof of the agreement’s effect. None of the Justices concerned himself with the agreement’s purpose. The Court’s careful distinction of Cement Manufacturers, however, suggests that proof of an actual purpose is not irrelevant in all cases. Where the information exchange occurs under a “controlling circumstance,” such as the purpose of preventing fraud in Cement Manufacturers, the exchange can be upheld under the Sherman Act, despite a proven or presumed effect on price.9
The question before us is whether appellants’ alleged desire to establish a defense, to price discrimination charges, of compliance with section 2(b) of the Robinson-Pat-man Act would create a similar “controlling circumstance” that would legitimate an agreement otherwise prohibited by section 1 of the Sherman Act. If it would, then the trial court’s instruction was improper in permitting the jury to convict appellants merely by finding an effect on prices, with no consideration of purpose. The Government denies that compliance with section 2(b) amounts to a “controlling circumstance.” Container, claims the Government, limited “controlling circumstances” to prevention of fraud.
We do not read that case so narrowly. If the Container Court had sought to restrict the exception to “prevention of fraud,” it could have distinguished Cement Manufacturers by noting simply that the Container case did not involve an attempt to thwart fraudulent purchasing practices, as had Cement Manufacturers. Instead, the Court used that much broader term “controlling circumstance,” one of which was the desire to prevent fraud present in the latter case. This suggests that there are “controlling circumstances” other than the one delineated in Cement Manufacturers.
Price verification, insist appellants, must be considered a controlling circumstance. They claim that otherwise the very activity dictated by the Robinson-Patman Act could be used as evidence of conspiracy to convict under the Sherman Act. This conflict between the two acts would leave them with [124]*124their three unappealing alternatives: (1) forego the price cut and probably lose a sale; (2) cut the price and risk a Robinson-Patman charge, if the buyer was lying; or (3) verify the price and risk the Sherman Act charge involved here.
The Government replies that nothing in the Robinson-Patman Act condones discussions about prices among competitors, nor has any court ever required such discussions as a condition of establishing a section 2(b) defense. Section 2(b), says the Government, does not force appellants to verify. The Government poses its alternatives for the appellants: (1) refrain from cutting prices to the single buyer; (2) cut prices as to all buyers. In neither event will the seller break either antitrust law.
Even the Government concedes that this choice is “difficult.” We fear that so limiting sellers’ alternatives would discourage price competition in industries like gypsum board, where the fact that price cutting occurs off list prices and invoices makes corroboration of reported competitor’s offers difficult or impossible. In such industries, the Government’s alternatives would eliminate the section 2(b) defense. This, it seems to us, would be more likely to put a floor under prices than to induce wholesale cutting, since a firm in a concentrated industry would be unlikely to reduce prices across the board simply to close a single sale. Report of the Attorney General’s National Committee to Study the Antitrust Laws 181 (1955). And it is through isolated price reductions that established price levels are eroded, precipitating widespread price cuts. P. Areeda, supra, at 231. The Government’s position, then, could induce greater price rigidity.
Because the Government’s resolution of the conflict between the Sherman and Robinson-Patman Acts would require us to read the Sherman Act as discouraging the very price competition it was designed to promote, we reject that resolution. Appellants, however, seek an accommodation of the two acts that would recognize any bona fide attempt to comply with section 2(b) of Robinson-Patman as a “controlling circumstance” for Sherman Act purposes. Anti-competitive dangers exist on this side, too. Such a resolution would countenance a broad range of data dissemination schemes and communications among competitors, which might provide camouflage for illegitimate agreements. With this problem in mind, we turn to the cases that have already addressed this issue.
Except for the court below, each court to consider the issue has agreed with appellants that the attempt to establish a 2(b) defense amounts to a “controlling circumstance” that shields those exchanging information from Sherman Act liability. The first and most apposite case involved a civil action alleging essentially the same conspiracy charged here. In Wall Products Co. v. National Gypsum Co., 326 F.Supp. 295 (N.D.Cal.1971), the court found that manufacturers of gypsum board, because their products were homogeneous, deemed it essential to meet lower offers by competitors. Purchasers aggressively played one manufacturer against another, whipsawing a series of price cuts through the industry by falsely reporting lower offers. When purchasers told sellers’ field representatives of competitors’ lower offers, the representatives asked for invoices or other written corroboration, but these requests usually proved fruitless. In order to insulate themselves against possible Robinson-Patman liability, specific company officials were given authority to communicate with the competitor in question. This communication took place “[only] after exhausting other means of confirming the truth of such reported deviation.” Id. at 309. Specific confirmation was obtained, and no reported offers were met unless verified. The Wall Products court held that this activity amounted to a “controlling circumstance” contemplated by Container. 326 F.Supp. at 312.
The three other cases that involved this issue shared similar facts. In each, a major oil company conferred with its competitors during a gasoline price war, in order to verify its competitors’ prices before granting temporary allowances to its own local retailer. Although it is not made clear in [125]*125Webster v. Sinclair Refining Co., 338 F.Supp. 248 (S.D.Ala.1971), whether the defendant communicated directly with competitors or merely observed publicly posted prices, the court must have thought that there was direct communication. It stated that obtaining information from competitors does not violate section 1 of the Sherman Act if the “controlling circumstance” of Robinson-Patman compliance is present, citing Wall Products, 338 F.Supp. 251-52.
In Belliston v. Texaco, Inc., 455 F.2d 175 (10th Cir.), cert. denied, 408 U.S. 928, 92 S.Ct. 2494, 33 L.Ed.2d 341 (1972), the competitors exchanged price information that was also publicly available through trade journals and financial dailies. Id. at 181. The court cited Wall Products, however, for the broad proposition that price information exchanges for the purpose of establishing a Robinson-Patman defense amount to a “controlling circumstance.” Id. at 182.
The third gasoline war case was Gray v. Shell Oil Co., 469 F.2d 742, 746-47 (9th Cir. 1972), cert. denied, 412 U.S. 943, 93 S.Ct. 2773, 37 L.Ed.2d 403 (1973). There the court held that the jury had to decide whether the purpose of Shell’s communications with its competitors was to comply with the Robinson-Patman Act. Because the jury had found that purpose, the court upheld a verdict of no liability, citing Belliston, supra.
First, it should be noted that the three gasoline war cases significantly expanded the Wall Products holding. None contained any suggestion that alternative means of corroboration, such as invoices or public trade papers, were unavailable, as they had been in Wall Products. Indeed, the Belli-ston court expressly found that they were available. None indicated that the gasoline retailers had achieved the same reputation for mendacity that the Wall Products court ascribed to wallboard buyers. The gasoline war cases, then, suggest that the category of price exchanges that qualify as “controlling circumstances” is quite broad. So free a license to exchange price information might undermine the Container holding.,
Second, these cases diluted the pro-competitive strength of the Sherman Act in order to preserve the force of Robinson-Pat-man’s requirement of solid corroboration. Yet Automatic Canteen Co. v. Federal Trade Commission, 346 U.S. 61, 73 S.Ct. 1017, 97 L.Ed. 1454 (1953), suggests that when policies of the Sherman and Robinson-Patman Acts conflict, it is the Robinson-Patman Act that should give way. In Automatic Canteen, a buyer was charged with violating section 2(f)10 of the Robinson-Patman Act, which forbids inducement of a discriminatory price. The question before the Court was which party — the FTC or the defendant — had the burden of producing evidence on the issue of the seller’s costs, information that might demonstrate that the varying prices were not in fact discriminatory. The Court refused to place the burden of discovering the seller’s costs upon the buyer because, among other things, “it would almost inevitably require a degree of cooperation between buyer and seller, as against other buyers, that may offend other antitrust policies,” id. at 69, 73 S.Ct. at 1022, that is, section 1 of the Sherman Act.
We are therefore reluctant to embrace the accommodation fashioned by the gasoline war cases, which seems to eviscerate the Sherman Act for the benefit of Robinson-Patman. The best way to resolve the dilemma framed by appellants in this case, in the light of Automatic Canteen, would be to ease their burden under Robinson-Pat-man, instead of facilitating evasions of the Sherman Act’s procompetitive strictures as did the gasoline war cases. If Robinson-Patman is interpreted to permit a seller lacking other corroboration to cut his price without taking the ultimate corroborative step of verifying the purchaser’s report with his competitor, then the Sherman Act [126]*126and Container retain more vigor. We think Wall Products, Viviano, and a recent case in the Sixth Circuit outline such a solution to appellants’ alleged dilemma.
In Kroger Co. v. Federal Trade Commission, 438 F.2d 1372 (6th Cir.) cert. denied, 404 U.S. 871, 92 S.Ct. 59, 30 L.Ed.2d 115 (1971), the Sixth Circuit affirmed an FTC finding that Kroger had induced a discrimination in price, in violation of section 2(f) of the Robinson-Patman Act. In soliciting bids for sale of private label milk to its chain of grocery stores, Kroger falsely told Beatrice Foods that it had received a bid twenty percent off list price from another dairy. The FTC found that Beatrice attempted to corroborate Kroger’s report, but did not contact its competitor directly. Beatrice Foods Co., 76 F.T.C. 719, 809-11 (1969). Relying on Kroger’s false report, Beatrice cut its price to Kroger. Both Beatrice and Kroger were charged with violating the Robinson-Patman Act. The FTC, however, held that Beatrice had established the “good faith meeting competition” defense under section 2(b). Despite this finding that the price was not illegal as to the seller, Kroger was held to have induced a discriminatory price. The FTC declared that Beatrice could not be held liable for the discriminatory effect of the price, because it had no reason to disbelieve Kroger. The Sixth Circuit affirmed the ruling that Kroger could nevertheless be held to have known that the price it induced was in fact discriminatory. Kroger, supra, 438 F.2d at 1376-77.
The Kroger case, when combined with the Wall Products decision and our Viviano holding, offers a solution to the quandary in which appellants claim to find themselves. Kroger indicates that sellers can establish a section 2(b) defense without taking the ultimate corroborative step of communicating directly with their competitors. It accomplishes this by holding the purchaser to a high standard of good faith bargaining under section 2(f). Thus, as long as the seller has taken all other reasonable steps in an attempt to corroborate his purchaser’s report of lower offers, he need not fear Robinson-Patman liability, provided that — as in Kroger — he has no independent reason to doubt the purchaser’s reliability.
Viviano, though, requires the seller to investigate that reliability. Vivano, supra, 411 F.2d at 259. This holding protects the integrity of the section 2(b) defense. If the seller could discriminate in price when he had good reason to believe that the buyer — even a buyer held to a high standard by Kroger — was lying about a competitor’s supposedly lower offer, then the “good faith” requirement of section 2(b) would be rendered nugatory. Therefore, if previous experience or the investigation required by Viviano reveals that the purchaser cannot be trusted, as was the case in Wall Products, supra, 326 F.Supp. at 309, then, and only then, the seller may communicate with his competitor to test the buyer’s veracity. The presence of this doubt as to the purchaser’s truthfulness, combined with the seller’s inability to corroborate the buyer’s reports independently, creates the “controlling circumstance” that exempts price information exchanges from the Sherman Act ban, under Container. This is a narrow “controlling circumstance,” not the broad one of the gasoline war cases.
There was some evidence of this narrow “controlling circumstance” in this case. Several witnesses testified that the purpose of the “verification” calls was to establish a Robinson-Patman defense and that their scope was confined by that purpose. There was also testimony concerning the unreliability of purchasers in the gypsum board market. Therefore, appellants were entitled to an instruction that their verification practice would not violate the Sherman Act if the jury found: (1) the appellants engaged in the practice solely to comply with the strictures of Robinson-Patman; (2) they had first resorted to all other reasonable means of corroboration, without success; (3) they had good, independent reason to doubt the buyers’ truthfulness; and (4) their communication with competitors was strictly limited to the one price and one buyer at issue. Trial court charged instead that if the jury found that verification had [127]*127an effect on prices, they had to find guilt under the Sherman Act. Because this charge improperly ignored Container’s exception for price information exchanges carried out under a “controlling circumstance,” reversal is required.
B.
Appellants also contend that, on the state of the evidence, they were entitled to a more precise charge on the purpose and scope of the conspiracy any particular defendant was alleged to have joined. The Government insists that the trial court’s charge on this point was sufficient and that appellants waived their objections in any event. I agree with appellants that the court’s charge was deficient. Finding no waiver, I would order reversal on this point.11
Over a period of five months, the Government introduced masses of evidence concerning a multitude of seemingly unrelated events and attempted to link them together to show a single, overall conspiracy to stabilize the prices and marketing conditions of gypsum wallboard. Most of this evidence dealt with actions prior to the start of the applicable limitations period. In these circumstances, it was necessary specifically to inform the jury that any particular defendant could be found guilty of conspiracy only if he understood that he had joined the single, overall agreement charged in the indictment; participation in any of the subsidiary events was not enough. United States v. Peoni, 100 F.2d 401, 403 (2d Cir. 1938); accord, United States v. Purin, 486 F.2d 1363, 1369 (2d Cir. 1973), cert. denied sub nom. Da Silva v. United States, 416 U.S. 987, 94 S.Ct. 2392, 40 L.Ed.2d 764 (separate appeal), cert. denied sub nom. Purin v. United States, 417 U.S. 930, 94 S.Ct. 2640, 41 L.Ed.2d 233 (1974) (separate appeal). The great number of activities in evidence and the huge cast of characters involved made it essential to impress upon the jury that each defendant could be found guilty of the conspiracy charged only if that overall design was within his reasonable contemplation when he engaged in any of the isolated activities. Specifically, the jury should have been asked to decide whether a given defendant, by engaging in one or more of the apparently isolated events, was associating himself with the overall agreement charged in the indictment. United States v. Borelli, 336 F.2d 376, 385 (2d Cir. 1964), citing Peoni, supra, 100 F.2d at 403. This the trial court did not do.
Trial court did — as did the trial court in Borelli, supra — repeatedly instruct the jury that each defendant could be found guilty only if he knowingly joined the conspiracy charged and that guilt was to be determined on an individual basis.12 But this [128]*128left the jury — like the jury in Borelli — free to find that a defendant automatically associated himself with the overall conspiracy charged merely by participating in one or a few of the acts alleged to be individual elements of the grand design. In this sense, the findings of guilt may have been “all-or-none”: once an overall conspiracy was found as to some defendants, all of the defendants could have been connected to the conspiracy by their involvement in one of the minor agreements allegedly part of it. The court failed to emphasize that joining a subsidiary agreement would not constitute guilt unless it was done with the intent to join the overall conspiracy. This “all-or-none” basis for determining guilt is what bothered the Borelli and Peoni courts.
The Borelli court described the problem:
The view that if the evidence warrants the finding that some defendants were parties to a single agreement to sell contraband for a nine-year period, it necessarily does so as to every defendant who has conspired with them at any time for any purpose, is thus a considerable oversimplification. Rinaldo’s testimony thoroughly warranted the finding of an agreement among himself, Joe and Rosario Magavero beginning in 1950 and continuing through 1959. But it does not follow, for example, that his recital of two deliveries to Tantillo in March and July, 1951, on the instructions of Locascio in one instance, and of Rosario Mogavero in the other, would justify a finding that Tantillo had thereby made himself a part of whatever narcotics enterprises Locas-cio, the Mogaveros and their future partners might engage in for the rest of their lives, provided only that these were uninterrupted — a theorem that would lead to the inevitable but extraordinary conclusion that Tantillo’s two purchases entered him into a “partnership in criminal purposes,” United States v. Kissel, 218 U.S. 601, 608, 31 S.Ct. 124, 54 L.Ed. 1168 (1910), with Castiglia, who supplied a source to and took deliveries from an altered core seven years later, and vice versa.
Borelli, supra, 336 F.2d at 384. Appellants here raise a similar concern. For example, it does not follow that appellant Brown’s participation in the allegedly conspiratorial 1965 industry-wide price increase and 1966 change in credit terms — the only actions with which any evidence connected him— means that he associated himself with an overall conspiracy to stabilize prices beginning before 1960 and lasting until 1973. Each of the other appellants was likewise connected by the evidence only to a few of the dozens of allegedly conspiratorial episodes that the Government sought to weave into one overriding agreement. Yet the jury was never told that participation in one episode, without more, was insufficient to prove association in the latter enterprise.
In view of the ambiguity surrounding the scope of any agreement joined by each of the defendants, the trial court should have given the charge they requested on this point.13 Conspiracy is “an elastic, sprawl[129]*129ing and pervasive offense.” Krulewitch v. United States, 336 U.S. 440, 445, 69 S.Ct. 716, 719, 93 L.Ed. 790 (1949) (Jackson, J., concurring). Trial courts must take great pains to prevent the seine of a conspiracy charge from entangling a defendant only tangentially related to the agreement in question.
The Government’s assertion that appellants waived any objection to the trial court’s charge on this point is without merit. Defense counsel three times expressed their willingness to make a record by “particularizing” objections to each charge the trial court denied. On each occasion, the court observed that particular objections would consume too much time and that one blanket objection sufficed to make a record for all. Where the trial judge declares that he does not require further elucidation of the requests he has refused, there can be no waiver. See Green v. Reading Co., 183 F.2d 716, 719 (3d Cir. 1950); 3 C. Wright, Federal Practice and Procedure § 842 (1969). Hence, the lack of a specific objection to the refusal to give this specific charge does not bar appellants from raising it as error on appeal.
Contrary to the Government’s claim, appellants submitted their requested charge in a timely manner. After telling counsel on June 27, 1975, what the substance of his charge would be, the trial judge invited written objections by July 2. On June 30, the court advanced that deadline one day, so that he could rule on the objections before closing arguments began. On July 1, appellants filed numerous objections and supplemental requests designed to meet the court’s previous comments. Among them was the charge at issue here. Because of time constraints, the trial court did not then rule on the lengthy submissions. He indicated, however, that he would refer to them in formulating his charge.14 Because the court acknowledged on the record that appellants had timely called his attention to the points they wanted charged, the Government’s assertion of waiver must fail.
C.
Appellants also question the sufficiency of the trial court’s charge with respect to withdrawal from the alleged conspiracy. Where a continuing conspiracy is proven, each defendant who by affirmative action has completely disassociated himself from the agreement before the start of the limitations period is entitled to acquittal. Hyde v. United States, 225 U.S. 347, 369, 32 S.Ct. 793, 56 L.Ed. 1114 (1911). This abandonment may be effected by communicating it “in a manner reasonably calculated to reach co-conspirators.” Borelli, supra, 336 F.2d at 388. Proof of such communication need not involve evidence that the defendant directly informed each co-conspirator of his intent to withdraw.
The formation of a criminal conspiracy or adherence to a criminal scheme may be shown by evidence either direct or circumstantial, in practice often taking a wide range, and we see no reason why the withdrawal should not be held susceptible to proof of the same character. It might, of course, be shown by a writing, or by an express oral agreement, and we think by conduct wholly inconsistent with the theory of continuing adherence.
Buhler v. United States, 33 F.2d 382, 384 (9th Cir. 1929); accord, Marino v. United States, 91 F.2d 691, 698 (9th Cir. 1937), cert. [130]*130denied sub nom. Gullo v. United States, 302 U.S. 764, 58 S.Ct. 410, 82 L.Ed. 593 (1938). See also United States v. Goldberg, 401 F.2d 644, 648-49 (2d Cir. 1968), cert. denied, 393 U.S. 1099, 89 S.Ct. 895, 21 L.Ed.2d 790 (separate appeal), cert. denied sub nom. Tannenbaum v. United States, 394 U.S. 932, 89 S.Ct. 1202, 22 L.Ed. 461 (1969) (separate appeal); Developments in the Law — Criminal Conspiracy, 72 Harv.L.Rev. 920, 958 (1959).
Appellants introduced evidence of the resurgence of price competition, and other forms of competition, which could have convinced the jury that one or more of the defendants had abandoned any conspiracy to stabilize prices and other conditions before the start of the limitations period. They also requested a charge to the effect that active competition on price might amount to the “affirmative action” necessary to accomplish a withdrawal.15 Trial court’s charge, however, negated this theory of withdrawal. The court charged as follows:
In order to find that a defendant abandoned or withdrew from a conspiracy pri- or to December 27, 1968, you must find from the evidence, that he or it took some affirmative action to disavow or defeat its purpose. Mere inaction would not be enough to demonstrate abandonment. To withdraw, a defendant either must have affirmatively notified each other member of the conspiracy he will no longer participate in the undertaking so they understand they can no longer expect his participation or acquiescence, or he must make disclosures of the illegal scheme to law enforcement officials.
Thus, once a defendant is shown to have joined a conspiracy, in order for you to find he abandoned the conspiracy, the evidence must show that the defendant took some definite, decisive step, indicating a complete disassociation from the unlawful enterprise.
(Emphasis added).
As the court explained the meaning of “withdraw,” it comprised only affirmative notification — which connotes direct announcement — and disclosure to law enforcement authorities. Either might constitute withdrawal, but the court, by limiting withdrawal to these two, improperly excluded other possibilities, such as conduct inconsistent with the theory of continued adherence.
The court did permit appellants to argue their theory of withdrawal to the jury. That opportunity did not salvage appellants’ rights, however, because the court in its subsequent charge told the jury that they could not consider appellants’ theory. Reversal is, in my opinion, required on this point.16
IV.
The judgments of conviction will be reversed and the case remanded to the court below.