OPINION OF THE COURT
GARTH, Circuit Judge.
The O. Hommel Company brought suit against the Ferro Corporation, claiming that Ferro had violated § 2 of the Sherman Act, 15 U.S.C. § 2 (1976) and § 2 of the Clayton Act as amended by the RobinsonPatman Act, 15 U.S.C. § 13 (1976). Between 1973 and 1977, Ferro sold porcelain enamel frit to three customers at a price lower than the price it charged other customers, and below its average total cost. The district court refused to enter a directed verdict for Ferro, and the jury found that Ferro had violated the Robinson-Pat-man Act, 15 U.S.C. § 13, but not the Sherman Act. Ferro’s subsequent motion for judgment notwithstanding the verdict on the Robinson-Patman count was denied.
We hold that there was insufficient evidence upon which to predicate RobinsonPatman liability, and thus we reverse and direct that judgment be entered for Ferro.
I.
A.
Both Ferro Corporation and 0. Hommel Corporation (Hommel) manufacture and sell two kinds of frit. Porcelain enamel frit is the basic ingredient in the coating of such products as bathtubs, stoves, washing machines and refrigerators. These frits:
are sold by plaintiff and defendant in conjunction, at times, with other frits and always with “additions” or “additives” of clay, electrolytes and other substances necessary to produce a porcelain enamel coating with the buyer’s desired properties. The frits and additions or additives are assembled by the customer on a predetermined formula in what is called a “batch” which is milled or ground with water to produce a porcelain enamel “slip”. The slip is applied to the steel appliance part by spraying, dipping or flow coating. It is then dried and fused by heat which produces the porcelain enamel finish.
I App. at 55-56.
The companies also manufacture ceramic frit which is the basic ingredient in the glaze coating on products such as dishes. The jury’s verdict on Robinson-Patman discrimination, however, was based only on sales of the porcelain enamel frit.1 There is no major difference in the method of manufacturing the two types of frit (II App. at 413), and “generally speaking” the Ferro and Hommel production methods were the same. (II App. at 416).
Hommel has one plant in Carnegie, Pennsylvania. Frit (both porcelain enamel and ceramic) is its principal product, constituting 53% to 68% of sales. Its net worth is about $2.5 million, and its annual sales during the 1973 to 1977 period generally were in the range of $7,000,000. Hommel’s income before taxes increased from $12,902 in 1973 to $256,615 in 1977. (IV App. at 212-16). Its gross profits on the sale of porcelain enamel frit were approximately $475,-000 in 1973 and $540,000 in 1977, although when selling, laboratory, and general and administrative expenses were allocated, it showed net losses each year. (IV App. at 217-21). Its net loss, however, on these frit sales was $19,000 in 1977 compared to $259,-000 in 1973. Id.
Ferro is a multinational corporation with several divisions. Its net worth is about $150,000,000 and during the applicable period its sales averaged $334,000,000. Ferro [343]*343manufactures frit in the United States at plants in Los Angeles, Nashville, and Cleveland. . The alleged illegal sales all concerned frit manufacture at the latter two plants.
There are five manufacturer-sellers of porcelain enamel frit. Ferro is the largest producer and Hommel is one of the two smallest. Chicago Vitreous and Perneo, the next two largest manufacturers after Ferro, are both divisions of multinational conglomerates larger than Ferro. During the 1973-77 period, the years during which the alleged violations occurred, the sales of porcelain enamel frit (in thousands of pounds) were as follows:
Industry Ferro Chicago Vitreous
1973 180,188 76,365 41,064
1974 144,243 61,995 32,905
1975 111,484 46,507 27,804
1976 125,936 51,067 30,915
1977 134,397 54,200 31,921
Pomco Hommel IngrahamRichardson
1973 38,086 12,247 12,426
1974 31,628 9,278 8,437
1975 24,816 6,480 5,877
1976 30,907 7,615 5,432
1977 32,662 9,162 6,452
These figures resulted in the following market shares for sales of porcelain enamel frit:
Ferro Chicago Vitreous Perneo Hommel IngrahamRichardson
1973 42% 23% 21% 7% 7%
1974 43 23 22 6 6
1975 42 25 22 6 5
1976 41 25 25 6 4
1977 40 24 24 7 5
These figures do not include production by four concerns that produce porcelain enamel frit for their own production needs. (Ill App. at 617).
As can be seen from the production figures, during the 1973 to 1975 period the frit industry went into a steep decline from which it only partially recovered. The 1977 porcelain enamel frit production was less than 75% of the 1973 production. Ferro suffered a 29% loss of sales during the five year period, and its market share fell from 42% to 40%. Hommel had a 25% decline in sales during that same period, and a 7% market share in both 1973 and 1977. Only part of the decline in sales can be attributed to the 1974 recession. The porcelain enamel frit industry was facing increasing competition from other products. Paint and plastic products were being substituted for porcelain enamel frit. An increase in the imports of products to which frit is applied, also hurt the demand for domestic frit.
Prices for frit generally rose during the 1973 to 1977 period. (See, e. g., IV App. at 393). All five of the firms had published price lists, but neither Hommel nor Ferro put all their price quotations on their published lists. (II App. at 341). Ferro also had a confidential, “special price list” for porcelain enamel frit. (IV App. at 15-22, Joint Exhibits 10-13). Prices varied with the type of porcelain enamel frit sold.
B.
Robinson-Patman liability was based on sales to three customers: General Housewares, Briggs, and General Electric. Porcelain enamel frit was sold to these customers below the published list price and below Ferro’s average total cost (III App. at 484-488),2 but above Ferro’s average variable cost (which Ferro approximates as its direct manufacturing cost).3 (Joint Exhibits 718, 719, IV App. at 392d-394).
The discriminatory sales made to General Housewares, Briggs and General Electric constituted no more than 5% of Ferro’s national porcelain enamel frit sales during this 1973-1977 period, (Ferro brief at 7), and 2% of the domestic sales made by all frit companies during this period. Id.
1. General Housewares
General Housewares manufactures pots and pans. In 1966 Ferro convinced General Housewares to discontinue being a self-smelter of frit. Between 1973 and 1977 Ferro’s frit sales to General Housewares declined about 40%, from 5,660,000 pounds to 3,650,000 pounds. As Ferro’s marketing [344]*344manager explained, in order to sell to a firm that was a self-smelter, as General Housewares had been, prices had to be very close to direct manufacturing costs. Although General Housewares had discontinued its production of frit, it could reestablish its smelting operation for less than $500,000. (Ill App. at 618). General Housewares was also very resistant to high frit prices because the domestic pot and pan business faced “tremendous foreign competition.” (Ill App. at 617). Several American companies in the field — including at least one former Ferro customer — had gone out of business. Hommel had never sold any frit to General Housewares.
2. Briggs
Briggs, a bathtub manufacturer, had been a Ferro customer for many years. In 1974, Briggs moved its plant and made Hommel its frit supplier because Hommel frit could be applied at a lower, cost-saving temperature. Quality problems developed at the new Briggs plant — the enamel began to flake off the steel. The president of Briggs testified that in searching for a solution to this problem, he decided to purchase frit again from Ferro, although he did not discontinue purchasing the groundcoat from the same supplier. Ferro thereafter supplied half of Briggs’s frit requirements. The frit sold to Briggs was identical to other frit sold by Ferro to other customers under a different label and at a higher price. It was later determined that the steel in the Briggs tubs was faulty, and Briggs considered resuming the purchase of frit from Hommel. Briggs demanded from Hommel a price competitive with Ferro’s price, price protection for four, six or more months, and mentioned “something special” given to Briggs by Ferro — which turned out to be yearly price protection, assured by Ferro on a year-to-year basis. Hommel never offered any price protection and did not sell to Briggs after 1974.
3. General Electric
In April, 1976 Ferro extended a 5% discount to General Electric on certain frits if General Electric would buy 6,000,000 pounds of frit during 1976. (IV App. at 362-364). General Electric annually purchased approximately 15,000,000 pounds of frit and had obtained 4,500,000 pounds from Ferro in 1975. General Electric only ordered 3,800,000 pounds of frit in 1976.. Ferro unsuccessfully sought to rescind the 1976 discount and discontinued the discount for 1977 sales. Hommel sold no frit to General Electric in 1976 and 1977. Between 1973 and 1975 Hommel’s sales to General Electric had averaged about 200,000 pounds a year. Hommel’s loss of sales was due in large part, if not totally, to a change in coating materials made by two General Electric divisions. A Hommel employee did testify that he attempted to sell frits to General Electric but General Electric refused because of its arrangement with Ferro.
C.
Hommel charged Ferro with attempting to monopolize the frit industry in violation of § 2 of the Sherman Act, 15 U.S.C. § 2 (1976), and with price discrimination in violation of § 2(a) of the Robinson-Patman Act, 15 U.S.C. § 13(a) (1976). On June 21, 1979 the district court denied Ferro’s motion for summary judgment, accepting Hommel’s argument that sales below total cost in some circumstances could permit an inference of predatory intent. O. Hommel v. Ferro Corp., 472 F.Supp. 793 (W.D.Pa.1979). The court relied on Utah Pie v. Continental Baking Co., 386 U.S. 685, 87 S.Ct. 1326, 18 L.Ed.2d 406 (1967) in arriving at its decision. Citing Zenith Radio Corp. v. Matsushita Electric Indus. Co., 402 F.Supp. 244 (E.D.Pa.), aff’d mem., 521 F.2d 1399 (3d Cir. 1975), it also ruled that Hommel could not introduce evidence demonstrating that Ferro had subsidized low frit prices in the United States through foreign sales.
The theory upon which the district court proceeded was explained in its charge to the jury:
Where one seller sells below cost at a price less than direct cost plus an allocation for overhead, and a competing seller who is undercut is injured, the effect upon competition may be found to be traceable to such discrimination.
[345]*345You should, of course, in this connection consider the instructions I have given you previously as to a declining industry.
(Ill App. at 920-21). The judge’s previous instructions, given on the § 2 Sherman Act count, were:
Now, I charge you that, where there is competition among members of a relatively stable industry, particularly where one of the members manufactures only the product in question, in this case, frit, porcelain or glaze, each manufacturer should charge prices which, over a period of, say, four years, as in this case, should equal the total cost; and, if a price less than the total cost has been effected or charged by the manufacturer, then, such pricing may amount to predatory pricing; that is, pricing intended to injure a competitor. Of course, to continue to charge less than total cost in this period of time could force one or more of the competitors in the frit industry into bankruptcy. You must make all your costs or you are going to go broke.
However, were you to find that the market for frit was declining during the years 1973 to 1977, and that the frit industry was a declining industry, and there was a large amount of unused capacity, then, the prices charged by manufacturers need only equal the average variable costs; that is, the direct manufacturing costs.
(Ill App. at 912-913).4
The jury found Ferro liable under the Robinson-Patman Act, but not under the Sherman Act. The jury calculated the damages to be $1,500,000, which the district court trebled to $4,500,000.5
On June 16, 1980, the district court, in a supplementary judgment order, awarded $483,350.48 in costs and attorney’s fees to Hommel. Within ten days Ferro filed a motion for a judgment notwithstanding the verdict (it had moved for a directed verdict at the close of all the evidence,. Ill App. at 802), and a motion to alter the judgment so as to strike all damages. Ferro also filed a protective notice of appeal from the supplementary judgment order.
This court remanded the case for determination of all pending motions, jurisdiction of the appeal being retained. The district court found the post trial motions untimely and in the alternative denied them on the merits. It held that predatory intent could be found in Ferro’s price discrimination and in Ferro’s below total cost pricing. The present appeal resulted.
II.
In this case we are concerned with price discrimination at the primary or seller level. The Supreme Court wrote in FTC v. Anheuser-Busch, Inc., 363 U.S. 536, 546, 80 S.Ct. 1267, 1272, 4 L.Ed.2d 1385 (1960) that “an independent and important goal of § 2(a) is to extend protection to competitors of the discriminating seller.”
In relevant part, § 2(a) of the Clayton Act, as amended by the Robinson-Patman Act, 15 U.S.C. § 13(a), reads:
(a) It shall be unlawful for any person engaged in commerce, in the course of such commerce, either directly or indirectly, to discriminate in price between different purchasers of commodities of like grade and quality, where either or any of the purchases involved in such discrimination are in commerce, where such commodities are sold for use, consumption, or resale within the United States or any Territory thereof or the District of Columbia or any insular possession or other place under the jurisdiction of the United States, and where the effect of such discrimination may be substantially to lessen competition or tend to [346]*346create a monopoly in any line of commerce . .. .6
(emphasis added).
The price discrimination referred to in Section 2(a) is merely a price difference. Anheuser-Busch, 363 U.S. at 549, 80 S.Ct. at 1274. A difference in prices charged by a seller to different customers, for the same product, constitutes price discrimination.
Price discrimination is not illegal per se. At the primary level, the plaintiff must show that the “effect of such discrimination may be substantially to lessen competition or tend to create a monopoly in any line of commerce . . ..” See Anheuser-Busch, 363 U.S. at 542-43, 553, 80 S.Ct. at 1270-1271, 1276; 4 Van. Kalinowski, Antitrust Laws and Trade Regulation §§ 28.08, 29-01[4] (1980); Utah Pie Co. v. Continental Baking Co., 386 U.S. 685, 87 S.Ct. 1326, 18 L.Ed.2d 406 (1967); Dean Milk Co. v. FTC, 395 F.2d 696, 700 (7th Cir. 1968).7
In defining the scope of liability under the Robinson-Patman Act, the larger goals of antitrust law must be considered. The Supreme Court has cautioned us “that the Robinson-Patman Act should be construed consistently with broader policies of the antitrust laws.” Great Atlantic & Pacific Tea Co. v. FTC, 440 U.S. 69, 80 n.13, 99 S.Ct. [347]*347925, 938, 59 L.Ed.2d 153 (1979). A court must be wary in imposing liability that would result in “price uniformity and rigidity.” Id. at 80, 99 S.Ct. at 933; Automatic Canteen Co. v. FTC, 346 U.S. 61, 63, 73 S.Ct. 1017, 1019, 97 L.Ed. 1454 (1953). Secret price discrimination may be the only effective means for a firm to undercut published oligopolistic prices. See ABA Antitrust Section, Monograph No. 4, The Robinson-Patman Act: Policy and Law, Vol. I (1980) at 29. The Robinson-Patman Act is not to be used as a means to maintain artificially high prices. In defining the scope of Robinson-Patman liability, it must be remembered that vigorous price competition is a central goal of the antitrust laws. See International Air Industries, Inc. v. American Excelsior Co., 517 F.2d 714 (5th Cir. 1975) cert. denied, 424 U.S. 943, 96 S.Ct. 1411, 47 L.Ed.2d 349 (1976). “A focus on detrimental effects on competition, rather than a concern with individual competitors, is fundamental to a reconciliation of the Robinson-Patman Act with overall antitrust policies.” F. Rowe, PRICE DISCRIMINATION UNDER THE ROBINSON-PAT-MAN ACT 130 (1962) (emphasis added).
There are two basic means of meeting the competitive injury requirement: (1) actual competitive injury shown by market analysis; and (2) predatory intent from which competitive injury may be inferred.
It is not disputed in this case that Ferro sold porcelain enamel frit to three customers below the prices at which it sold that frit to other customers. The central issue is whether these price differentials substantially lessened competition or tended to create a monopoly.
There was clearly no actual competitive damage in this case. The market share of Ferro dropped, while Hommel retained its market share. The market shares of Chicago-Vitreous and Perneo, both increased. There was thus no evidence of any trend to monopolization — nor was there evidence of a less competitive industrial structure. There was no collapsing price structure, a condition found to constitute competitive harm in Utah Pie v. Continental Baking Co., 386 U.S. 685, 87 S.Ct. 1326, 18 L.Ed.2d 406 (1967). The speculative possibility that Hommel may have increased its sales had Ferro not engaged in price discrimination, cannot constitute harm to competition within the meaning of § 2(a). Even the most generous reading of the full record fails to disclose any competitive harm which Hommel suffered (its market share remaining stable, and its income and gross profits on the sale of porcelain enamel frit having increased between 1973 and 1977), much less “detrimental effects on competition.”
Since there is no actual competitive harm that has been demonstrated in this case, our analysis must focus on whether there was evidence of Ferro’s predatory intent from which competitive harm can be inferred. Predatory intent can be shown by two means: (a) by express evidence; and (b) by inference from “below-cost” pricing. This second means of proving predatory intent is known as the “double-inference” test. “From below-cost pricing, predatory intent is inferred; from the finding of predation, injury to competition is inferred.” Pacific Engineering & Prod. Co. v. Kerr-McGee Corp., 551 F.2d 790, 798 (10th Cir.), cert. denied, 434 U.S. 977, 98 S.Ct. 543, 54 L.Ed.2d 472 (1977).
Initially, we must define predatory intent. The legislative history of the Clayton Act indicates that predatory intent is an intent to destroy a rival with the ultimate purpose of acquiring a monopoly in a particular locality. See H.R.Rep.No. 627, 63d Cong., 2d Sess. 8 (1914); S.Rep.No. 698, 63d Cong., 2d Sess. 2-4 (1914).8
[348]*348More recent definitions of predatory intent focus on sacrificing present revenues with the purpose of achieving monopoly profits in the future. The Ninth Circuit has defined predatory pricing as pricing that indicates that a defendant was “foregoing present profits in order to create a market position in which it could charge enough to obtain supranormal profits and recoup its present losses.” Hanson v. Shell Oil Co., 541 F.2d 1352, 1358 (9th Cir. 1976), cert. denied, 429 U.S. 1074, 97 S.Ct. 813, 50 L.Ed.2d 792 (1977). See also Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848, 856 (9th Cir. 1977), cert. denied, 439 U.S. 829, 99 S.Ct. 103, 58 L.Ed.2d 122 (1978); Areeda and Turner, Predatory Pricing and Belated Practices Under Section 2 of the Sherman Act, 88 Harv.L.Rev. 697, 698 (1975). The concept of predation under the Robinson-Patman Act does not differ from the Sherman Act concept of predation.9 See generally Areeda and Turner, 88 Harv.L.Rev. at 727.
Hommel claims that there was sufficient evidence of predatory intent upon which a jury could base liability. It points to what it considers to be direct evidence of predatory intent and asserts that the double inference test was correctly employed here — from the pricing below average total cost, the jury could infer predatory intent. We conclude, however, that there is no express evidence of predatory intent and that the use of any double inference test, on the facts of this case, was improper.
To be relevant, express evidence of predatory intent should demonstrate the alleged predator’s intention of sacrificing present revenues with the hope of obtaining monopoly profits. Hommel’s direct evidence of predatory intent stressed the secrecy of sales made by Ferro to Briggs, General Housewares, and General Electric. The prices of the frit that Ferro sold to these concerns were not published. Hommel emphasizes that identical frit was placed in different numbered bags and sold at different prices to other customers. This secrecy and possible deception, Hommel asserts, is evidence of predatory intent.
The secrecy here, however, cannot be viewed as evidence of predation. Keeping the “below-cost” sales to these three customers secret does not indicate that Ferro intended to drive Hommel out of business with the hope of obtaining monopoly profits. By keeping its prices secret, Ferro permitted Hommel to continue selling to Hommel’s own customers at Hommel’s higher prices. If Ferro had made known its lower prices, then Hommel and the other frit manufacturers would have had to meet those prices or risk losing customers.10 Such conduct by Ferro would be more likely to injure competition than did the price discrimination here. (See infra for a discussion of across-the-board versus isolated, non-geographic price discrimination).
Moreover, the price discriminations here could not be accomplished without secrecy. [349]*349As Hommel admits, it is difficult to offer a product at a lower price to some favored customers without offering the same lower prices to other customers, or risking the loss of their business, when the less-favored customers learn of the discounts. Additionally, as mentioned earlier, secret discount pricing may be the only effective means to undercut an oligopolistic price. We thus fail to understand why price secrecy is indicative of predatory intent, nor has Hommel called our attention to any cases that would explain that relationship.
Hommel also considers it highly significant that an outside economic expert witness for Ferro testified that because of the continuing excess capacity in the frit industry, some frit firm might be forced to halt production in the future. The expert was of the opinion that if demand did not increase, it would be the most efficient use of economic resources for one of the firms to leave the frit business. Not only do we find it difficult to accept the testimony by a non-Ferro employee at trial in 1979, as revealing Ferro’s intent in setting prices in 1973-1977, but we regard that testimony as bearing no relationship to predatory intent. Nor are we persuaded that any of the other evidence presented by Hommel, constituted proof of predatory intent.
Having determined that no express evidence of predatory intent is revealed by this record, we turn next to Hommel’s argument that Ferro by pricing below its average cost11 has exhibited the predatory intent which § 2(a) proscribes. Hommel, relying on Utah Pie Co. v. Continental Baking Co., 386 U.S. 685, 87 S.Ct. 1326, 18 L.Ed.2d 406 (1967) and Williamson, Predatory Pricing: A Strategic and Welfare Analysis, 87 Yale L.J. 284, 321-22 (1977), asserts that from the below-average cost pricing in this case, competitive harm can be inferred. Ferro counters, that since there is no actual competitive damage, none can be inferred, and in the alternative, that competitive harm can only be inferred from prices below marginal cost or average variable cost.12 We agree with Ferro that on the [350]*350facts of this case, the jury should not have been permitted to infer predatory intent from the evidence of Ferro’s “below-cost” pricing.
The main evil at which § 2 of the Clayton Act was aimed was geographic price discrimination. Congress feared that nationwide concerns would lower costs in one region with the intent of destroying local competitors thereby hoping to acquire a monopoly in that locality. See S.Rep.No. 698, 63d Cong., 2d Sess. 2-4 (1914); H.R. Rep.No. 627, 63d Cong., 2d Sess. 8 (1914). The House Report stated:
Section 2 of the bill ... is expressly designed with the view of correcting and forbidding a common and widespread unfair trade practice whereby certain great corporations and also certain smaller concerns which seek to secure a monopoly in trade and commerce by aping the methods of the great corporations, have heretofore endeavored to destroy competition and render unprofitable the business of competitors by selling their goods, wares, and merchandise at a less price in the particular communities where their rivals are engaged in business than at other places throughout the country. ... In the past it has been a most common practice of great and powerful combinations engaged in commerce — notably the Standard Oil Co., and the American Tobacco Co., and others of less notoriety, but of great influence — to lower prices of their commodities, oftentimes below the cost of production in certain communities and sections where they had competition, with the intent to destroy and make unprofitable the business of their competitors, and with the ultimate purpose in view of thereby acquiring a monopoly in the particular locality or section in which the discriminating price is made....
Id. The Senate bill expressed similar concerns. Congress believed that price discrimination in one locality would be financed by higher prices elsewhere. “Every concern that engages in this evil practice must of necessity recoup its losses in the particular communities or sections where their commodities are sold below cost or without a fair profit by raising the price of the same class of commodities above their fair market value in other sections or communities.” S.Rep. at 3.13 Whatever the merits of this recoupment theory, it is clear that it only applies in a geographic price discrimination setting. The theory presupposes that there is a geographic market for the commodity where the concern engaging in such “evil” practices faces no price competition.
The implicit congressional belief that non-geographic price discrimination does not present the same threat of economic harm at the primary level is shared by current commentators. Areeda and Turner would not even prohibit selective price-cutting:
2. Price Discrimination in the Same
Geographic Market.—
If, as we have contended, a monopolist should be permitted to make a general price reduction so long as his price equals or exceeds marginal cost, we are unable to see a persuasive case for prohibiting selective price-cutting to retain or obtain particular customers. Selective price-cutting cannot possibly be more harmful to small competitors than a general price reduction to the same level.54 And since any additional sale at a price at or above marginal cost does not decrease shortrun net returns, the necessary element of predation is missing.
Areeda & Turner, 88 Harv.L.Rev. at 725-26. See also Van Kalinowski, supra at § 29.04, 29-94-95; Note, Competitive Injury [351]*351Under the Robinson-Patman Act, 74 Harv. L.Rev. 1597, 1610-11 (1961).
The theoretical evil that lies behind predatory pricing is that a large firm will engage in the illegal price-cutting tactics which Congress feared, thereby forcing smaller concerns also to cut their price, or risk losing all their customers to the initial price cutter.14 If a price-cutter offers a lower price to only a few set number of firms, then its price schedule cannot attract customers from its competitors. Those competitors will be able to retain their customers, because their customers will not be able to receive the discount price given by the price-cutter. Thus, the competitors will not be forced to lower their prices to meet the price-cutter. This type of price discrimination would be difficult to maintain without secrecy because otherwise disfavored customers of the price-cutter might demand equal treatment. This secrecy, as explained earlier, helps competitors to retain their customers without meeting the price-cutter’s price. Unless a customer knows of the lower predatory price, and can buy at that price, price discrimination will not likely have an adverse effect on competition.
Moreover, an additional sale at above marginal cost will increase short run net returns. A seller faced with a choice of making a sale at above marginal cost but below total cost, or foregoing the sale, will choose to make the sale. Such a profit maximizing sale cannot be indicative of predatory intent. We are thus impelled to conclude that in a non-geographie selective discriminatory pricing case when actual competitive harm has not been shown, Robinson-Patman primary line liability cannot be imposed merely upon evidence of selective below-average cost pricing.
Utah Pie is entirely consistent with our analysis. In Utah Pie, national manufacturers of frozen dessert pies challenged the price of a small (18 employees, family operated), local dessert pie manufacturer, Utah Pie, by lowering prices in the locality where Utah Pie sold its products. The prices set by the national pie manufacturers were all below their prices in other regions, and there was evidence that many of the prices were below “cost” — a term which the Court, at one point, apparently defined as “less than . . . direct cost plus an allocation for overhead.” Utah Pie, 386 U.S. at 698, 87 S.Ct. at 1333. The price of frozen pies, including those of Utah Pie, fell dramatically. Utah Pie’s share of the local market fell from 66.5% to 45.3%. Utah Pie’s sales, however, nearly tripled and its net income increased.
Upon this evidence, the Supreme Court, reversing the Court of Appeals, reinstated the jury’s finding of liability against the national manufacturers. The Court held that the competitive harm requirement had been satisfied because: (1) The “drastically declining price structure” constituted actual competitive harm. Id. at 703, 87 S.Ct. at 1336. (2) There was evidence of predatory intent which bore “on the likelihood of injury to competition.” Id. at 702, 87 S.Ct. at 1335 (footnote omitted). This predatory intent was shown by the defendants’ charging lower prices in the Salt Lake City market, and apparently, though it is not made explicit, by pricing below average cost.
Utah Pie thus does not stand for the proposition that in a case such as the present one, which is a selective, non-geographic price discrimination case, pricing below average cost is sufficient evidence upon which a jury can infer competitive harm when no actual competitive harm has been shown. The Supreme Court did not indicate that the competitive harm requirement could be satisfied merely by a showing of below-average cost pricing. It only stated that “the existence of predatory intent might bear on the likelihood of injury to competition,” id. (footnote omitted), in the context of a case where it also found actual competitive harm. When inferring predatory intent from “below-cost” pricing, the Court also emphasized that the defend[352]*352ants sold pies in the Salt Lake City market at a lower price than elsewhere in the United States. Id. at 690, 694, 698, 701, 87 S.Ct. at 1329, 1331, 1333, 1335. Therefore, even if Utah Pie would support the theory that predatory intent, and hence competitive harm, could be shown merely by below-average cost pricing, such a theory would only apply in a geographic price discrimination case.
The Williamson cost theory15 relied on by Hommel and apparently by the district court, does not convince us that in a selective price discrimination case, predatory intent can be inferred from below cost pricing. Williamson would have a court find pricing to be predatory if it is below average cost over a long term when there is a loose oligopoly and when demand is not declining. However, we do not read his analysis to extend to the selective price discrimination context presented in this case. Whatever the merits of the Williamson theory as it pertains to across-the-board predatory pricing, we have difficulty in relating that theory to the present circumstances, which involve no more than lower pricing for a limited number of customers. Indeed, although Williamson responds to the Areeda and Turner article, Williamson’s analysis does not address that aspect of Areeda and Turner which would free selective price discrimination from regulation. Thus, we are forced to conclude that reliance on the Williamson formulation would be misplaced in a case such as this one where we are faced with selective price discrimination.16
Although we have discussed various economic theories as they have been urged upon us by the parties, we need not choose among them to conclude that in this case insufficient evidence existed to warrant submission of the Robinson-Patman claim to the jury. While we are inclined to accept the basic premise of the Areeda and Turner thesis that predatory intent may not be inferred from sales at or above average variable cost, in light of the issue that we must resolve here and the record which presents it, we need not declare our adherence to that or any other economic theory proposed. The record here discloses that Ferro’s discrimination in prices caused no competitive harm. The selective price discrimination practiced by Ferro resulted in no gain in Ferro’s market share. Indeed, between 1973 and 1977, Ferro’s share of the market decreased. During that same period, Hommel retained its share of the market and its overall economic position improved. No express evidence of predatory intent appears in the record, and Hommel thus sought to draw inferences of predatory intent from Ferro’s pricing, relying on Robinson-Patman cases which involved not selective price discrimination, but geographic price discrimination. Yet, here the record reveals that Ferro’s price discrimination was dispersed throughout the national market. Indeed, the price discriminations which are the focal point of Hommel’s action, were limited to three customers and affected only 5% of Ferro’s national sales.
In such circumstances we have grave reservations whether any sales priced below [353]*353any cost formulation could, without more, be deemed indicative of predatory intent because our review of the record shows it to be barren of any substantial lessening of competition or any tendency to create a monopoly17 caused by Ferro’s discriminatory pricing.
Thus, without relation to any particular cost formula, we are satisfied that no liability under Robinson-Patman was incurred by Ferro. If, however, we were to look to the Areeda and Turner thesis to which we have been attracted, it is evident that our conclusion in this respect would be buttressed, providing, of course, that Areeda and Turner would even apply their theory to a selective price discrimination case. 88 Harv.L. Rev. at 725-26. We observe that the prices at which Ferro sold frit to General Housewares, Briggs and General Electric were all above, not below, average variable cost, which under the analysis of Areeda and Turner could not lead to any inference of predatory pricing.
We conclude therefore, that “as a matter of law, the record is critically deficient of that minimum quantum of evidence from which a jury might reasonably afford relief.” Denneny v. Siegel, 407 F.2d 433, 439 (3d Cir. 1969).
III.
Hommel contends that even if we find the evidence to be insufficient for a jury to find liability, we cannot enter judgment for Ferro because Ferro moved for judgment notwithstanding the verdict out of time. According to Hommel, in this circumstance, we can only remand for a new trial. See Johnson v. New York, New Haven & Hartford R.R., 344 U.S. 48, 54, 73 S.Ct. 125, 128, 97 L.Ed. 77 (1952). Rule 50(b) of the Federal Rules of Civil Procedure provides:
No later than 10 days after entry of judgment, a party who has moved for a directed verdict may move to have the verdict and any judgment entered thereon set aside and to have judgment entered in accordance with his motion for a directed verdict.
Ferro moved for a judgment notwithstanding the verdict within ten days of the district court’s Supplementary Judgment Order of June 16, 1980, which set the amount of attorneys’ fees. It did not move for an NOV judgment within ten days of the jury verdict and the entry of judgment thereon.
We agree with both parties that “judgment” means final judgment. Rule 54 defines judgment as used in the Federal Rules of Civil Procedure as “includpng] a decree and any order from which an appeal lies.” Rule 4(a)(4) of the Federal Rules of Appellate Procedure assumes that an NOV motion will be made after final judgment:
(4) If a timely motion under the Federal Rules of Civil Procedure is filed in the district court by any party: (i) for judgment under Rule 50(b); (ii) under Rule 52(b) to amend or make additional findings of fact, whether or not an alteration of the judgment would be required if the motion is granted; (iii) under Rule 59 to alter or amend the judgment; or (iv) under Rule 59 for a new trial, the time for appeal for all parties shall run from the entry of the order denying a new trial or granting or denying any other such motion. A notice of appeal filed before the disposition of any of.the above motions shall have no effect. A new notice of appeal must be filed within the prescribed time measured from the entry of the order disposing of the motion as provided above. No additional fees shall be required for such filing.
Rule 50(b) had been amended in 1963 to provide that the motion for a judgment NOV should be made within ten days of the entry of judgment, rather than ten days after the reception of the verdict, so as to be consistent with Rules 52(b) and 59(b). See Notes of Advisory Committee on Rule 50(b), 18 U.S.C.A. at 376 (1963). The First Circuit has held that “judgment” in Rule 59(b) means final judgment. Warner v. Rossignol, 513 F.2d 678, 684 n.3 (1st Cir. [354]*3541975). We thus conclude that entry of judgment should be construed as entry of final judgment.
A judgment is not final until attorneys’ fees have been determined. Richerson v. Jones, 551 F.2d 918 (3d Cir. 1977). Hommel argues that Richerson was overruled by De Long Corp. v. Raymond International, Inc., 622 F.2d 1135, 1138-39 n.3 (3d Cir. 1980). Yet a panel of this court cannot overrule a prior panel precedent. United States Court of Appeals for the Third Circuit, Internal Operating Procedures, Ch. VIII, § C (1978). To the extent that it is inconsistent with Richerson, De Long must be deemed without effect.18 Since this judgment was not final until the order of June 16, 1980, which established attorneys’ fees, Ferro’s NOV motion was timely, and judgment for Ferro may thus be entered.
IV.
On the record before us, Ferro’s below-average cost pricing to three of its customers has caused no competitive harm within the meaning of 15 U.S.C. § 13 to the porcelain frit industry. Nor does the record disclose explicit evidence of any predatory intent from which^ competitive harm could be inferred. All Hommel has shown is selective below-average cost pricing to three customers of Ferro. This evidence falls far short of establishing that Ferro was sacrificing short-term profits in the hope that it would recoup monopoly profits. If, on this record, we were to sustain Robinson-Pat-man liability, we would be encouraging the very price rigidity and uniformity which the Supreme Court warned us against in Great Atlantic & Pacific Tea Co. v. FTC, 440 U.S. 69, 80, 99 S.Ct. 925, 933, 59 L.Ed.2d 153 (1979). Our holding would thus run counter to the broader policies of the antitrust laws. Id. at 80, n.13, 99 S.Ct. at 933. Having concluded that evidence of competitive harm and of predatory intent is lacking, we will reverse the district court and direct that judgment be entered for Ferro.