WALKER, Circuit Judge:

A Quaker State television commercial asserts that "tests prove" its 10W-30 motor oil provides better protection against engine wear at start-up. In a thoughtful opinion reported at 1992 WL 47981 (S.D.N.Y. March 2, 1992), the United States District Court for the Southern District of New York (Charles S. Haight, Judge) held that plaintiff-appellee Castrol, Inc. ("Castrol") had proven this advertised claim literally false pursuant to § 43(a) of the Lanham Act, 15 U.S.C. § 1125(a) (1988). The district court issued a March 20, 1992 Order preliminarily enjoining defendants-appellants Quaker State Corporation, Quaker State Oil Refining Corporation, and Grey Advertising Inc., ("Quaker State"), from airing the commercial. We agree that Castrol has shown a likelihood of success in proving the commercial literally false. We accordingly affirm.

BACKGROUND

Judge Haight's March 2, 1992 opinion thoroughly recites the facts of this case. We describe only those facts essential to the disposition of this appeal.

The voiceover to Quaker State's 10W-30 motor oil commercial states:

Warning: Up to half of all engine wear can happen when you start your car.

At this critical time, tests prove Quaker State 10W-30 protects better than any other leading 10W-30 motor oil.

In an overwhelming majority of engine tests, Quaker State 10W-30 flowed faster to all vital parts. In all size engines tested, Quaker State protected faster, so it protected better.

Get the best protection against start up wear. Today's Quaker State! It's one tough motor oil.

Visually, the commercial begins with a man entering a car and then shows a bottle of Quaker State 10W-30 motor oil. Large, block letters, superimposed over the bottle, "crawl" across the screen with the words:

AT START UP QUAKER STATE 10W-30 PROTECTS BETTER THAN ANY OTHER LEADING 10W-30 MOTOR OIL.

Originally, this "crawl" used the words "tests prove" instead of "at start up," but shortly after the filing of the current lawsuit Quaker State revised the message. The commercial then shows an engine, superimposed over which are bottles of Quaker State and four competing motor oils (including Castrol GTX 10W-30) and a bar graph depicting the speed with which each oil flowed to components of a Chrysler engine. The Quaker State bar is higher than all four competitors indicating that it flowed faster. The commercial closes with the words: "ONE TOUGH MOTOR OIL."

Polymethacrylate or "PMA," an additive intended to quicken oil flow to engine parts, is the source of Quaker State's superiority claim. The competitors listed in its commercial use olefin copolymer or "OCP," another additive. Two laboratory tests, the first run in 1987 and the second in 1991, have compared Quaker State's PMA-based oil with competing OCP-based oils. Rohm and Haas, the Pennsylvania corporation which manufactures PMA, conducted both tests.

Rohm and Haas' 1987 tests measured two performance indicators: "oiling time," or the time it takes for oil to reach distant parts in a just-started engine, and engine wear, measured through the amount of metal debris observed in the oil after the engine had run. Rohm and Haas technicians filled engines, in all other respects similar, with either Quaker State's PMA-based 10W-30 oil, or with a generic OCP-based oil known as "Texstar." During numerous engine starts, Quaker State's oil demonstrated a substantially faster oiling time, reaching distant engine parts as much as 100 seconds earlier than the Texstar competitor. Contrary to expectations, however, this did not translate into reduced engine wear. A Rohm and Haas report stated that "[a]fter 64 starts . . . the Quaker State oil gave marginally better results, but there was no significant difference in wear metals accumulation between the two oils."

Rohm and Haas initially attributed the poor engine wear results to the presence of "residual oil" remaining from the prior engine starts. They theorized that this oil might be lubricating the engine in the period between ignition and arrival of the new oil, and so might be preventing the faster flowing Quaker State oil from demonstrating better protection that is statistically significant. To address this, they conducted additional engine starts with a warm-up between each run so as to burn off the residual oil. The Rohm and Haas report, however, concluded that "[w]ear metals analysis for this test cycle also failed to differentiate significantly between the two oils. . . ." Thus, while the 1987 Rohm and Haas tests demonstrated faster oil flow, they could not prove better protection against engine wear that is statistically significant.

The 1991 Rohm and Haas tests compared Quaker State's oiling time with that of four leading OCP-based competitors, including Castrol GTX 10W-30. Again, Quaker State's PMA-based oil flowed significantly faster to engine parts. Using a 1991 2.2 liter Chrysler engine with a sump temperature of minus 20 degrees Fahrenheit, for example, the Quaker State oiling time was 345 seconds, as compared to the competing oils' times of 430, 430, 505 and 510 seconds. In the 1991 tests, as opposed to the 1987 studies, Rohm and Haas made no attempt to measure whether this faster oiling time resulted in reduced engine wear.

Quaker State broadcast their commercial in November, 1991. On December 19, 1991, Castrol initiated the present action. Castrol asserted that no studies supported the commercial's claim that "tests prove" Quaker State's oil provides better protection, and that this claim of test-proven superiority constituted false advertising. It sought preliminary and permanent injunctive relief and damages pursuant to § 43(a) of the Lanham Act, 15 U.S.C. § 1125(a), New York General Business Law §§ 349, 350, and common law unfair competition.

At the hearing on the motion for a preliminary injunction, Quaker State relied on the Rohm and Haas tests. It argued that the Rohm and Haas oiling time findings support the advertised claim of better protection because oil which flows faster to engine parts necessarily protects them better. Dr. Elmer Klaus, Quaker State's sole expert witness, explained this "faster means better" theory as follows: Prior to start-up "the metal parts [of an engine] are not separated by a film of oil. The solid members are sitting on each other," a condition referred to as "boundary lubrication." Upon ignition, engine wear begins to occur. Soon, however, the movement of the parts generates a film of lubrication from the "residual oil" remaining from a prior running of the engine and engine wear ceases. But the heat of the running engine thins the residual oil which can no longer keep the parts sufficiently apart. The engine returns to a condition of boundary lubrication and wear again occurs until the arrival of the new oil. Dr. Klaus concluded that the faster the new oil flows to the engine parts, the better job it does of minimizing this second period of boundary lubrication. Faster oil flow, therefore, means better protection.

Castrol's three experts focused on the role of residual oil. They testified that the small amount of residual oil left from a prior running of an engine provides more than adequate lubrication at the next startup. Moreover, they asserted that this residual oil remains functional for a significant period of time so that both PMA-based and OCP-based 10W-30 motor oils reach the engine parts before this residual oil burns off. Thus, they maintained, there is no second boundary lubrication period and Quaker State's faster oiling time is irrelevant to engine wear.

Castrol's experts supported their residual oil theory with a Rohm and Haas videotape, produced in the course of its tests, which shows the residual oil present on the cam lobe interface of a Chrysler 2.2 liter engine. Dr. Hoult, who narrated the tape for the court, explained that "as the film goes on the lubricant there will never go away[,] which means it's lubricated throughout the starting process and that's the basic reason that the time for the replenishment oil to reach these parts is not related to wear[,] because the parts have already lubricated okay."

The experts also cited the near absence of catastrophic engine failure since the imposition of mandatory "pumpability" standards, known as "J300" standards, in the early 1980's. Pumpability refers to the ease with which the pump can spread oil throughout the engine. As pumpability increases, oiling times decrease. Prior to the J300 standards, certain oils became unpumpable in cold weather. This, the experts testified, caused engines to suffer catastrophic failure within a "fraction" of a second after the residual oil had burned off. The J300 standards, however, required increased pumpability and have virtually eradicated reported cases of engine failure. The experts inferred that all 10W-30 oils, which are required to meet the J300 standards, must therefore be reaching the engine before the residual oil burns off. At best, there is only a "fraction" of a second between residual oil burn-off and catastrophic failure during which a faster flowing oil could conceivably reduce engine wear.

The district court assessed the parties' conflicting testimony in its March 2, 1991 opinion. Judge Haight found Dr. Klaus' testimony lacking in credibility because

[Dr. Klaus'] current research programs at Pennsylvania State University are funded in significant part by Quaker State or an industry association to which Quaker State belongs. Klaus arrived at his opinions not on the basis of independent research but by digesting technical papers furnished to him by Quaker State and Rohm and Haas in preparation for his testimony; and he acknowledged that he reached his conclusion concerning Quaker State's better protection before even being made aware of the contrary 1987 Rohm and Haas tests.

Judge Haight credited the testimony of Castrol's three experts. In addition, he found their testimony corroborated by three key facts: (1) the failure of the 1987 Rohm and Haas tests to demonstrate reduced engine wear; (2) the Rohm and Haas technician's 1987 hypothesis that the presence of residual oil might be the reason for the failure to show better engine wear protection that is statistically significant; and (3) the virtual disappearance of catastrophic engine failure following the imposition of the J300 standards. Judge Haight accordingly "accept[ed]" the residual oil theory put forth by these experts. The court explained that an engine is like "a fort besieged by an encircling and encroaching enemy." The enemy is engine wear; the fort's supplies are residual oil; and a relief column on its way to reinforce the fort is the new oil. "If that relief column does not reach the bearing surfaces before the residual oil is burned away, the engine will suffer not only wear but catastrophic failure. . . . [T]he Quaker State commercial is false because the evidence shows that during the time differentials demonstrated by the [Rohm and Haas] oiling tests, residual oil holds the fort."

Judge Haight concluded that because residual oil "holds the fort," Rohm and Haas' faster oiling time findings did not necessarily prove better protection. He consequently held that "Castrol has established the likelihood of proving at trial the falsity of Quaker State's claim that tests prove its oil protects better against start-up engine wear." On March 20, 1992, 1992 WL 73569 the district court entered an Order granting preliminary injunctive relief. Quaker State appeals.

DISCUSSION

A party seeking preliminary injunctive relief must show (a) that it will suffer irreparable harm if relief is denied, and (b) either (1) a likelihood of success on the merits or (2) sufficiently serious questions going to the merits to make them fair ground for litigation and a balance of hardships tipping decidedly in plaintiff's favor. See Procter Gamble Co. v. Chesebrough-Pond's, Inc., 747 F.2d 114, 118 (2d Cir. 1984); Coca-Cola Co. v. Tropicana Prods., Inc., 690 F.2d 312, 314-15 (2d Cir. 1982); United States v. Siemens Corp., 621 F.2d 499, 505 (2d Cir. 1980). We will presume irreparable harm where plaintiff demonstrates a likelihood of success in showing literally false defendant's comparative advertisement which mentions plaintiff's product by name. See McNeilab, Inc. v. American Home Products Corp., 848 F.2d 34, 38 (2d Cir. 1988); Nester's Map Guide Corp. v. Hagstrom Map Co., 760 F.Supp. 36, 36 (E.D.N.Y. 1991); Valu Eng'g, Inc. v. Nolu Plastics, Inc., 732 F.Supp. 1024, 1025 (N.D.Cal. 1990).

Section 43(a) of the Lanham Act, 15 U.S.C. § 1125(a) (1988), pursuant to which Castrol brings this false advertising claim, provides that

Any person who, on or in connection with any goods or services . . . uses in commerce any . . . false or misleading description of fact, or false or misleading representation of fact, which —

* * * * * *

(2) in commercial advertising or promotion, misrepresents the nature, characteristics, qualities, or geographic origin of his or her or another person's goods, services, or commercial activities,

shall be liable in a civil action by any person who believes that he or she is or is likely to be damaged by such act.

To succeed under § 43(a), a plaintiff must demonstrate that "an advertisement is either literally false or that the advertisement, though literally true, is likely to mislead and confuse consumers. . . . Where the advertising claim is shown to be literally false, the court may enjoin the use of the claim `without reference to the advertisement's impact on the buying public.'" McNeil-P.C.C., Inc. v. Bristol-Myers Squibb Co., 938 F.2d 1544, 1549 (2d Cir. 1991) (quoting Coca-Cola, 690 F.2d at 317) (citations omitted). Here, Castrol contends that the challenged advertisement is literally false. It bears the burden of proving this to a "likelihood of success" standard.

As we have on two occasions explained, plaintiff bears a different burden in proving literally false the advertised claim that tests prove defendant's product superior, than it does in proving the falsity of a superiority claim which makes no mention of tests. In Procter Gamble Co. v. Chesebrough-Pond's, Inc., 747 F.2d 114 (2d Cir. 1984), for example, Chesebrough alleged the literal falsity of Procter's advertised claim that "clinical tests" proved its product superior. Id. at 116. Procter, in return, challenged as literally false a Chesebrough commercial which, making no mention of tests, asserted that its lotion was equal in effectiveness to any leading brand. Id. We explained that in order to prove literally false Procter's claim of "test-proven superiority," Chesebrough bore the burden of "showing that the tests referred to by P G were not sufficiently reliable to permit one to conclude with reasonable certainty that they established the proposition for which they were cited." Id. at 119. We held that Procter could prove false Chesebrough's advertisement, however, "only upon adducing evidence" that affirmatively showed Chesebrough's claim of parity to be false. Id.

We drew this same distinction in McNeil-P.C.C., Inc. v. Bristol-Myers Squibb Co., 938 F.2d 1544 (2d Cir. 1991). Bristol-Myers initially advertised to trade professionals that "clinical studies" had shown its analgesic provided better relief than McNeil's. Id. at 1546. Bristol-Myers' later televised commercial made the product superiority claim but "did not refer to clinical studies." Id. We held that, with respect to the initial trade advertising, "McNeil could . . . meet its burden of proof by demonstrating that these studies did not establish that AF Excedrin provided superior pain relief." Id. at 1549. With respect to the televised commercial, however, McNeil bore the burden of generating "scientific proof that the challenged advertisement was false." Id.

A plaintiff's burden in proving literal falsity thus varies depending on the nature of the challenged advertisement. Where the defendant's advertisement claims that its product is superior, plaintiff must affirmatively prove defendant's product equal or inferior. Where, as in the current case, defendant's ad explicitly or implicitly represents that tests or studies prove its product superior, plaintiff satisfies its burden by showing that the tests did not establish the proposition for which they were cited. McNeil, 938 F.2d at 1549. We have held that a plaintiff can meet this burden by demonstrating that the tests were not sufficiently reliable to permit a conclusion that the product is superior. Procter, 747 F.2d at 119; see also Alpo Petfoods, Inc. v. Ralston Purina Co., 720 F.Supp. 194, 213 (D.D.C. 1989), aff'd in part and rev'd in part on other grounds, 913 F.2d 958 (D.C.Cir. 1990); American Home Prods. Corp. v. Johnson Johnson, 654 F.Supp. 568, 590 (S.D.N.Y. 1987); Thompson Medical Co. v. Ciba-Geigy Corp., 643 F.Supp. 1190, 1196-99 (S.D.N.Y. 1986). The Procter "sufficiently reliable" standard of course assumes that the tests in question, if reliable, would prove the proposition for which they are cited. If the plaintiff can show that the tests, even if reliable, do not establish the proposition asserted by the defendant, the plaintiff has obviously met its burden. In such a case, tests which may or may not be "sufficiently reliable," are simply irrelevant.

The district court held that Castrol had met this latter burden, stating that "Castrol has established the likelihood of proving at trial the falsity of Quaker State's claim that tests prove its oil protects better. . . ." In this Lanham Act case, we will reverse the district court's order of preliminary injunctive relief "only upon a showing that it abused its discretion, which may occur when a court bases its decision on clearly erroneous findings of fact or on errors as to applicable law." Procter, 747 F.2d at 118.

I. The district court committed no errors of law.

Quaker State contends that the district court improperly shifted the burden of proof to the defendant when it stated that "the claim that tests demonstrate . . . superiority is false because no test does so and [Dr.] Klaus' analysis fails to fill the gap." It argues that plaintiff bears the burden in a false advertising action and there should be no "gap" for defendant to fill.

Where a plaintiff challenges a test-proven superiority advertisement, the defendant must identify the cited tests. Plaintiff must then prove that these tests did not establish the proposition for which they were cited. McNeil, 938 F.2d at 1549. At the hearing, Quaker State cited the 1987 and 1991 Rohm and Haas oiling time tests in conjunction with Dr. Klaus' theory of engine wear at the second boundary lubrication period. Castrol's burden was to prove that neither the Rohm and Haas tests alone, nor the tests in conjunction with Dr. Klaus' theory, permitted the conclusion to a reasonable certainty that Quaker State's oil protected better at start-up. The district court's statement that "no test [demonstrates superiority] and Klaus' analysis fails to fill the gap" is a finding that Castrol, through its residual oil theory, met its burden. It is, in substance, a finding that the Quaker State tests, which proved faster oiling time, are irrelevant to their claim that Quaker State's oil protects better at start-up. Therefore, we need not consider the tests' reliability. The district court's statement does not shift the burden to defendant.

Quaker State also contends that the district court should have subjected the 1987 Rohm and Haas engine wear results to Procter's "sufficiently reliable" test before relying on them. It argues, in other words, that the Procter standard applies not only to the studies offered to support defendant's claim of test-proven superiority, but also to plaintiff's evidence offered to rebut this claim.

Quaker State misreads Procter. In that case, we established that a plaintiff proves false a test-proven superiority claim when it shows that "the tests referred to by [defendant were] not sufficiently reliable. . . ." Id. at 119. This phrase merely establishes plaintiff's burden of proof with respect to defendant's tests. It in no way limits the evidence which plaintiff may use in meeting this burden. Such evidence is governed by the usual standards of admissibility. It was not error for the district court to consider the 1987 tests in this regard.

II. The district court's findings as to the role of residual oil were not clearly erroneous.

Quaker State asserts that the district court's factual findings as to the role of residual oil are clearly erroneous. See Fed.R.Civ.P. 52(a). We disagree.

The Supreme Court has explained that "[i]f the district court's account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it even though convinced that had it been sitting as the trier of fact, it would have weighed the evidence differently. Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous." Anderson v. City of Bessemer City, 470 U.S. 564, 573-74, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985); see also ABKCO Music, Inc. v. Harrisongs Music, Ltd., 944 F.2d 971, 978 (2d Cir. 1991). We owe particularly strong deference where the district court premises its findings on credibility determinations. "[W]hen a trial judge's finding is based on his decision to credit the testimony of one of two or more witnesses, each of whom has told a coherent and facially plausible story that is not contradicted by extrinsic evidence, that finding, if not internally inconsistent, can virtually never be clear error." Anderson, 470 U.S. at 575, 105 S.Ct. at 1512; see also ABKCO, 944 F.2d at 978 (trial court's credibility determinations entitled to "considerable deference"). A district court finding is clearly erroneous only where "`although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.'" Anderson, 470 U.S. at 573, 105 S.Ct. at 1511 (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)).

In this case, the district court heard five days of expert testimony. Its credibility determinations in favor of Castrol's experts and against Quaker State's support its finding that "residual oil holds the fort." This finding also receives support from the videotape of residual oil in an engine, the absence of catastrophic engine failure following the imposition of the J300 standards, and Rohm and Haas' 1987 failure to demonstrate reduced engine wear. Nothing in the record convincingly contradicts the district court's conclusion. Under the applicable legal standards, we are hard pressed to hold Judge Haight's residual oil finding clearly erroneous.

Quaker State argues that the residual oil theory flies in the face of decades of technical literature documenting the existence of start-up wear. It reasons that if residual oil truly lasted until the new oil arrived, start-up wear would not be possible.

Dr. Hoult, a Castrol expert, answered this point. He testified that the term "start-up," as used in the cited technical papers, refers not to the period between ignition and full oil flow but to the time between ignition and the achievement of equilibrium temperature in the engine. The relatively cool engine temperature during the start-up period, thus defined, results in increased wear due to certain chemical properties best described by Dr. Hoult himself:

When the engine is driven for a short period of time under cold conditions, it never gets fully warm. And in the combustion process of reciprocating engines, there are acids which are formed, typically nitrate acid; if the fuels have sulfur, sulfuric acid. In a cold engine, there is more acids that mix with the lubricants than there is in a hot engine, because in a hot engine, the parts are hot enough that the acid doesn't condense on them. So that when an engine is colder[,] when it's started up from cold conditions, the engine chemistry is different. And the general understanding [in the field] is that that changes engine wear rate.

This credited testimony effectively rebuts Quaker State's objection. Viewing the record as a whole, we are not left with a "`definite and firm conviction that a mistake has been committed.'" Anderson, 470 U.S. at 573, 105 S.Ct. at 1511 (citation omitted). We accordingly reject Quaker State's contention that Judge Haight's findings on the role of residual oil are clearly erroneous.

III. Is the district court's injunction overly broad?

In a March 20, 1992 memorandum opinion accompanying its simultaneously-issued Order of Preliminary Injunction, the district court explained its intent "to enjoin preliminarily Quaker State from claiming `that tests prove its oil protects better against start-up engine wear.'" The injunction, however, goes beyond this limited intent. Paragraph 2 of the injunction states that

Defendants . . . are preliminarily enjoined from broadcasting, publishing or disseminating, in any manner or in any medium, any advertisement, commercial, or promotional matter . . . that claims, directly or by clear implication, that:

(a) Quaker State 10W-30 motor oil provides superior protection against engine wear at start-up;

(b) Quaker State 10W-30 motor oil provides better protection against engine wear at start-up than other leading 10W-30 motor oils, including Castrol GTX 10W-30; or

(c) Castrol GTX 10W-30 motor oil provides inferior protection against engine wear at start-up.

This paragraph enjoins Quaker State from distributing any advertisement claiming that its oil provides superior protection against engine wear at start-up, whether or not the ad claims test-proven superiority. As explained above, Castrol bears a different burden of proof with respect to this broader injunction than it does in seeking to enjoin only commercials which make the test-proven superiority claim.

The district court expressly found that Castrol had met its burden with respect to any test-proven superiority advertisement. It stated that "Castrol has established the likelihood of proving at trial the falsity of Quaker State's claim that tests prove its oil protects better . . . ." Its injunction would be too broad, however, absent the additional finding that Castrol had met its burden with respect to superiority advertisements that omit the "tests prove" language. As we have noted above, Castrol meets this burden by adducing proof that Quaker State's oil is not, in fact, superior.

Judge Haight made this additional finding. Castrol submitted the report from the 1987 Rohm and Haas tests as proof that Quaker State's oil did not protect better. This submission was proper under our holding that "[plaintiff can] rel[y] on and analy[se] data generated by [defendant] as scientific proof that the challenged advertisement was false." McNeil-P.C.C., 938 F.2d at 1549. The district court, referring to this document, stated that "the record makes it crystal clear that to the extent tests were performed to demonstrate better wear protection (as opposed to faster flowing), the tests contradict, rather than support the claim. I refer to the 1987 Rohm and Haas tests . . . ." (emphasis added). The court went on to find that "Quaker State presents no convincing argument to counter the unequivocal conclusion of Roland [author of the 1987 report], a Rohm and Haas scientist, that the 1987 tests failed to demonstrate a superiority in protection against engine wear. . . ." These statements amount to a finding that Castrol has met the additional burden. The injunction is not overly broad.

Quaker State also asks us to limit the injunction to advertisements based on the 1987 and 1991 Rohm and Haas tests. It contends that it should not be barred from advertising a superiority claim if later tests should support it.

Any time a court issues a preliminary injunction there is some chance that, after the issuance of the order but prior to a full adjudication on the merits, changes in the operative facts will undercut the court's rationale. We will not, however, require the district court to draft a technical and narrow injunction to address the possibility of additional tests which are, at this time, purely hypothetical. If tests supporting its claim do come to light, Quaker State may move to modify or dissolve the injunction. See Flavor Corp. of Am. v. Kemin Indus., Inc., 503 F.2d 729, 732 (8th Cir. 1974); 11 C. Wright A. Miller, Federal Practice and Procedure, § 2961 at 604 (1973). We will likely have jurisdiction to review the district court's disposition of such a motion, and can consider the issue at that point if necessary. See 28 U.S.C. § 1292(a)(1) (1988); United States v. City of Chicago, 534 F.2d 708, 711 (7th Cir. 1976) (denial of motion to dissolve preliminary injunction is appealable pursuant to 28 U.S.C. § 1292(a)(1); Int'l Brotherhood of Teamsters v. Western Penn. Motor Carriers Ass'n, 660 F.2d 76, 80 (3d Cir. 1981) (denial of motion to amend injunction is appealable).

CONCLUSION

We affirm the district court's March 20, 1992 Order granting the preliminary injunction.