dissenting in part and concurring in part:
In a four-day trial that was fair to both sides, the defendants successfully defended the case Crofton asserted against them, which included a CERCLA claim that they dumped drums contaminated with TCE on a parcel of land in Anne Arundel County, Maryland. The majority nevertheless concludes that a remand of the CERCLA claim is necessary because the district judge did not understand the bases of liability under the Act and therefore failed to assess the evidence properly in making his factual findings. I respectfully disagree. While the evidence was being presented at trial, Crofton’s theory was that the defendants dumped the TCE-filled drums on the site during the time they owned or operated it. During closing arguments, however, Crofton attempted to raise a new theory of liability based on passive disposal, that is, the leaking of TCE. I agree with the district judge’s refusal to indulge this belated change in theory because of the prejudice to the defendants, who had defended a different case. In focusing on the case that was actually tried, the judge concluded that Crofton failed to prove by a preponderance of the evidence that contaminated drums were placed on the site during the time the defendants were responsible. This finding was not clearly erroneous. I would therefore affirm the judgment in favor of the defendants.
I.
The remand ordered by the majority gives Crofton the chance to have its case reconsidered under a new (and perhaps less burdensome) theory, that TCE leaked from the drums during the time that the defendants owned or operated the site. This is fundamentally unfair because Crof-ton waived the passive disposal theory of liability by failing to raise it until closing argument in a five-day trial.
In an extended closing argument session, Crofton’s lawyers spent most of their time trying to convince the district judge that the defendants were responsible for dumping the drums found at the site. Then, at the very end, one of Crofton’s lawyers said, “I want to take a step back ... and kind of drop a bomb shell on this Court .” He explained that he wanted to include the passive disposal (or leaking) theory of liability in his argument. The defendants’ lawyer objected to the introduction of this theory, saying that Crofton had not based its case on passive disposal. The defense pointed out, for example, that there had been “[n]o discovery about passive disposal.... If there had been, we would have had lots of testimony on it.” The district judge, who was immediately concerned because the defendants had no notice of the passive disposal theory, said:
It is a little bit strange here, this [passive disposal] concept for the first time in final argument. Why did we go through four days of testimony without hearing about it? I don’t understand. ... Why didn’t you mention this before? ... It would [be] nice if they [defendants] knew what they were trying — if they really knew they were going to try that issue.
The judge nevertheless allowed Crof-ton’s lawyer to attempt to identify evidence of leaking. In the end, however, the judge returned to the notice problem:
I cannot believe in final argument you are going to be able to change this case from what everybody tried it to something that nobody tried.... I am sorry. It is absolutely not the slightest notice that this [passive disposal] was going to be tried.... It is not an additional argument. It is a totally different case than what they [defendants] thought they *302were trying. You look at the pretrial order. You don’t see the slightest reference to this.... I got to tell you I didn’t have to listen to all that stuff I heard for the last three years. It is a totally different ease. Secondly, it wasn’t even raised in argument with regard to liability; so let’s proceed. I cannot allow you to amend the pleadings and do this at this time. This is unconsci[o]nable; un-consci[o]nable. Unconsci[o]nable. I can’t do it.
It was thus clear to the trial judge that leaking was never an issue in the case. The record supports the judge’s understanding. The passive disposal theory did not surface in discovery or in any other pretrial proceeding. And most important, at trial Crofton did not indicate during opening argument or at any other time before the close of evidence that it intended to base liability on leaking.
I agree with the majority that a number of events or acts can constitute “disposal” for purposes of CERCLA liability. This, however, does not excuse a plaintiff from giving a defendant notice of the particular theory or theories of disposal being pressed in a given case.1 CERCLA provides that “any person who at the time of disposal of any hazardous substance owned or operated any facility at which such hazardous substances were disposed of’ is liable for cleanup costs and other damages. 42 U.S.C. § 9607(a)(2). The term “disposal” means:
the discharge, deposit, injection, dumping, spilling, leaking, or placing of any solid waste or hazardous waste into or on any land or water so that such solid waste or hazardous waste or any constituent thereof may enter the environment or be emitted into the air or discharged into any waters, including ground waters.
42 U.S.C. § 6903(8). Different cases present different grounds for disposal liability. Not every case involves facts suggesting both active disposal (for example, dumping or placing) and passive disposal (for example, leaking or spilling). See, e.g., Kaiser Aluminum, & Chem. Corp. v. Catellus Dev. Corp., 976 F.2d 1338, 1342 & n. 7 (9th Cir.1992) (declining to consider the question of passive migration because plaintiff alleged active disposal, specifically, that defendant excavated contaminated soil and spread it over a clean part of the property); Nurad, Inc. v. William E. Hooper & Sons Co., 966 F.2d 837, 844-46 (4th Cir.1992) (holding individual defendant liable because hazardous substances were leaking from tanks during the time that he *303owned the property, even though he did not handle the tanks). Crofton chose to frame and try its case as one involving active disposal only. The defendants responded accordingly. They did not pursue discovery on the issue of leaking, nor did they present evidence at trial to rebut a theory of leaking. This was not an oversight by the defendants, who legitimately believed that they were only facing a charge of active disposal. Crofton should not be permitted to argue now that this is a passive disposal case requiring only a showing that the TCE leaked into the environment during the time of the defendants’ ownership or operation. Indeed, it is unprecedented to allow Crofton to make such a prejudicial change in its theory — in its own words, to “drop a bomb shell” — at this late stage. See, e.g., Savarin Corp. v. Nat’l Bank of Pak., 447 F.2d 727, 732 (2d Cir.1971). The district judge was correct to conclude that Crofton waived the passive disposal theory of liability.
Moreover, the district judge was correct in his observation (during closing argument) that there was insufficient evidence to establish that TCE leaked from the drums during the time the defendants would have been responsible. Crofton points to the drum log and groundwater test results for evidence of leaking. See Br. of Appellant at 26; Reply Br. of Appellant at 6. The drum log describes drums in various states of damage or deterioration, including crushed, split, or leaking. The log, however, revealed only what was seen. The log was not based on any scientific analysis. Furthermore, because the log was prepared at the time of the cleanup in 1995, it does not address whether any leakage occurred earlier when the defendants owned or operated the site. The groundwater test results do show TCE contamination, but the testing was conducted after the drums had been removed. Tr. 328-29. The expert reports indicate that Crofton had not done any soil or groundwater sampling before the cleanup. See Expert Report of Bruce Monteith, Pl.’s Ex. 35 A., at 6-7; Expert Report of Joseph P. Lewandowski, Def. G & H’s Ex. 39, at 5, 10. In addition, the groundwater test results are irrelevant to the issue of leaking. That evidence was introduced for the sole purpose of proving that the drum removal complied with the National Contingency Plan (NCP), an issue that the district judge never reached. Before Crof-ton’s expert testified as to the groundwater test results, the judge ruled that the evidence could only be used to prove compliance with the NCP. Tr. 327, 329-30. All of this reveals that even if the issue of leaking is considered, the evidence, as the district judge recognized, is not sufficient to establish that the defendants are liable for passive disposal.
II.
It is beside the point to say, as the majority does, that the facts presented by Crofton “easily permit a factfinder to conclude” that TCE was placed on the site when defendants owned or operated it. Ante at 299. The record reveals that the district judge understood the bases of liability under CERCLA and that he made factual findings that were keyed to the case presented and the liability standards spelled out in the Act. We review factual findings for clear error. See Fed.R.Civ.P. 52(a) (“Findings of fact ... shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses.”). The district court’s factual findings are not clearly erroneous.
The majority contends that the district judge erred because he required “Crofton to show that the defendants themselves actively dumped the TCE at the site or otherwise were involved in the dumping .” Ante at 299. Although the district judge focused on whether Crofton proved that *304the defendants dumped the TCE on the site, this does not mean that the judge misunderstood the law. The judge was merely analyzing the case according to Crofton’s theory, specifically, that the defendants had produced the TCE and then illegally dumped it on another part of their property after 1977. See R. vol. 9 at 51 (closing argument by Crofton’s lawyer) (“We believe [the drums] all came from the asphalt plant that was constructed by Mr. Mitchell in 1977_”).2 Based on the case that Crofton presented, the judge framed the “central question” as “whether Plaintiff has proved the alleged dumping in the post 1976 period.” J.A. 826. See also J.A. 825 n. 14 (“[A]s discussed herein, the Court concludes that Plaintiff has failed to prove that there was any TCE dumping on the Site after 1977....”). The judge rejected some of the evidence because it did not shed light on when the dumping occurred. See J.A. 833 (“[I]t would be more likely that drum dumping took place prior to 1977, while the dump site area was denuded from strip mining, rather than later, close to 1984, after the vegetation had regrown.”); J.A. 834 (“The presence of [the fully corrugated] drums indicates that the dumping (at least of these drums) took place a long, long time ago.”); J.A. 835 (“[I]t would have been simple, yet potentially significant, to have preserved evidence of ... the vintage of the tires.”).
In a 31-page opinion the district judge explained in detail why he concluded that Crofton did not prove its active disposal case by a preponderance of the evidence. The first problem was “the absence of reliable proof of the location and size of the dump site.” One of Crofton’s assertions was that before 1977 heavy vegetation surrounding the dump site prevented trucks from driving to the site and unloading the drums, but that by 1984 the area was cleared and easily accessible. This, according to Crofton, allowed the inference that the drum dumping took place after 1977. But there was confusion in Crof-ton’s case about the location of the site. Crofton’s expert on aerial photographs placed the dump site at a location on the 1984 aerial photograph that was different from the location he identified on the 1968 and 1977 photographs. See R. vol. 8, ex. 3B-3D. Because of this confusion, the *305judge concluded that Crofton’s “argument [that the dumping took place after 1977] based upon the alleged progressive changes in the dump site [from heavily vegetated to cleared] totally lacked an evi-dentiary foundation.” The second problem was that much of the evidence that would have established the timing of the dumping either was not fully analyzed or was lost during the cleanup process. Crofton failed to present an expert analysis on the degradation of the drums and failed to specify how many drums were fully corrugated (in use in the 1930s and 1940s) and how many were semi-corrugated (in use later). Crof-ton likewise did not preserve evidence removed from the dump site that would have shed light on when the dump site was in use. This evidence included tires, mud flaps, road signs, and junked household appliances. The final problem was that Crofton’s evidence regarding the defendants’ involvement in the dumping was too speculative. The deposition of a former employee of the defendants supported Crofton’s case, but the judge considered the employee’s testimony unreliable because the defendants had fired him after he was caught stealing. The judge concluded that the remaining evidence created no more than a suspicion that the defendants dumped their TCE waste at the site. Crofton has not pointed to any record evidence that would fill the gaps in proof that the district judge identified.
Not once in part II does the majority mention the clearly erroneous standard of review, nor does it apply that standard. It is not enough to say, as the majority does, that a factfinder could conclude that TCE was placed on the site while the defendants owned or operated it. This is not the standard that governs our review. The district judge, who heard four days of testimony and studied almost one hundred exhibits, found that Crofton did not prove a case of active disposal. After reviewing the evidence and giving “due regard ... to the opportunity of the trial court to judge of the credibility of the witnesses,” I am firmly convinced that the district judge did not commit clear error in finding that Crofton did not prove its case. Fed. R.Civ.P. 52(a). See also Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985).
III.
This appeal at the core is not about CERCLA. It is about whether the defendants won according to rules governing the trial of a lawsuit and any appeal from the judgment. These rules are that a defendant is entitled to fair notice of the plaintiffs theory of liability at trial and that findings of fact are reviewed on appeal for clear error. Because the defendants won the case according to these rules, the majority’s remand of the CERCLA claim is simply wrong. I therefore respectfully dissent from parts I and II of the majority opinion, which govern the remand. I concur in part III, which affirms the judgment in favor of the defendants on Crofton’s common law claims.
. I respectfully disagree with the majority's assertion that the ''bombshell'' Crofton "dropped” in closing argument did not signal a new theory but rather was "advice to the district court that the court was reading CERCLA too narrowly.” Ante at 298 n. 3. Crofton was advancing a new theory, and the district judge did not misunderstand the law. As the judge recognized, "If any of the liquid leaks out, any of the TCE leaked out of the drum before [Crofton] moved it, that there would have been passive disposal.” Thus, he did not read CERCLA too narrowly. The judge's ultimate rejection of the passive disposal or leaking theory rested on Crofton’s delay in introducing the specific theory. In response to the judge's inquiry about why Crofton had never before mentioned leaking, its lawyer said only that the complaint, which used the general statutory term "disposed of,” was sufficient. The general language in the complaint, however, did not excuse Crofton from giving the defendants and the court reasonable notice as to which of the several statutory definitions of "disposal” it was basing its case upon. Leaking is not a "lesser included burden of proof," as the majority suggests, ante at 298 n. 3, but a separate ground for liability. It might be easier to prove leaking than dumping in some cases, but the defendant is still entitled to know whether the plaintiff's theory of the case includes leaking.
. The trial judge's comments that the majority includes in its footnote 3 did not result from any "limited interpretation of CERCLA.” Ante at 298 n. 3. Rather, they were made in reference to Crofton's theory that the defendants themselves placed the contaminated drums on the site. See R. vol. 9 at 7 (closing argument by Crofton's lawyer) ("[M]y argument it would be what's in the drums, who put them there, who knew about it and who cleaned it up.”); R. vol. 9 at 93-94 (judge's statement during closing argument) ("[Crof-ton’s lawyer] just went through a long list which said it is evidence from which I should find that you [defendants] put it there.”). Notwithstanding Crofton’s theory, the judge recognized that Crofton could prove an active disposal case by showing that some person, not just the defendants, placed the contaminated drums on the site while the defendants owned or operated it. As a result, the judge focused his analysis on the timing of the dumping. See, e.g., R. vol. 9 at 11 ("Why don’t we start with the proof that it was put there between '77 and '88.... ”).
According to the majority, the judge's comment that Crofton had no "responsibility” reveals the "court's misperception of the statute's scope.” Ante at 299 n. 3. The majority, I believe, has misconstrued the judge’s statement. It is true that as the current landowner Crofton is liable for cleanup costs under CERCLA. 42 U.S.C. § 9607(a)(1). But the judge was not referring to this technical statutory meaning. Instead, he was using the ordinary meaning of “responsibility.” See R. vol. 9 at 4-5 ("[Crofton] is innocent as I use the term. That may not be precisely the right word when you get down to it.”). In fact, the judge's ruling against Crofton even though he considered Crofton "innocent” shows that the judge understood CERCLA's strict liability scheme.