Plaintiff-appellee Olin Corporation ("Olin") brought this action seeking indemnification from various insurance companies, including defendants-appellants Certain Underwriters at Lloyd's London and London Market Insurance Companies (collectively London Insurers or "LI") for pollution liability. The district court separated these claims into groups for trial; this appeal concerns the subset of claims for costs of state-required pollution remediation at four sites near Niagara Falls, New York.1

Based on the district court's instructions regarding the meaning of property damage, the jury specifically found the years in which property damage occurred at each site, as well as the year in which the entire remediation that was later ordered became necessary. As to each site, the jury found the final year in which property damage occurred also to be the year in which the full remedy became necessary. Accordingly, the district court allocated the costs of remediation equally over the years in which the jury found property damage occurred, resulting in liability being attributed to LI's excess insurance policies in each of those years. LI argue that this method of allocation was based on an incorrect understanding of property damage, which they contend continued at each site until the date the remediation commenced there. They assert on appeal, as they did in the district court, that the evidence supports entry of judgment in their favor as to the extent of the damages and the resulting allocation of costs.

We hold that (1) costs of remediation should be allocated over the period in which property damage occurred, as nearly as possible according to the amount of property damage that occurred in each policy period, and (2) property damage includes the passive migration or spread of contaminants. Because the district court incorrectly instructed the jury on the definition of property damage, we remand to the district court for reconsideration of LI's motion for judgment as a matter of law, and if that motion is denied, then to conduct a new trial. We further hold that, on remand, LI should be permitted to raise the defense of late notice. However, we find LI's remaining assignments of trial error to be without merit.

Olin has been required by the New York State Department of Environmental Conservation to remediate environmental contamination at four sites near Niagara Falls, New York, under the Comprehensive Environmental Response, Compensation, and Liability Act, 42 U.S.C. § 9601 et seq. The relevant facts as to each sites are as follows.

Olin manufactured chemical products at its Niagara Falls plant site from 1897 to the mid-1990s. Olin's primary focus was the manufacture of chlorine and caustic soda, a process that creates a brine sludge containing mercury. Olin disposed of the brine sludge by placing it on the ground at the plant site. Olin also released mercury vapor into the air at a rate of one to two pounds per day.

A second plant at that site was built in 1950 to produce the pesticide benzene hexachloride ("BHC"). A byproduct of BHC production, which enriches gamma isomers to make a more potent pesticide, is degraded alpha and beta isomers. Olin disposed of these isomers on the ground at the plant site in what was known as the alpha beta pile. In 1956, an explosion destroyed that plant, and it was not rebuilt. The explosion released 65 tons of sulfuric acid, 6,500 gallons of benzene, two tons of sodium hydroxide, and a ton of soda ash. Firefighters hosed down the site for a couple days; some of the contamination was washed into adjoining Gill Creek, while the rest soaked into the ground and migrated into the ground water.

Olin started an investigation into the pollution at the plant site in 1978 and began conducting remediation in 1997, which was completed in 1998. Both the soil and ground water at the plant site were found to be contaminated with BHC, mercury, and other organic compounds. To address the soil contamination, Olin capped an area of the site with an asphalt cap to prevent precipitation from seeping into the ground and continuing to wash contaminants into the groundwater.2 To address the groundwater contamination, Olin pumped water out of the ground, treated it with an airstripper, and then sent the treated water to the City of Niagara Falls sewer system. The parties agree that this treatment did not remove any mercury; there was also some testimony that it did not remove BHC. The cost of remediating the contamination of the plant site was $5,027,722.

The second site involves property located at the intersection of Pine and Tuscarora streets near Niagara Falls. Olin deposited waste materials on the ground at this site for two months in 1957, including hexaclorobenzene and alpha-beta cake, byproducts of BHC manufacturing. This led to BHC contamination of the soil and groundwater at the site, as well as off-site groundwater and creek sediment. Beginning in 1989, Olin remediated this contamination by rerouting nearby Cayuga Creek, building a retaining wall or slurry wall to block the flow of groundwater, and constructing a cap over the disposal site to prevent rainwater from spreading the contaminants; remediation was completed in 1990. The cost of this remediation was $3,346,582.

The third site at issue, the Industrial Welding site, is a vacant section of land across the street from Olin's BHC plant. The parties dispute when disposal started at this site, but the jury found contamination began in 1950. Olin disposed of mercury-containing brine sludge, a byproduct of its chlorine production, on this site until 1966. This site was found to be contaminated with BHC and mercury. Beginning in 1998, Olin remediated this contamination by building a cap over the site and surrounding it with a clay wall to prevent the spread of contamination; remediation was completed in 2000. Olin also was required to dredge some of the sediment from a section of nearby Gill Creek to remove contamination and protect creatures living in the creek. The cost of remediating this site was $5,557,549.

The fourth site at issue is the mouth of Gill Creek, which runs near the plant site. Olin's operations released chemicals into the creek. In 1958, the State of New York, incident to constructing a parkway, built cofferdams that blocked the creek from flowing into the Niagara River and altered the location of the mouth of the creek. These cofferdams were removed in 1960. Olin's expert testified that the swift moving Niagara River would have carried away any contaminants except during the period when the cofferdams were in place. The contaminants found in the mouth of Gill Creek included BHC, mercury, and other organic compounds. Although LI contend that there was no significant source for mercury pollution until 1960, the jury found that property damage at the site was complete in 1960, implicitly rejecting that theory. In 1992, Olin and DuPont, which had also contributed to the contamination, split the cost of remediating the site, which involved removing the contaminated soil. Olin's portion of the costs amounted to $6,959,422.

Between 1950 and 1970, LI issued insurance policies providing Olin excess coverage. Because these were excess policies, they did not apply until Olin had exhausted its primary insurance coverage and its liability had reached the attachment point for the excess policies. From 1956 to 1970, the attachment point for each of the LI policies was $300,000 or greater; for a few of the earlier policies, it was slightly lower.

The specific policy language varies slightly among the policies issued in different years. Some of the earlier policies cover "accidental damage to or destruction of property" "in respect of accidents occurring during the [policy period]," and define accident as "an accident or series of accidents arising out of one event or occurrence." Other policies use similar language but substitute "occurrence" in place of "accident," and define occurrence as "one happening or series of happenings, arising out of, or due to one event taking place during the term of this policy." Later polices cover "Property Damage . . . caused by or arising out of each occurrence," and define occurrence as "an accident or a happening or event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally result in personal injury, property damage or advertising liability during the policy period."

While the language of the policies differs somewhat, all the polices are "occurrence policies," meaning that they cover events during the policy period, no matter how many years later the claim arises. Under New York law, such occurrence-based comprehensive general liability policies are "triggered by occurrence of the property damage during the policy period."3 Olin Corp. v. Ins. Co. of N. Am., 221 F.3d 307, 321 (2d Cir.2000) [hereinafter Olin I]. The parties agree that the policies cover property damage that occurred during the policy period, but they dispute what constitutes damage to property.

The policies also require Olin to give notice to LI of accidents or occurrences that could implicate the policies as soon as practicable after they occur. Olin notified LI of potential claims involving the Niagara Falls plant and Industrial Welding sites in February 1984. Olin notified LI of potential claims involving the mouth of Gill Creek in August 1988. The timing of the notice as to the Pine and Tuscarora site is not at issue.

Olin sought reimbursement for the remediation expenses associated with these sites from its primary and excess insurers between 1950 and 1970. Olin settled with the primary carriers, and only the excess claims went to trial. Although the parties agreed that the liability would be allocated over the years in which property damage occurred, they disputed whether property damage continued after the contamination reached a point at which the full remediation became necessary. The district court explicitly agreed with Olin that once the remediation became necessary, no further policy years would be implicated for reimbursement. However, the district court instructed the jury to decide both when property damage occurred, specifically explaining that continued groundwater contamination is not property damage, and to determine when the full remediation became necessary. The jury found as to the Niagara Falls plant site that damage occurred from 1950 to 1957, as to the Industrial Welding site, 1950 to 1966, as to the mouth of Gill Creek, 1958 to 1960, and as to the Pine and Tuscarora site, 1957. For each site, the jury found the full remedy became necessary in the final year of property damage. On the basis of these findings, the district court allocated Olin's damages as to each site over the years in which the jury found property damage occurred, and then entered judgment against LI for the amount in each year exceeding the policy attachment point. Despite the fact that there were other claims for remediation at other sites still to be determined, the district court certified under Federal Rule of Civil Procedure 54(b) that there is no just reason for delay in the entry of final judgment as to these claims.4 LI filed a motion for judgment as a matter of law and a new trial asserting that this method of allocation was incorrect, as well as other trial errors, which the district court denied. LI filed a timely notice of appeal.

In Olin I, 221 F.3d at 307, and Consolidated Edison Co. v. Allstate Insurance Co., 98 N.Y.2d 208, 746 N.Y.S.2d 622, 774 N.E.2d 687 (2002) [hereinafter Con Ed], both this Court and the New York Court of Appeals, respectively, made clear that when continuous property damage takes place over a number of policy periods, the liability for that injury is allocated over the time during which property damage occurred. Olin I, 221 F.3d at 324-25; Con Ed, 746 N.Y.S.2d at 629-30, 774 N.E.2d 687. In so holding, these courts rejected the proposition that the insurers faced joint and several liability that would allow the insured to select any triggered policy for compensation of the entire amount and leave it to the insurer to seek contribution. Olin I, 221 F.3d at 324; Con Ed, 746 N.Y.S.2d at 629-30, 774 N.E.2d 687.

In Olin I, the insured had been depositing dry and liquid pesticide components onto the soil as a result of its pesticide production, which in turn resulted in expansive soil and groundwater contamination. We held that "public policy and equitable considerations" "clearly indicate allocation" as the proper method for distributing liability. 221 F.3d at 324. We found allocation preferable because it is a more efficient apportionment of damages among those responsible for coverage than a separate contribution action, and would prevent the insured from receiving a windfall for uncovered damages—for example, those damages that were non-accidental or occurred during a period of self-insurance. Id. We, therefore, upheld a determination that property damages had occurred in each year of a thirty-year span, and we countenanced a pro-rata allocation of the indemnifiable costs that spanned that time. Id. at 327. In Con Ed, the New York Court of Appeals reached a similar conclusion, reasoning that the joint and several approach would be "inconsistent with the unambiguous language of the policies" at issue, while pro-rata allocation, although not mandated, would be consistent with that policy language. 746 N.Y.S.2d at 629-30, 774 N.E.2d 687.

Neither party here has argued that a joint and several model should be applied, but the parties do dispute the years over which damage should be allocated. While both Olin I and Con Ed upheld the use of pro-rata allocation of damages over each year in which property damage occurred, both also left open the possibility that other methods of allocation could be used. See Olin I, 221 F.3d at 325 (noting that "[a]llocation need not be based strictly on the number of years during which injury was incurred and the number of years that the insurers and the insured were on the risk," and suggesting, for example, that where the record gives some reason to believe more injury occurred during a particular year the allocation model could reflect that inequality); Con Ed, 746 N.Y.S.2d at 630, 774 N.E.2d 687 (observing that "this conclusion does not foreclose pro rata allocation among insurers by other methods"). For instance, if it could be determined exactly how much property damage occurred in each year, then the indemnifiable remediation costs could be allocated proportionately to the amount of damage, rather than equally among each year, and such a more accurate allocation would clearly be preferable. Regardless, a critical aspect of this allocation is determining in which policy periods property damage actually occurred, so we now turn to that question.

II. Jury charge on the definition of property damage

LI assert the district court improperly charged the jury on the definition of the term "property damage." As a result of the jury's application of that definition, they argue the jury improperly limited the number of years in which they could find "property damage" to have occurred. This error, in turn, caused the district court to allocate liability incorrectly because the number of years over which it pro-rated remediation expenses was too few.

As to the definition of property damages, the district court instructed the jury:

All I want to do by way of definition is to say that what is being talked about[] . . . is really additional property damage: Was there additional pollution? was there additional contamination? was there a spread of contamination? and during what years did such additional damage occur? We are not talking here about a static condition.

Let me pose this. Let's suppose that the ground underneath the . . . plant became polluted in a certain period of time. Let's suppose that pollution reached its maximum and the [plant] operation was over with, there was no more contaminant being put into the ground and the pollution reached its maximum. Let's suppose ground water pours past and is contaminated by this polluted ground.

That isn't additional damage. Obviously, every time ground water passes polluted soil, it is going to get some pollution. That is not what we are talking about. We are talking about a situation where there is additional property damage in the sense of additional contamination of a pollution-creating situation.

. . . If there is additional contamination being introduced to the ground so that the pollution of the ground waters is going to become more and more severe, that is additional damage. . . .

Simply a static condition in the terms that I have spoken of, if the ground through which the ground water runs is fully contaminated and no more contamination [occurs], the fact that ground water keeps passing through it, maybe it passes through it year after year after year, that is not additional damage in terms of this question. If sediment is fully contaminated and new fish go by and eat the contaminated substance, that isn't additional damage in terms of what we are talking about.

LI argue this charge improperly limited the jury by excluding from their consideration the ongoing damage to groundwater that continued after Olin stopped releasing contaminants into the soil and until Olin began remediation. As such, argue LI, because property damage, properly defined, did occur during years which could not be included under the court's charge, the allocation model used to distribute Olin's liability was necessarily flawed. Olin, in contrast, suggests the jury charge correctly embodied New York law on property damage. The charge, posits Olin, allowed the jury to limit its consideration to the time periods ending when "full remediation" plans were required to address the contamination sources already in place but no longer being actively augmented.

The insurance policies at issue here define "property damage" as the "loss of or direct damage to or destruction of tangible property (other than property owned by the Named Assured)." In the context of ongoing environmental contamination, "property damage" includes damage to or destruction of tangible property, but defining when such damage or destruction from contamination ceases5 also requires us to analyze the characteristics of the injury resulting from such contamination. We note at the outset that parties to a contract can define this aspect of property damage in the same manner in which they can define other contract terms. Nonetheless, where parties contest a term not defined in a contract, the court system must supply the answer. It is to that task we now turn.

LI assert that property damage includes not only injury caused by active polluting—the actual stockpiling of the contaminant—but also that injury caused by passive migrating contamination. LI maintain that migrating contamination, and thus the property damage, continued in this case until Olin actually began remediation. Under this theory, the per-year allocation of dollars to costs of remediation is at its lowest because the property damage, which includes the additional migrating contamination, necessarily occurs for a greater number of years. Olin's proposal, in contrast, advocates a definition of property damage that ceases at the time full remediation would be required to clean up the contamination that has been set in motion—usually, this is the point at which active pollution (generally meaning the unconfined stockpiling of contaminants) is no longer occurring. Olin's proposed definition, therefore, does not include the type of contamination caused by the migration of already existing pollutants from a point source into the adjacent soil and water. Because the time within which damages could be said to occur under Olin's so-called "full-remedy" model is significantly shortened, the per-year allocation of the cost to clean up those damages is much higher.

In their arguments to this Court, both parties refer less to property damage and more to the allocation of damages over time. What the parties are in fact arguing is how property damage should be defined under the insurance policies at issue so the fact finder can properly determine when it began and when it ceased. In other words, does the property damage at the four sites include ongoing passive contamination, through the leaching and migration of pollutants, of land and water that is already polluted from the same source; and does it include that same sort of passive contamination to previously uncontaminated land and water?

III. Definition of property damage in the context of environmental contamination

We review the jury instructions de novo, reversing only where they failed to inform or misled the jury regarding the applicable legal rule. See United States v. Ford, 435 F.3d 204, 209-10 (2d Cir.2006). Because, as discussed above, allocation must be over all years in which property damage occurred, if the district court's instruction was a correct statement of the law, then the district court properly allocated the costs of remediation over the years in which the jury found property damage occurred. However, because we cannot agree that the passive spread of contamination would not constitute new property damage, we cannot uphold the allocation used by the district court.

Both Con Ed and Olin I assumed continuing damage over a large number of years after active contamination ceased. However, despite LI's assertions that these cases are dispositive, neither case directly addressed the question of what constitutes "property damage" for the purposes of allocation in the context of migrating contamination. Although New York has not directly addressed whether migrating contamination constitutes additional property damage to trigger liability coverage, we are guided initially by the general tenet under New York law that where the precise meaning of insurance policies is ambiguous, their provisions are to be construed in favor of finding coverage. Handelsman v. Sea Ins. Co. Ltd., 85 N.Y.2d 96, 623 N.Y.S.2d 750, 751-52, 647 N.E.2d 1258 (1994); Ovando v. Hanover Delivery Serv. Inc., 13 A.D.3d 780, 786 N.Y.S.2d 608, 609 (N.Y.App.Div.2004); see also New York v. Blank, 27 F.3d 783, 788 (2d Cir.1994) (looking to New York law to interpret the insurance contract provisions).6

More pertinently, we find the forum courts' implicit assumption that migrating contamination is within the definition of property damage to be persuasive evidence that under the insurance policies at issue, New York would not limit damage merely to the placing of pollutants at a source where they will leach into the environment thus requiring full remediation. For instance, the New York Court of Appeals assumed in Con Ed that the continued contamination after the insured's active pollution constituted property damage. 746 N.Y.S.2d at 628, 774 N.E.2d 687. Additionally, in Long Island Lighting Co. v. Allianz Underwriters Ins. Co., 301 A.D.2d 23, 749 N.Y.S.2d 488, 495 (App. Div. 1st Dep't 2002) [hereinafter LILCO], the court denied liability for the migrating contamination "damage" of the insured's property based on the particular coverage language of the insurance policy that was triggered only when an event that occurred during the policy period caused the damage. In denying coverage under that particular circumstance, the court noted "[i]t would be illogical to deem the continuing migration of preexisting contaminants to be both the damage itself and the cause of the damage" thereby explicitly referencing continued migration of contamination as "damage." Id. at 495 (emphasis added). Partially basing its decision to preclude coverage on the fact that the migrating contamination was damage, as distinguished from the initial event causing the damage, LILCO supports our conclusion that property damage includes passive migrating contamination.

Further, we find additional cases, many of which Olin relies upon for support of its argument, actually support the contrary position and suggest passive migrating contamination is property damage. Olin relies on Stonewall Insurance Co. v. Asbestos Claims Management Corp., 73 F.3d 1178 (2d Cir.1995) and Maryland Casualty Co. v. W.R. Grace & Co., 23 F.3d 617 (2d Cir.1993), which both held that the property damage to a building containing asbestos occurs and is complete at the time of installation. See Stonewall, 73 F.3d at 1210; W.R. Grace, 23 F.3d at 628. In referencing the surcease of damage, these cases repeatedly distinguish the injury caused by the leaking and migration of hazardous waste at issue here from other types of injury. The Stonewall court found asbestos damage complete at installation, but held that where there are successive injuries, as is the case with progressive diseases like asbestosis, successive policies may be triggered. 73 F.3d at 1195-96. Similarly, in W.R. Grace, where we held, citing New York law, that asbestos damage is complete at installation, we explicitly contrasted the installation of asbestos with more gradual damage, such as the leaking of hazardous waste or the progression of cancer. See id. at 627-628 ("[A]sbestos property damage is unlike the gradual leaking of hazardous waste from a landfill or the gradual cracking and shifting of a building's foundation."). Thus, these cases support a definition of property damage that includes continuing migration of chemicals in the groundwater

The remaining cases on which Olin relies, Owens-Illinois, Inc. v. United Insurance Co., 138 N.J. 437, 650 A.2d 974 (1994) and E.R. Squibb & Sons, Inc. v. Lloyd's & Cos., 241 F.3d 154 (2d Cir.2001), are of no greater help. While these cases extended the responsibility of insurers, for differing reasons, to damage or injuries beyond the policy period, neither established the converse—that those injuries would not trigger the policies for the years in which the injuries occurred as well. See Owens-Illinois, 650 A.2d at 988-89; E.R. Squibb, 241 F.3d at 169. In fact, Owens-Illinois explicitly left open the possibility that damages that develop over time from long-term exposure could be allocated among multiple policies in effect at the time of the damage. See 650 A.2d at 989 ("[T]he argument that all sums to be assessed because of long-term exposure to asbestos could have been established in any one of the policy years is intuitively suspect and inconsistent with our developing jurisprudence in the field of toxic torts."). Similarly, E.R. Squibb, while extending liability beyond injuries during the policy period to additional injuries that were the consequence of injuries during the policy period, did not purport to limit recovery from later policies, which were not at issue. See 241 F.3d at 169.

Furthermore, Olin's "full remedy model" is problematic because under this theory, damage to property caused by continued dumping of contaminants or migration of contamination, could easily continue after full remediation is required and yet not be considered damage that would trigger a policy or add to the allocation period. Such a damages model is illogical and not supported by the case law.

LI's preferred allocation model is not entirely enunciated. They argue in this case that evidence showed the property damage continued until Olin began to remediate. Olin characterizes LI's allocation approach as one based on a definition of damages that assumes injury to the property, and thus property damage, necessarily continues until remediation actually commences. To the extent LI posit such a model, we note that this model is problematic because it does not contemplate an end to the damage accruing to the environment and implicitly assumes that contamination continues at a constant rate. These assumptions are untenable. At some point, perhaps many years after active contamination ends, the pollution must reach its maximum point. We doubt that it will continue to spread indefinitely. Moreover, if property damage necessarily continues until remediation, then the amount of coverage under each policy will change depending on when the remediation occurs, again making coverage dependent on events occurring after the policy period.

We adopt an intermediate approach, and hold that property damage occurs as long as contamination continues to increase or spread, whether or not the contamination is based on active pollution or the passive migration of contamination into the soil and groundwater. This appears to comport with the majority view that passive contamination migration does constitute property damage. See Tod Zuckerman & Mark Raskoff, Environmental Insurance Litigation: Law and Practice § 4:4. This definition is within the parameters of New York case law on the issue, which clearly has assumed as a matter of law that continued passive contamination after active pollution can constitute property damage. See, e.g., Con Ed, 746 N.Y.S.2d at 628, 774 N.E.2d 687; LILCO, 749 N.Y.S.2d at 495. Moreover, we uphold the guiding principle that ambiguities in contracts should generally be construed in favor of finding coverage. Handelsman, 623 N.Y.S.2d at 751-52, 647 N.E.2d 1258. We acknowledge that, without expert testimony, it may be difficult for judges or juries to understand when contamination increases or spreads. For instance, we think it is possible for previously clean groundwater to migrate through contaminated soil and flow on to contaminate new areas. On the other hand, we also think it might be possible that newly contaminated groundwater migrating through contaminated soil will not increase or spread the total extent of contamination so, notwithstanding the passage of time, no additional property damage results. However, if that newly contaminated water does spread the contamination to new areas, property damage has clearly occurred.7

Because the district court's jury instruction did not allow the jury to consider as property damage the passive spread of contamination up to the point the contamination ceased to increase or spread, we find that this instruction was erroneous. In order to properly allocate the cost of remediation, a jury must determine in what years property damage, as defined above, actually occurred. On appeal, LI have argued that this Court should order judgment as a matter of law in its favor. We conclude in light of the erroneous jury instruction, and the holding herein, that the period in which property damage occurred appears to be an open question for a factfinder, which should be decided in the district court in the first instance. Nadel v. Isaksson, 321 F.3d 266, 271-72 (2d Cir.2003) (noting entry of judgment as a matter of law is proper only where there can be but one conclusion as to the verdict that reasonable persons could have reached); see also Atl. States Legal Found., Inc. v. Pan Am. Tanning Corp., 993 F.2d 1017, 1022 (2d Cir.1993) (remanding after reversal for district court to decide summary judgment motion and noting "sound practice" of allowing district court to rule). Thus, although a new trial is almost certainly required, we defer to the district court's greater familiarity with the voluminous record in this case as to whether Olin has introduced any evidence by which a rational juror could conclude property damage, including the passive spread of contaminants, ended before remediation.

LI argue that they should be permitted to raise the defense of late notice on remand. Because the insurance policies at issue require notice of an occurrence or accident that may give rise to coverage as soon as practicable, such notice is a prerequisite to coverage. The district court ruled that prejudice was an element of the late-notice defense and that it would so instruct the jury. LI had no evidence of prejudice, and therefore decided not to submit the late notice defense to the jury. However, subsequent cases have made clear that under New York law, prejudice is not an element of the late notice defense. See Argo Corp. v. Greater N.Y. Mut. Ins. Co., 4 N.Y.3d 332, 794 N.Y.S.2d 704, 706-07, 827 N.E.2d 762 (2005); Great Canal Realty Corp. v. Seneca Ins. Co., 5 N.Y.3d 742, 800 N.Y.S.2d 521, 522, 833 N.E.2d 1196 (2005). Olin does not dispute that the district court erred in holding that LI had to show prejudice to prevail on the late notice defense; instead, Olin raises two arguments that the defense has been waived. We hold that LI should be permitted to raise this defense on remand.

First, Olin argues that by voluntarily withdrawing the defense and not submitting the underlying factual questions to the jury LI waived the late notice defense. Generally, specifically stating an objection to the court's ruling is sufficient to preserve that issue for appeal. Fed. R.Civ.P. 46. Although failure to submit a defense to the jury may waive that defense, see Kerman v. City of New York, 374 F.3d 93, 118 (2d Cir.2004), here, once the district court ruled that prejudice was an element of the late notice defense, LI had no chance of prevailing on the defense because they had presented no evidence of prejudice. Furthermore, LI explicitly stated that they intended to preserve the objection, and the district court suggested that the parties would be able to proceed on the record of this issue for the appeal. Because the parties and the district court had initially agreed that the underlying factual questions would be submitted to avoid a new trial in case of reversal on appeal, it is unclear why LI changed course. However, Olin should not be encouraged to sandbag by sitting silently while LI explicitly say they are preserving the issue, and then on appeal claim it has not been preserved. We therefore find that this issue has been adequately preserved. Because it is clear the district court incorrectly required LI to show prejudice, on remand they are entitled to a determination of their late notice defense under the correct standard.

Second, Olin argues that the defense was waived because LI failed to assert it in their letters denying coverage. It is true that "an insurer may be barred from raising defenses not asserted in communications to the insured denying coverage." Lauder v. First Unum Life Ins. Co., 284 F.3d 375, 380 (2d Cir.2002) (internal quotation marks omitted). However, the parties dispute whether the correspondence between them raised only specific defenses or reserved other defenses as well. We express no opinion on this argument and leave it to the district court to resolve this factual dispute in the first instance on remand. See Coogan v. Smyers, 134 F.3d 479, 486-87 (2d Cir.1998) (noting that appellate court will not consider an issue raised for the first time on appeal unless there is no need for additional fact-finding).

LI raise several other purported trial errors. Although these issues are, for purposes of this appeal, mooted by our disposition, we address them briefly because they may recur on remand. See, e.g., LNC Invs., Inc. v. First Fidelity Bank, 173 F.3d 454, 464 (2d Cir.1999). We find each to be without merit.

Olin offered expert testimony from Frank Rovers, the head of an environmental civil engineering firm, the designer of the remedial program for the plant site, and a monitor of the Pine and Tuscarora site. LI argue that Rovers's testimony should not have been admitted because it was not sufficiently reliable. We review the district court's decision to admit expert testimony for abuse of discretion. See Gen. Elec. Co. v. Joiner, 522 U.S. 136, 141, 118 S. Ct. 512, 139 L. Ed. 2d 508 (1997). We will not reverse such a decision unless it is manifestly erroneous. Id. at 142, 118 S. Ct. 512.

The district court's inquiry into the reliability of expert testimony is flexible. See Amorgianos v. Nat'l R.R. Passenger Corp., 303 F.3d 256, 266 (2d Cir.2002). The trial judge has great latitude not only in deciding whether an expert's testimony is reliable, but in deciding how to make that inquiry. See Kumho Tire Co. v. Carmichael, 526 U.S. 137, 152, 119 S. Ct. 1167, 143 L. Ed. 2d 238 (1999). However, "the district court must determine whether the proffered testimony has a sufficiently reliable foundation to permit it to be considered," and "when an expert opinion is based on data, a methodology, or studies that are simply inadequate to support the conclusions reached, Daubert [v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993)] and Rule 702 mandate the exclusion of that unreliable opinion testimony." Amorgianos, 303 F.3d at 265-66 (internal quotation marks omitted).

Here, the district court found Rovers to be extremely well-qualified, a conclusion that LI have not contested. In fact, Rovers testified that he specializes in hydrogeology, the study of how groundwater moves through soil, and that he focused his masters degree work on disposal of waste on land and the fate and transport of chemicals, in other words the migration of water through contaminants and the impact on the environment, precisely the issue in this case.

The district court did suggest that some of Rovers's opinions were not particularly well-supported; specifically, with respect to Rovers's opinion about the Industrial Welding site, the judge commented, "I have no idea the basis for that." Nevertheless, the district court found other portions of Rovers's testimony helpful and decided that the problems went to the weight, rather than admissibility of the testimony. This determination was not an abuse of discretion. There was some evidence in the record to support all of Rovers's conclusions as to the date remediation was required, with the possible exception of the Industrial Welding site. Rather than selectively excluding the portions of Rovers's testimony that the court thought were not well supported, the judge considered the testimony as a whole. This was not error.

The district court was aware that the weaknesses in Rovers's opinion would be pointed out to the jury by LI's cross-examination and experts, and that such a technique is an appropriate way of attacking weak expert testimony, rather than complete exclusion. See Amorgianos, 303 F.3d at 267. Furthermore, in this case LI's attack was clearly effective because the jury rejected the portion of Rovers's testimony related to the Industrial Welding site. Therefore, we hold that the district court did not abuse its discretion in admitting Rovers's testimony.8

LI complain that the judge instructed the jury that they could consider what the predominant chemical at a site was in determining whether the contamination was accidental. The judge discussed briefly the primary sources of contamination at each of the sites, a discussion that LI argue focused too much on BHC to the exclusion of mercury. Our review of the instruction indicates that the judge did not at all limit the ability of the jury to consider mercury contamination. The judge's suggestions about the predominant chemicals did mention BHC, but also mentioned the mercury-containing brine sludge. Furthermore, after this discussion the judge stated that:

[W]hat I am talking to you about now is a suggestion as to how you can approach your job, but you've heard all the evidence. If I mention a particular chemical or I mention a particular source, and in your deliberations you say well, I think there was also this important source, you're the jury . . . . what those predominant chemicals are and what the predominant sources are, if you wish to go that route, it is up to you to decide.

Finally, when the jury sent a note asking "Can we get clarification on the Mercury consideration? We were told that Mercury isn't under consideration in the verdict," the judge responded that:

I had no intention of saying that Mercury was not to be considered . . . . I did mention certain, certain chemicals I think in trying to illustrate certain types of issues about accident, but I did not intend to either direct your attention to what were the relevant chemicals or to say that any chemicals are not relevant. That is the issue that is up to you. It is up to you, and I didn't intend to have it any other way.

Looking at these instructions as a whole, there is no basis to believe the jury would have been confused about its ability to consider the mercury contamination. Therefore, LI's argument is meritless.

C. Exclusion of evidence of intentional pollution

LI further argue that the district court should not have excluded a 1979 misdemeanor conviction of Olin and three employees for knowingly polluting the environment in violation of law, disposing of impermissible amounts of mercury into the Niagara River in the 1970s, and knowingly submitting false reports to the government that understated Olin's mercury releases. We review the district court's decision to exclude this evidence only for an abuse of discretion. See United States v. Payton, 159 F.3d 49, 56-57 (2d Cir.1998).

Under Federal Rule of Evidence 609, a witness may be impeached with evidence of a prior conviction that involves dishonesty or false statement, as a conviction for submitting false reports clearly does. Fed.R.Evid. 609(a). LI cite to Stone v. C.R. Bard, Inc., No. 02 Civ. 3433, 2003 WL 22902564, *3 (S.D.N.Y. Dec. 8, 2003), for the proposition that a corporate representative may similarly be impeached with convictions of a corporation. See id. However, even assuming that is correct, generally such convictions are not admissible if they are more than ten years old, unless the probative value substantially outweighs the prejudicial impact. Fed. R.Evid. 609(b). The 1979 conviction was obviously outside this time frame.

Although the conviction may have been probative to impeach the statement of corporate representatives that Olin would never intentionally pollute, the 1979 conviction dealt with a different time period than much of the initial pollution at issue in the case, which occurred in the 1950s and 1960s, and may have been highly prejudicial to Olin in suggesting that the contamination at issue was intentional. Thus, it was entirely reasonable for the district court to conclude the probative value does not substantially outweigh the prejudicial impact, and it was not an abuse of discretion for the district court to exclude evidence of the conviction.

For the foregoing reasons, the judgment of the district court is vacated, and the case is remanded to the district court for further proceedings consistent with this opinion.