The division of damages for large, long-term environmental claims is a matter of considerable debate and great interest; at least it is in the rarified and blessedly small arena of insurance litigation. This topic seized the Supreme Judicial Court’s attention in the recent case of Boston Gas Company v. Century Indemnity Company, 454 Mass. 337 (2009). Justice Cordy’s thoughtful decision set forth the general method of apportioning damages between insurers and insureds in these rather difficult disputes where the timing and extent of environmental damage are difficult to determine. Applying the Court’s general rules to the factual specifics of a case is a trial judge’s job. After all, this filling of the interstices of a rule by way of the accretion of trial court decisions is the stuff of common law. Therefore, one can liken the following to Oliver Wendell Holmes Sr.’s nautilus slowly building another chamber in its spiral shell. Unlike the nautilus’s efforts in the poem, this decision will not honor some higher being, but it hopefully will produce a just result.
The jury returned a verdict in favor of the plaintiff insured in the amount of $120,276. In so doing, the juiy found that the insurer had breached both its duty to defend and its duty to indemnify. The damages returned related only to the issue of indemnification. The jury found that the defendant had failed to prove that a significant portion of the pollution at a Superfund site was not the result of a “sudden and accidental” discharge. The verdict figure equaled the amount earlier paid by the insured to satisfy an EPA demand for a de minimus settlement regarding pollution at this Superfund site.
The Supreme Judicial Court has adopted a “time on the risk” pro rata method of damages for so-called “long-tail claims” resulting from long-term or progressive environmental damage. The SJC determined that the most “probable fiction” and the fairest assessment of damages would be to pro-rate the damages by determining the “time on the risk” divided by the years during the “triggered period.” The Court was not asked to define what triggered an insurance policy or over what total period of time the damages should be spread out. The decision does note that trial courts *579have the flexibility to alter any apportionment of damages if the facts of the case permit a fairer apportionment than the numerator/denominator approach of Boston Gas. The fairest method, as noted by the Court, would be a “fact-based allocation” of damages, but that is often impossible in certain environmental damage claims that arise many years after the “triggering event.”
Transposed on this pro rata damages construct is the strict liability theory of CERCLA. Under that federal law, any generator of waste to a Superfund site is theoretically liable for the entire pollution at the site. See Highlands Insurance Company v. Aerovox, Inc., 424 Mass. 226, 234 (1997). It is important, however, to separate out the two distinct policies at work. Boston Gas attempts to develop a damages rule that most fairly apportions risk between different insurers (and the policyholder during periods of under-insurance or no insurance). CERCLA, on the other hand, is a federal law that represents a societal decision to shift toxic clean-up costs to all generators of waste at a particular site, making those generators jointly and severally liable for the entire amount of pollution damage.
The facts of this case emphasize the difference between the two concepts. The Superfund site in this case began receiving waste in 1924 and continued to do so until 1996 when it was taken over by EPA. When spills, leaks, and other pollution began at the site is unknown, and perhaps unknowable. There is evidence that the site owners used an open lagoon in approximately 1965 to store waste oil and this probably produced ground water pollution relatively soon after the lagoon began operations.1 Given the multiple sources of pollution and the numerous plumes of pollution that permeated the groundwater, pollution undoubtedly came from numerous point sources and began at some indefinite time long ago. Even though the plaintiff is financially responsible for this early pollution under CERCLA, none of this pollution related to the plaintiff. The plaintiff did not even exist until sometime in the mid-1960s and never shipped a single drop of waste oil to the site until 1982. It is the 1982 shipment of waste oil to the Beede Waste Oil site in Plastow, New Hampshire that “triggered” the Commercial Union insurance policy. Commercial Union, which apparently insured the plaintiff in the 1970s, is not claiming that any of its earlier policies were triggered by the discovery of pollution at the Beede site.
The plaintiff was one of hundreds of generators who lawfully shipped waste oil to the site. The plaintiffs shipments covered a ten-year period, 1982 to 1992. From sometime in the 1970s until 1997, the plaintiff purchased general comprehensive liability policies from American Employers, a company doing business as Commercial Union. These policies contained coverages for certain “occurrences” of bodily injury or property damage. During at least 1981 through 1987, the pollution exclusion in the policies contained a “sudden and accidental” exception. Sometime in 1990, the policy was amended to add an absolute pollution exclusion.
The EPA sent a PRP letter to plaintiff in June 2001 and this was forwarded to the insurer in March 2002. The site cleanup is currently ongoing and it is unknown when it will end. It is also unknown if the cleanup will actually cease the environmental damage.
The first issue is the date that “time on the risk” begins. Under the Boston Gas decision, one must determine the “triggered” policies that had “time on the risk” (the numerator) and divide that by the “total number of years of triggered coverage” (the denominator). So one first figures out the numerator. The beginning of the numerator period is not in substantial dispute. Both sides agree that it is the 1981-1983 policy that was initially “triggered” by the PRP letter which announced that EPA had found groundwater pollution in a neighboring well in 1983. The triggering event appears to be the 1982 shipments from the insured to the Beede site. This is relatively self evident as the insured would have no CERCLA liability had it not begun shipping to the site in 1982.
Where the “time on the risk” period (the numerator) ends is a bit more problematic due to a failure of evidence. Both sides now agree that the succeeding policy (1984-1987) is triggered by the ongoing shipments and ongoing pollution.2 Nor is there disagreement that the succeeding policies are “triggered” (i.e. those Commercial Union policies from 1988 through at least 1992). From July 1990 onward, both sides agree that a so-called “absolute” or broad-based pollution exclusion was contained in the policies. Despite lengthy arguments by the plaintiff that the post-July 1990 pollution exclusions might possibly allow coverage, it is clear that the post July 1990 policies preclude any pollution coverage for this claim. Thus the “time on the risk” definitively ends July 1990.
What is at issue is whether the 1987-July 1990 policy contained an “absolute” pollution exclusion or the “sudden and accidental” language of the earlier policies.3 The person handling the insurance claim for Commercial Union, Deborah Dyer, confirmed in letters that the “sudden and accidental” provision existed in the plaintiffs policies until July 1990. See Exhibits 28 and 40. The defendant is now in the uncomfortable position of disputing its agent’s representation and Commercial Union claims that Dyer was mistaken and that the standard practice for policies written in the July 1987 to 1990 period was to include an “absolute” pollution exclusion. See Affidavit of Edward C. Albanese. One cannot be certain, however, because no one can find the policy in question and Commercial Union cannot find the underwriting file for the policy. See Affidavits of Albanese and David R. Young. Given the delay of time before the policy became relevant, it is understandable why this small company and its for*580mer insurance agent would not retain a copy of a fourteen-year-old policy, but it is less understandable why the insurer would not retain at least the underwriting file. Apparently, Ms. Dyer (whose representations are binding on the principal) had evidence of coverage before she made the assertions in her 2004 letters. Mr. Albanese’s assurances that Commercial Union, as a general matter, began using the “absolute” pollution exclusions in its policies as of November 1985 (either by way of policy language or endorsements) does not shed any light in this case due to the undoubted fact that the plaintiffs 1984-1987 policy was not altered by an endorsement and the policy definitively does not contain the “absolute” exclusion. More importantly, the evidence submitted by .the parties indicates that the Commercial Union “absolute” pollution exclusion (drafted in 1988) was not approved by the Massachusetts insurance commissioner until 1990. See paragraph 5 of Young affidavit and Exhibit E to plaintiffs Memorandum Re: Issue of Allocation of Damages. In this situation, where there is evidence of coverage, the insurer bears the burden of contradicting that evidence and either producing the policy or some similarly persuasive evidence. Commercial Union has not adequately done that in this case. Therefore, the “sudden and accidental" coverage (or the “time on the risk”) extends to July 1990. Thus, the “numerator” is approximately April 1982 (month of first shipment) to July 1990 or eight years.
That brings one to the more difficult determination of the “total number of years of triggered coverage,” i.e. the total years the risk/damage/claim existed regardless of who was insuring the plaintiff. This figure is the “denominator” for purposes of the pro rata calculation.
The insurer suggests that the years of triggered coverage begins when pollution first occurred at the Beede site. At trial, Commercial Union, somewhat randomly, picked the year 1965 and cited the creation of the lagoon in the mid-1960s. Now, using the date of a photograph of the same lagoon, Commercial Union suggests the date of 1970.4 This approach is equivalent to randomly picking a year out of a hat as the start date due to the overwhelming evidence of long-term pollution at the Beede site for most of the twentieth centuiy. Such an approach does not achieve the purposes of the Boston Gas pro rata approach to promote “judicial efficiency, engender stability and predictability in the insurance market, provide incentive for responsible commercial behavior, and produce an equitable result.” Id. at 366. Far from being predictable and equitable, this approach encourages speculation, unfairness to a company that did not even exist before 1965, and a windfall to the insurance company. Likewise, this approach ignores a matter that the SJC focused upon in Boston Gas, i.e. the insurer should only be responsible for “occurrences” under the particular policy. The much fairer approach is to begin the triggered coverage period when the first “occurrence” actually triggers the liability, i.e. the 1982 shipment of waste to the Beede site. Commercial Union’s argument that this approach does not take into account the all encompassing liability of CERCLA (i.e. all generators are jointly and severally liable for the entire clean up cost — a sum that exceeded 71 million dollars in this case) is not persuasive. This is particularly true in this case because the pollution damages paid by the insured to EPA under CERCLA were not all encompassing. Instead, the damages were limited by the de minimus settlement amount ($120,276) which took into account the limited amount of waste oil shipped to the site by the insured over a relatively brief period of time. Because the damages are lessened by the insured’s limited involvement with the site, the insurer should not reap the advantage of stretching the “denominator” to some exponential number.
If 1982 is the start of the triggered period of years, when does the period end? The options are numerous. One could argue that the period ends when the pollution ends. But that future date is impossible to even estimate. Indeed, given the extent of the pollution and its infiltration into the ground water, it may be an infinite date. Commercial Union suggests the date of the jury verdict, but that is plainly a random date that does nothing to ensure predictability or fairness. The dates of the PRP letter5 or notice to the insurer also seem equally random as they have nothing to do with the triggered period or the time of damage/risk. Another possibility is the date the EPA took over the site (1996). This date has some advantage for once EPA controls the site, the clean-up is beyond the control of the insured. But, in the circumstances of this case, the clean-up of a site owned by a third party in another state was always beyond the knowledge and control of the insured. The best date appears to be when the insured had its last contact with the polluted site. In this case, that would be the date of the last shipment, i.e. 1992. This approach best approximates the Boston Gas case which assumed that the triggered period of years ended when Boston Gas sold or abandoned the polluted site. Likewise, it best approximates the “fact-based allocation” praised in Boston Gas. In this case, one knows that the insured did nothing to further pollution of the site after the last shipment in 1992. Therefore, given the facts of this case, the fairest “triggered period of years” is from 1982 to 1992, or ten years.
Due to the absolute pollution exclusion in the July 1990 policy, the insured had no coverage for the Beede site pollution for two years of the “triggered period.” Given these determinations, the fairest pro rata equation in this case is eight years over ten years. In other words, Commercial Union is responsible for paying 8/10ths (or 80%) of the juiy’s verdict. If the court’s mathematical skills are not too rusty, this equates to a damage award of $96,220.80. Interest is to run on that portion of the verdict from the date of breach. The date of breach was the deadline date set by EPA for *581the settlement payment, August 4, 2004 (the day the insured made the payment). The final judgment shall be amended to reflect this ruling.
During trial, defendant’s counsel argued that 1965 was the fairest date to choose for the start date of pollution damage based on the rather general testimony of its expert. In its post-trial submissions, the defendant now claims a date of 1970 because this is the year of the photograph of the lagoon. The shifting date exemplifies the arbitrary and speculative nature of defendant’s approach to a start date.
Initially Commercial Union’s representative took the position that only the 1981-1983 policy was “potentially applicable” and that “policies in effect 1/1/1984 and subsequent would not apply to this claim...” Ex. 28 (July 15, 2004 letter of Deborah Dyer). Apparently cooler and more intelligent heads have prevailed and the insurer now concedes that at least some of the later policies are “triggered.”
In its post-trial submissions, the defendant un-persuasively argues that the plaintiff already has stipulated that only two policies are at issue when determining Commercial Union’s “time on the risk” (I.e. the 1981 and 1984 policies). The defendant cites a factual stipulation contained in the parties’ joint pre-trial memorandum. The pre-trial memorandum was submitted before the Boston Gas case had been decided by the Supreme Judicial Court. Understandably, given the recent vintage of the Boston Gas decision, the defendant did not raise the issue of a Boston Gas-style pro-rata division of damages until the eve of trial. The court does not accept plaintiffs argument that the defendant’s late disclosure of this “affirmative defense” should estop the defendant from raising the issue. Still, once the Boston Gas issue was raised, the relevance of other “triggered” policies became very relevant. In practical fact, the total number of “triggered” policies only became relevant after the jury returned its verdict finding that the defendant had breached its duly to indemnify and awarding damages equal to the full EPA settlement paid by the insured. The defendant’s argument that the plaintiff is somehow bound by the much earlier stipulation regarding the applicability of two policies (which was relevant only in determining liability, not in proportioning the damages) is particularly disingenuous.
Even if Commercial Union was correct in using the start date of 1970, the evidence indicates that Commercial Union insured the plaintiff during the 1970s. What little evidence exists suggests that the policies in the 1970s did not contain “absolute” pollution exclusions. Thus, even accepting Commercial Union’s argument, the insurer’s proportional liability would not change.
The court initially indicated at trial that the PRP letter might be the best ending point because it represented when the EPA took control of the site. Upon reflection, this proposed solution was wrong. First, the date of the PRP letter, particularly in this case, does not remotely coincide with the date of the EPA takeover of the site. Second, the purposes of fairness and predictability are not served in any way by the rather random date that the EPA decided to take over a polluted site or send a PRP letter.