RICHLI, J.

In this action, the State of California (the State) seeks to recover from its liability insurers the amounts that a federal court has ordered it to pay for the cleanup of the Stringfellow hazardous waste site. Some insurers were granted summary judgment; the propriety of that ruling is currently before the California Supreme Court in State of California v. Underwriters at Lloyd's London (2006) 146 Cal.App.4th 851 [ 54 Cal.Rptr.3d 343 ], review granted April 18, 2007, S149988. Other insurers settled with the State.

By the time the trial court entered the judgment that is the subject of this appeal, there were only six insurers left standing: Continental Insurance Company (Continental), Continental Casualty Company (Casualty), Employers Insurance of Wausau (Wausau), Horace Mann Insurance Company (Horace Mann), Stonebridge Life Insurance Company (Stonebridge), and Yosemite Insurance Company (Yosemite) (collectively the Insurers). Each of them had issued to the State an excess corporate general liability policy covering a two- or three-year policy period.

The trial court ruled that every policy in effect for any policy period during which the loss was occurring covered the entire loss which was at least $50 million, and could be as much as $700 million subject to the policy limits. However, it also ruled that the State could not recover more than the total policy limits for any one policy period; this effectively limited the State's recovery to $48 million. Finally, it ruled that the Insurers were entitled to a setoff for settlement amounts previously paid by other insurers. Because the State had already recovered approximately $120 million in settlements, the trial court entered a judgment awarding the State "$0" against the Insurers.

The State has appealed; the Insurers (other than Wausau) have filed a protective cross-appeal.

In the end, we will uphold (or find moot) all of the trial court's rulings, with two exceptions: The trial court did err by (1) ruling that the State could not recover more than the total policy limits in effect for any one policy period, and (2) admitting certain documents under the ancient documents exception to the hearsay rule (Evid. Code, § 1331). Accordingly, we must reverse and remand for further proceedings.

We hasten to add that we do not fault the trial court in any way. Each of the successive judges who have handled the case since it was first filed, way back in 1993, has done yeoman's service. In particular, Judge Erik Michael Kaiser (now retired), who handled the case throughout its final stages, including the jury trial, did an outstanding job of organizing, managing, and ultimately adjudicating this complex case.

Judge Kaiser ruled that the State could recover for only one policy period because he believed that he was bound to follow FMC Corp. v. Plaisted & Companies (1998) 61 Cal.App.4th 1132 [ 72 Cal.Rptr.2d 467 ] ( FMC ), which was the closest case on point. As an appellate court, however, we can and do respectfully disagree with FMC. It failed to follow other, closely analogous California cases, based on reasoning that we find to be flawed and unconvincing.

Similarly, in ruling on the ancient documents exception, Judge Kaiser entered uncharted territory, as this exception has not been the subject of an appellate opinion since it became effective, along with the rest of the Evidence Code, in 1967. We will construe it for the first time.

J.B. Stringfellow, Jr., owned a quarry near Glen Avon in Riverside County. In 1955, a state geologist inspected the quarry to determine whether it was suitable for use as an industrial waste disposal site. He reported that the site lay in a canyon, underlain by impermeable rock. He recommended that a concrete barrier dam be built to close a 250-foot gap in the canyon's natural walls. He concluded that, once such a dam was built, "the operation of the site for industrial wastes will not constitute a threat of pollution . . . ."

The State therefore proceeded to design the site and to supervise its construction. The site went into operation in 1956. More than 30 million gallons of industrial waste were deposited into unlined ponds at the site.

Actually, the site was badly flawed. First, an underground stream channel lay about 70 feet below the surface; it carried groundwater into and out of the site. Second, the underlying rock was fractured; contaminants could leak down through it and reach the groundwater. Third, the barrier dam was inadequate; it allowed contaminants to escape.

In 1969, heavy rains caused contaminants to overflow the dam. In 1972, groundwater contamination was discovered, and the site was closed. However, it continued to leak. In 1978, heavy rains once again made the ponds overflow; the State decided to allow a "controlled discharge" of contaminants into Pyrite Channel. Hazardous waste released from the site merged into a plume that ultimately extended miles away.

In 1983, the United States and the State filed suit against numerous defendants, including companies that had deposited waste at the site, as well as the hapless Mr. Stringfellow, alleging that they were liable for the resulting contamination. Certain defendants counterclaimed against the State.

In September 1998, the federal court found the State liable for, among other things, negligence in investigating the site, choosing the site, designing the site, supervising construction of the site, failing to remedy conditions at the site, and delaying the cleanup of the site. The State was held liable for all past and future remediation costs, which the State claims could be as much as $700 million.

The Insurers stipulated that the State was liable for at least $50 million.

Each of the Insurers (or their predecessors in interest) had issued one or more excess liability policies to the State, covering a multiyear policy period, as follows:

The State had drafted a master liability policy form, which it required its insurers to use. However, many provisions of the form used language that was standard in the industry. It is undisputed that the relevant language of each of the Insurers' policies was essentially the same, as follows:

In September 1993, the State filed an action against five named insurers, seeking indemnity for its liability in the underlying federal action.

The trial court ordered the case tried in a series of phases. On June 10, 1999, following a bench trial, the trial court (per Judge Cunnison) entered its statement of decision regarding phase II. It ruled, among other things, that the policy limits under policies with a multiyear policy period applied per occurrence, not annually (no-annualization ruling).

In April 2002, the trial court (per Judge Waters) ruled that the State's negligence in failing to remediate and delaying remediation at the site did not breach any duty to mitigate the defendant insurers' damages (no-mitigation ruling).

In September 2002, the State filed a second action, asserting similar claims against additional insurers, including the six that are parties to this appeal. In October 2003, the trial court consolidated the two actions. The defendants in the second action agreed to be bound by all previous rulings in the first action.

In November 2003, the case was assigned to Judge Kaiser.

The State and the defendant insurers stipulated that third party property damage resulting from the selection, design and construction of the site occurred continuously throughout all of the relevant policy periods.

In March 2004, the trial court ruled that each of the defendant insurers was potentially liable for the total amount of the loss (subject to their policy limits), rejecting their contention that they could be liable only for the portion of the loss attributable to their own policy periods (all-sums ruling). At the same time, however, it also ruled that the State could not recover the policy limits in effect for every policy period. Instead, the State had to choose one policy period, and it could recover only up to the policy limits of the policies in effect during that period (no-stacking ruling).

In February 2005, the trial court ruled that, for purposes of policy limits, there had been only a single occurrence, rejecting the State's contention that there had been as many as five occurrences (one-occurrence ruling). [2]

On March 28, 2005, a jury trial on phase III began. On May 16, 2005, the jury rendered special verdicts, finding, among other things, that the Insurers had breached their respective policies. At that point, the State had already entered into settlements with other insurers totaling approximately $120 million. The trial court ruled that these settlement amounts had to be set off against the Insurers' liability (setoff ruling). Under the trial court's one-occurrence, no-annualization and no-stacking rulings, the most the State could recover was $48 million. Accordingly, the trial court entered judgment nominally in favor of the State, but in the amount of "$0."

The State filed a timely notice of appeal. Except for Wausau, all of the Insurers filed timely notices of cross-appeal.

The State contends that the trial court erred by limiting it to the policy limits in effect for any one policy period.

In their protective cross-appeal, the Insurers [3] contend that the trial court erred by ruling that they could be liable for property damage that occurred outside their respective policy periods.

Because these contentions are related, and because the trial court ruled on them both at the same time, we consider them seriatim. For clarity, however, we address them in the reverse order.

In phase III, the parties stipulated that the trial court could resolve certain legal issues by motion. Accordingly, the State filed a motion for a ruling that, because it had been held liable for property damage that was continuous across multiple policy periods, it was "entitled to indemnity up to the combined limits of all policies in effect during those policy periods . . . ."

At the same time, the Insurers filed briefs asking the trial court to rule that each of their policies covered only property damage attributable to the stated policy period, as opposed to the entire continuous loss. Alternatively, they argued that, even assuming each policy was deemed to cover the entire loss, the State could not recover the policy limits in effect for more than one policy period.

In its all-sums ruling, the trial court ruled in favor of the State: "[O]nce coverage for . . . continuous . . . damage . . . is triggered under a liability policy, the insurer is required to pay for all sums (up to the policy limits) of the insured's liabilitynot just liability specifically allocable to damage during the policy period."

In its no-stacking ruling, however, it ruled in favor of the Insurers: "[The] State may not `stack' or combine policy periods . . . . [¶] . . . [¶] [The] State is entitled to select a single policy period triggered by continuing damage from the occurrence at the Stringfellow site. [It] may recover the full amount of the limits of the policies in that period . . . ." It explained, in part: "[I]t appears that the court is bound by the holding in FMC Corp. [ v ]. Plaisted & Companies [, supra, ] 61 Cal.App.4th 1132 , which seems to be the case most fully on point."

(1) We begin with Montrose Chemical Corp. v. Admiral Ins. Co. (1995) 10 Cal.4th 645 [ 42 Cal.Rptr.2d 324 , 913 P.2d 878 ] ( Montrose ). There, the California Supreme Court held that "bodily injury and property damage that is continuous or progressively deteriorating throughout several policy periods is potentially covered by all policies in effect during those periods." ( Id. at p. 655; see also id. at pp. 654-655, 675, 685-689.) In other words, it adopted the "`continuous injury' trigger of coverage." ( Id. at p. 655.)

In Montrose, seven insurers had issued a series of liability policies, collectively covering the period from 1960 to 1986. Admiral Insurance Company (Admiral) had issued polices covering only the last four years of this period. ( Montrose, supra, 10 Cal.4th at p. 656 .) The issue before the court was whether Admiral had a duty to defend actions alleging either continuous or progressively deteriorating bodily injury or property damage, resulting from toxic chemicals manufactured by the insured, that began before, but continued during, Admiral's policy periods.

Admiral argued that a "manifestation" trigger of coverage applied; in other words, the only relevant "occurrence," within the meaning of its policies, was when appreciable bodily injury or property damage first appeared. ( Montrose, supra, 10 Cal.4th at pp. 662-663, 669, 677 & fn. 17.) The Supreme Court disagreed. It noted that Admiral's policies defined "property damage" as "`physical injury to or destruction of tangible property which occurs during the policy period' "; similarly, they defined "bodily injury" as "`bodily injury, sickness or disease sustained by any person which occurs during the policy period . . . .'" ( Id. at p. 668.) They then defined "occurrence" as "`an accident, including continuous or repeated exposure to conditions, which results in bodily injury or property damage . . . .'" ( Id. at p. 669.) The court concluded: "[T]his policy language unambiguously distinguishes between the causative eventan accident or `continuous and repeated exposure to conditions' and the resulting `bodily injury or property damage.' It is the latter injury or damage that must `occur' during the policy period, and `which results' from the accident or `continuous and repeated exposure to conditions.'" ( Ibid. )

The Insurers do concede that, under Montrose, they are liable for any property damage that actually occurred during their respective policy periods. They deny, however, that they are liable for any property damage that occurred before or after their policy periods. They acknowledge that on the facts of this case neither side would be able to prove that any particular property damage occurred during any particular policy period. Hence, they urge us to adopt a rule allocating the total property damage pro rata, based on each insurer's time on the risk.

Technically, the issue in Montrose was the trigger of coverage, not the allocation of coverage. In other words, the court was only called upon to decide which policies provided any coverage for a continuous loss. It was not called upon to decide how much of the loss was covered under each policy.

(2)Nevertheless, Montrose did declare it to be a "settled rule that an insurer on the risk when continuous or progressively deteriorating damage or injury first manifests itself remains obligated to indemnify the insured for the entirety of the ensuing damage or injury." ( Montrose, supra, 10 Cal.4th at p. 686 , italics added.) It also cited Gruol Constr. Co. v. Insurance Co. (1974) 11 Wn.App. 632 [ 524 P.2d 427 ] with apparent approval, noting that "the holding of Gruol was that, when warranted by the facts, property damage should be deemed to occur over the entire process of the continuing injury. An insurer would become liable at any point in the process for the entire loss up to the policy limits, even though the continuing injury or progressively deteriorating damage may extend over several policy periods." ( Montrose, at p. 678, italics added.) Similarly, it cited California Union Ins. Co. v. Landmark Ins. Co. (1983) 145 Cal.App.3d 462 [ 193 Cal.Rptr. 461 ] ( California Union ) as holding that: "[A]n insurer's liability for a still insured and continuing event is not terminated by the expiration of the policy term. [Citations.] . . . `[I]n a "one occurrence" case involving continuous, progressive and deteriorating damage, the carrier in whose policy period the damage first becomes apparent remains on the risk until the damage is finally and totally complete, notwithstanding a policy provision which purports to limit the coverage solely to those accidents/occurrences within the time parameters of the stated policy term.' [Citation.]" ( Montrose, at p. 680, quoting California Union, at p. 476.)

Accordingly, in Armstrong World Industries, Inc. v. Aetna Casualty & Surety Co. (1996) 45 Cal.App.4th 1 [ 52 Cal.Rptr.2d 690 ] ( Armstrong ), the appellate court held squarely that every insurer that issued a liability policy for any period during which a continuous loss occurred was liable for "the full extent of the loss up to the policy's limits . . . ." ( Id. at p. 49.) Armstrong relied primarily on out-of-state cases involving liability coverage for asbestos-related diseases. Those cases, in turn, had relied on the words "all sums" in the insuring agreement: "`The policies at issue in this case provide that the insurance company will pay on behalf of [the insured] "all sums" that [the insured] becomes legally obligated to pay as damages because of bodily injury during the policy period. . . . [W]hen [the insured] is held liable for an asbestos-related disease, only part of that disease will have developed during any single policy period. The rest of the development may have occurred during another policy period or during a period in which [the insured] had no insurance. The issue that arises is whether an insurer is liable in full, or in part, for [the insured]'s liability once coverage is triggered. We conclude that the insurer is liable in full, subject to the "other insurance" provisions . . . .' [Citation.]" ( Id. at p. 49, quoting Keene Corp. v. Ins. Co. of North America (D.C. Cir. 1981) 667 F.2d 1034 , 1047 ( Keene ).)

Subsequent decisions by our sister courts have unanimously concurred with Armstrong and followed the all-sums approach. ( Stonelight Tile, Inc. v. California Ins. Guarantee Assn. (2007) 150 Cal.App.4th 19, 37 [ 58 Cal.Rptr.3d 74 ]; FMC, supra, 61 Cal.App.4th at pp. 1184-1187; Stonewall Ins. Co. v. City of Palos Verdes Estates (1996) 46 Cal.App.4th 1810, 1854-1855 [ 54 Cal.Rptr.2d 176 ] ( Stonewall ).)

(3)Moreover, in Aerojet-General Corp. v. Transport Indemnity Co. (1997) 17 Cal.4th 38 [ 70 Cal.Rptr.2d 118 , 948 P.2d 909 ], the Supreme Court itself followed the all-sums approach. It stated that the duty to indemnify "is triggered if specified harm is caused by an included occurrence, so long as at least some such harm results within the policy period. [Citation.] It extends to all specified harm caused by an included occurrence, even if some such harm results beyond the policy period. [Citation.] In other words, if specified harm is caused by an included occurrence and results, at least in part, within the policy period, it perdures to all points of time at which some such harm results thereafter." ( Id. at pp. 56-57, italics added, fns. omitted.)

The court added: "To illustrate by a hypothetical . . .: Insurer has a duty to indemnify Insured for those sums that Insured becomes legally obligated to pay as damages for property damage caused by its discharge of hazardous substances, up to a limit of $1 million. Insured discharges such a substance. It thereby causes property damage to Neighbor's land, in the amount of $100,000 (determined by the cost of returning the soil to its original condition), within the policy period of year 1. It causes further damage of this sort as the substance spreads under the surface, in the amount of $100,000 annually, in year two through year thirty. Insured must pay Neighbor $3 million in damages under judgment. Insurer must pay Insured the limit of $1 million for indemnification." ( Aerojet-General Corp. v. Transport Indemnity Co., supra, 17 Cal.4th at p. 57 .)

Admittedly, Aerojet-General, like Montrose, involved the duty to defend. The Insurers therefore argue that this language was dictum. But not so. The precise issue in Aerojet-General was whether the insurer could make the insured pay any part of the costs of defense. ( Aerojet-General Corp. v. Transport Indemnity Co., supra, 17 Cal.4th at pp. 45, 51, 55-56.) The court reasoned that the insurer would be liable to indemnify the insured against all claims that resulted from some "triggering harm" during the policy period, even if the claims arose after the policy period. ( Id. at pp. 59-60, 68-69.) The court therefore held that the insurer was liable to defend the insured, unless it could prove that those claims did not result from some triggering harm during the policy period. ( Id. at p. 71.) It added: "[T]he insurers assume that their contractual duty to defend is limited to only that part of a `mixed' claim that comes within a policy period because specified harm may possibly have been caused by an included occurrence therein. They are wrong. As explained above, the duty to defend embraces all the parts of such a claim in which some such harm may possibly have resulted, whether within the policy period or beyond. " ( Ibid., italics added; see also id. at p. 74.) Thus, the all-sums approach to the duty to indemnify was crucial to the court's holding regarding the duty to defend.

(4)In any event, "`"[e]ven if properly characterized as dictum, statements of the [California] Supreme Court should be considered persuasive. [Citation.]" [Citation.]' [Citation.]" ( People ex rel. Totten v. Colonia Chiques (2007) 156 Cal.App.4th 31, 39, fn. 6 [ 67 Cal.Rptr.3d 70 ], quoting Hubbard v. Superior Court (1997) 66 Cal.App.4th 1163, 1169 [ 78 Cal.Rptr.2d 819 ].)

The Insurers understandably rely on footnote 19 in Montrose, which stated: "We do not endorse that aspect of the California Union court's holding that both insurers in that case were jointly and severally liable for the full amount of damage occurring during the successive policy period. [Citation.] Allocation of the cost of indemnification once several insurers have been found liable to indemnify the insured for all or some portion of a continuing injury or progressively deteriorating property damage requires application of principles of contract law to the express terms and limitations of the various policies of insurance on the risk. [Citations.]" ( Montrose, supra, 10 Cal.4th at p. 681, fn. 19 .)

(5)The all-sums approach, however, is not literally joint and several liability. Admittedly, the outcome is much the same as if it were; hence, it is sometimes loosely referred to as such. Nevertheless, it is not. The insurers are not jointly liable on each other's policies; rather, each insurer is severally liable on its own policy. (See Rohr Industries, Inc. v. First State Ins. Co. (1997) 59 Cal.App.4th 1480, 1489 [ 69 Cal.Rptr.2d 872 ] [insurers are, at most, serial obligors on separate contracts, not co-obligors on a contract debt]; Topa Ins. Co. v. Fireman's Fund Ins. Companies (1995) 39 Cal.App.4th 1331, 1339-1340 [ 46 Cal.Rptr.2d 516 ] [same]; Hartford Accident & Indemnity Co. v. Superior Court (1995) 37 Cal.App.4th 1174, 1181 [ 44 Cal.Rptr.2d 126 ] [same].)

In Aerojet-General, the Supreme Court explained that this was all that it meant by footnote 19 in Montrose: "In Montrose, we also made plain that `successive' insurers `on the risk when continuous or progressively deteriorating [property] damage or [bodily] injury first manifests itself' are separately and independently `obligated to indemnify the insured': `[W]here successive. . . policies have been purchased, bodily injury and property damage that is continuing or progressively deteriorating throughout more than one policy period is potentially covered by all policies in effect during those periods.' [Citation.] The successive insurers are not `jointly and severally liable.' [Citation.]" ( Aerojet-General Corp. v. Transport Indemnity Co., supra, 17 Cal.4th at p. 57, fn. 10 , italics added, quoting Montrose, supra, 10 Cal.4th at pp. 686-687 & 681, fn. 19.)

To summarize, then, in California, when there is a continuous loss spanning multiple policy periods, any insurer that covered any policy period is liable for the entire loss, up to the limits of its policy. The insurer's remedy is to seek contribution from any other insurers that are also on the risk.

The Insurers' arguments to the contrary founder on the fact that we must follow the California Supreme Court's lead. For example, they argue that the trial court's ruling is inconsistent with the language of the applicable policies. That language, however, is not significantly different from the standard policy language that was at issue in Montrose and Aerojet-General. (See Aerojet-General Corp. v. Transport Indemnity Co., supra, 17 Cal.4th at p. 49 .) Similarly, they argue that it is "objectively unreasonable" to hold them liable for losses before or after their respective policy periods. The same argument, however, could have been made in Montrose and Aerojet-General. Finally, they argue that "[t]he majority of jurisdictions that have considered this issue have rejected the `all sums' approach in the indemnity context." Even if so, [4] California has firmly aligned itself with the minority.

We therefore conclude that the trial court correctly ruled that each of the Insurers covered the total amount of the State's liability for property damage (subject to their respective policy limits), including property damage that actually occurred before or after their policy periods.

As we have just held, each successive insurer is potentially liable up to the entire loss. Nevertheless, that liability is capped by the policy limits. Accordingly, the next question is whether the State is entitled to stack the policy limits of the different policy periods.

(6)"Stacking" is a useful shorthand term, but it can be ambiguous. Hence, we begin by defining what we mean by it. In its broadest sense, stacking means treating multiple policies that apply to a single loss as cumulativeas a "stack" of coveragerather than as mutually exclusive. Hence, stacking issues can arise almost any time multiple policies cover a single loss. In Wallace v. Farmers Ins. Group (1986) 177 Cal.App.3d 735 [ 223 Cal.Rptr. 171 ], for example, "stacking" was used to refer to a husband's recovery under both his and his wife's automobile policies. ( Id. at p. 740.) In Barrett v. Farmers Ins. Group (1985) 174 Cal.App.3d 747 [ 220 Cal.Rptr. 135 ], "stacking" was used to refer to recovery by the victims of an auto accident under three policies, each covering a different vehicle. ( Id. at p. 749.)

Most oftenas in this casestacking refers to stacking of policy limits. Thus, the California Supreme Court has defined stacking as "`the ability of the insured, when covered by more than one insurance policy, to obtain benefits from a second policy on the same claim when recovery from the first policy would alone be inadequate' to compensate for the actual damages suffered. [Citation.]" ( Wagner v. State Farm Mutual Auto. Ins. Co. (1985) 40 Cal.3d 460, 463, fn. 2 [ 220 Cal.Rptr. 659 , 709 P.2d 462 ], quoting Nationwide Ins. Co. v. Gode (1982) 187 Conn. 386, 388, fn. 2 [ 446 A.2d 1059 ].) However, as we will discuss in more detail in part III.C.5., post, there can also be issues as to stacking of deductibles.

Alsoand again, as in this casestacking is most often used to refer to the stacking of policy limits across different policy periods. As we discussed in part III.B., ante, under Montrose, a continuous loss that occurs across multiple policy periods may be covered under every policy applicable to every such period; moreover, under Aerojet-General, each such policy may provide coverage up to the entire amount of the loss. When the entire loss is within the limits of any one policy, there is no stacking issue; the insured can recover from that insurer, which will then be entitled to contribution from the other insurers. However, whenever the loss is greater than the limits of any one applicable policy, the insured will seek to stack the policy limits across the policy periods.

That is what the State seeks to do in this case. Accordingly, from now on, whenever we refer to "stacking," without any further qualification, we will be referring to the stacking of policy limits across policy periods.

Note that, in a jurisdiction that does not follow Montrose, holding instead that a single continuous loss can only be covered by the policies in effect during a single policy period, an issue of stacking simply cannot arise. Likewise, in a jurisdiction that does not follow Aerojet-General, holding instead that a single continuous loss must be prorated across the applicable policy periods, stacking cannot be an issue. Stacking is an issue only in a jurisdiction that, like California, has both a continuous injury trigger ( Montrose ) and an all-sums rule ( Aerojet-General ). Under these circumstances, some jurisdictions have permitted stacking. (E.g., Society Ins. v. Town of Franklin (Ct.App. 2000) 233 Wis.2d 207, 216 [ 607 N.W.2d 342 ]; J.H. France Refractories v. Allstate (1993) 534 Pa. 29, 42 [ 626 A.2d 502 ]; Cole v. Celotex Corp. (La. 1992) 599 So.2d 1058, 1077-1080 .) Others have not. (E.g., American Physicians Ins. Exch. v. Garcia (Tex. 1994) 876 S.W.2d 842, 854-855 ; Keene, supra, 667 F.2d at pp. 1049-1050.) [5]

"Whether a policyholder may stack the applicable policy limits . . . can drastically affect the amount of the policyholder's eventual recovery." (Note, The Allocation of Coverage Responsibility Among Multiple Triggered Commercial General Liability Policies in Environmental Cases: Life After Owens-Illinois (1996) 15 Va. Envtl. L.J. 525 , 533-534, fns. omitted.) Indeed, the choice between stacking and not stacking can have an even more drastic effect than the choice between an all-sums approach and a pro rata approach.

Take the following hypothetical (summarized in the table below): Polluter Corp. is held liable for $30 million in property damage, resulting from six years of continuous pollution. In year one, it was insured by insurer A, subject to policy limits of $1 million per occurrence. In each of years two and three, it was insured by insurer B, subject to policy limits of $10 million per occurrence. And in each of years four, five, and six, it was insured by insurer C, subject to policy limits of $5 million per occurrence.

In a jurisdiction that uses a pro rata approach based on time on the risk, $5 million of the risk is allocated to insurer A; however, insurer A's liability is limited to $1 million. $10 million of the risk is allocated to insurer B, and $15 million of the risk is allocated to insurer C. Thus, Polluter Corp. can recover $26 million.

In a jurisdiction that uses the all-sums approach and that also allows stacking, each insurer is potentially liable for the full $30 million. Insurer A's liability, however, is limited to $1 million; insurer B's liability is limited to $20 million; and insurer C's liability is limited to $15 million. Thus, Polluter Corp. can recover the full $30 million. This will be allocated among the insurers in accordance with their contribution rights. While the precise allocation may depend on the presence and the wording of any "other insurance" clauses in the policies, it is most likely to be pro rata, by policy limits. (Croskey et al., Cal. Practice Guide: Insurance Litigation (The Rutter Group 2008) ¶ 8:26.) In that event, Insurer A will contribute 1/36 of the $30 million, or $833,333; insurer B will contribute 20/36 of the $30 million, or $16,666,667; and insurer C will contribute 15/36 of the $30 million, or $12.5 million.

In a jurisdiction that uses the all-sums approach but prohibits stacking, however, Polluter Corp.'s recovery is limited to $10 million. Moreover, this amount will be allocated among the insurers in accordance with their contribution rights. Hence, no insurer will end up paying its own full policy limits. [6]

"Insurance policy interpretation is a question of law. [Citation.]" ( Hameid v. National Fire Ins. of Hartford (2003) 31 Cal.4th 16, 21 [ 1 Cal.Rptr.3d 401 , 71 P.3d 761 ].) Accordingly, "the interpretation of an insurance policy is reviewed de novo under well-settled rules of contract interpretation. [Citation.]" ( E.M.M.I. Inc. v. Zurich American Ins. Co. (2004) 32 Cal.4th 465, 470 [ 9 Cal.Rptr.3d 701 , 84 P.3d 385 ].)

(7) "`The fundamental rules of contract interpretation are based on the premise that the interpretation of a contract must give effect to the "mutual intention" of the parties. "Under statutory rules of contract interpretation, the mutual intention of the parties at the time the contract is formed governs interpretation. [Citation.] Such intent is to be inferred, if possible, solely from the written provisions of the contract. [Citation.] The `clear and explicit' meaning of these provisions, interpreted in their `ordinary and popular sense,' unless `used by the parties in a technical sense or a special meaning is given to them by usage' [citation], controls judicial interpretation. [Citation.]" [Citations.] A policy provision will be considered ambiguous when it is capable of two or more constructions, both of which are reasonable. [Citation.] But language in a contract must be interpreted as a whole, and in the circumstances of the case, and cannot be found to be ambiguous in the abstract.' [Citation.]" ( MacKinnon v. Truck Ins. Exchange (2003) 31 Cal.4th 635, 647-648 [ 3 Cal.Rptr.3d 228 , 73 P.3d 1205 ], quoting Waller v. Truck Ins. Exchange, Inc. (1995) 11 Cal.4th 1, 18 [ 44 Cal.Rptr.2d 370 , 900 P.2d 619 ], quoting Civ. Code, §§ 1636, 1639, 1644.)

(8) We therefore begin with the language of the policies. Under the standard policy definition of "occurrence"particularly as construed in Montrose and Aerojet-General a continuous loss that spans multiple policy periods is, nevertheless, only a single occurrence. Standard policy limits are worded in terms of a dollar amount per "occurrence." For example, in this case, the policy limits provisions stated, "The limit of [the insurer]'s liability shall be . . . [¶] . . . [¶] [A specified dollar amount] . . . each occurrence. . . ." (Capitalization omitted.) For this reason, it could be argued that, in the event of a continuing loss spanning multiple policy periods, the insured is not entitled to recover more than the policy limits under any one policy. (See Croskey et al., Cal. Practice Guide: Insurance Litigation, supra, ¶ 7:376.1 [summarizing arguments].)

However, this overlooks the fact that the policy language only purports to limit each particular insurer's liability under each particular policy. Insurer A's policy provides that insurer A will not have to pay more than $X per occurrence; insurer B's policy provides that insurer B will not have to pay more than $Y per occurrence; and insurer C's policy provides that insurer C will not have to pay more than $Z per occurrence. Under the all-sums approach, each of these insurers is liable up to the amount of the entire loss as a result of an occurrence, subject only to its own policy limits. Thus, even though there is only one occurrence, the insured should be entitled to recover against each insurer up to the limits of that insurer's policy. [7]

(9) We believe that this follows from the plain meaning of the policy language. However, even assuming for purposes of argument that that language is ambiguous, it must be broadly construed in favor of the insured. "[I]nsurance coverage is `"`interpreted broadly so as to afford the greatest possible protection to the insured, [whereas] . . . exclusionary clauses are interpreted narrowly against the insurer.'"' [Citation.]" ( MacKinnon v. Truck Ins. Exchange, supra, 31 Cal.4th at p. 648, quoting White v. Western Title Ins. Co. (1985) 40 Cal.3d 870, 881 [ 221 Cal.Rptr. 509 , 710 P.2d 309 ], quoting Reserve Insurance Co. v. Pisciotta (1982) 30 Cal.3d 800, 808 [ 180 Cal.Rptr. 628 , 640 P.2d 764 ], quoting State Farm Mut. Auto. Ins. Co. v. Partridge (1973) 10 Cal.3d 94, 101-102 [ 109 Cal.Rptr. 811 , 514 P.2d 123 ].) [8] "Provisions which purport to exclude coverage or substantially limit liability must be set forth in plain, clear and conspicuous language. [Citations.]" ( Thompson v. Occidental Life Ins. Co. (1973) 9 Cal.3d 904, 921 [ 109 Cal.Rptr. 473 , 513 P.2d 353 ].) Certainly it cannot be said that the policy language plainly forbids stacking. ( American Physicians Ins. Exch. v. Garcia, supra, 876 S.W.2d at p. 854 ["`[n]ot surprisingly, the policies do not explicitly provide a means of applying the limits of liability to injuries that are covered by multiple policies'"]; Note, The Allocation of Coverage Responsibility Among Multiple Triggered Commercial General Liability Policies in Environmental Cases: Life After Owens-Illinois, supra, 15 Va. Envtl. L.J. 525 , 562 [standard liability policies "do not speak directly to the stacking issue"].)

Thus, the Insurers rely not so much on the policy language as on FMC, supra, 61 Cal.App.4th 1132 , which held that an insured cannot stack limits across policy periods. The trial court understandably considered itself bound to follow FMC . As we will discuss below, however, the reasoning in FMC is flawed, and as a result, FMC 's holding is outside the mainstream of California case law.

When multiple policies apply during a single policy period, the insured is entitled to stack limits. ( State Farm Mut. Auto. Ins. Co. v. Partridge, supra, 10 Cal.3d at pp. 96-97, 106.) The sole exception is statutory. Under Insurance Code section 11580.2, subdivision (q), the policy limits of two or more applicable uninsured motorist policies cannot be stacked. However, the very existence of this statutory exception demonstrates that in other situations, stacking is the rule.

Sometimes, it is not the insured who is trying to stack limits, it is an excess insurer. Assume that an insured is covered under one excess policy and more than one primary policy. "Vertical exhaustion" means that the policy limits under just one particular primary policy must be exhausted before the excess policy kicks in. By contrast, "horizontal exhaustion" means that the limits under all applicable primary policies must be exhausted before the excess policy kicks in. Thus, horizontal exhaustion is essentially the same thing as stacking the limits of primary policies on behalf of the excess insurer.

As a general rule, California requires horizontal exhaustion. Vertical exhaustion applies if, and only if, the excess policy provides that it is excess to a specified primary policy. ( Community Redevelopment Agency v. Aetna Casualty & Surety Co. (1996) 50 Cal.App.4th 329 , 339-340 & fn. 6 [ 57 Cal.Rptr.2d 755 ]; see Croskey et al., Cal. Practice Guide: Insurance Litigation, supra, ¶¶ 8:236 to 8:237.) This is true even when the primary policies apply to different policy periods. (See Community Redevelopment Agency , at p. 334.)

Technically, the horizontal exhaustion rule only governs the relationship between the primary and excess insurers. Nevertheless, it necessarily implies that the insured, too, is entitled to stack the primary policies; otherwise, the primary policies would never be exhausted. (See Iolab Corp. v. Seaboard Sur. Co. (9th Cir. 1994) 15 F.3d 1500 , 1504 [applying California law; insured was not entitled to indemnity from excess insurers because it had not yet exhausted all primary policies].)

The Insurers assert that cases involving multiple policies applicable to a single policy period "are not relevant," because this case involves multiple policies applicable to multiple policy periods. We see a distinction, but not a meaningful one. In each instance, there is a single occurrence; in each instance, each insurer's liability is limited to a stated amount per occurrence. Even so, the insured is allowed to stack the limits of all the applicable policies.

The Insurers likewise assert that cases dealing with horizontal exhaustion are not relevant, essentially because they involve primary policies rather than excess policies. Once again, we see no meaningful difference.

In light of this background, it is not surprising that the first California case in which the issue of stacking of limits across policy periods was squarely presented held that it is allowed. [9] In Stonewall, supra, 46 Cal.App.4th 1810 (see pt. III.B., ante ), the insured settled an action alleging continuous property damage from 1971 through 1981. ( Id. at pp. 1822-1823.) One insurer (Jefferson) had issued one primary liability policy, covering a period from 1975 through 1978, with three annual endorsements; each endorsement provided for a policy limit of $300,000 per occurrence. ( Id. at pp. 1833, 1849.)

The appellate court held that "the Jefferson policy covered three separate periods with a limit of $300,000 for each period, an aggregate of $900,000 in coverage." ( Stonewall, supra, 46 Cal.App.4th at p. 1849 .) It explained: "Jefferson claims . . . that its policy language defining occurrence limits its exposure to $300,000. This language . . . states that all damage arising from continuous and repeated exposure is deemed a single occurrence. This argument ignores two points: (1) the policy covers liability for occurrences within a policy period; and (2) the Jefferson policy covers three separate periods. Reading (as we must) the Jefferson policy in a fashion resolving ambiguities against it, the language on which Jefferson relies must be construed as referring to a single occurrence in a policy period. The Jefferson policy covering three policy periods, the policy language amounts to a $300,000 per period limitationor, in the context of this case (involving a continuous trigger), a $900,000 limitation." ( Ibid. )

In addition to reasoning that any ambiguity had to be resolved against Jefferson, the court also noted that Jefferson had, in effect, resolved the ambiguity against itself. Jefferson had entered into a stipulation (albeit with an excess insurer, see Stonewall, supra, 46 Cal.App.4th at p. 1833 , not with the insured) providing that "`[t]he subject policies of insurance issued by . . . Jefferson . . . provided coverage of $300,000 per occurrence per year as respects property damage . . . .'" ( Id. at p. 1849, italics added.)

Just two years later, however, FMC rejected stacking of limits across policy periods. There, the insured had caused toxic contamination "over a period of many years," starting as early as 1959. ( FMC, supra, 61 Cal.App.4th at p. 1142 .) During this period, the insured had a number of different liability insurers. ( Id. at p. 1143.) One particular set of excess insurers (the London defendants) had issued a series of seven policies, more or less annually, covering a period from 1964 through 1970. Each of the London policies had a limit of $1 million per occurrence. ( Id. at pp. 1147-1148, 1188.) Hence, if the insured was allowed to stack, it could recover up to $7 million per occurrence from the London defendants; if not, it could recover only $1 million per occurrence from them. ( Id. at pp. 1188-1189.)

The appellate court held that the insured was not entitled to stack. It explained: "`[S]tacking' . . . has been criticized as affording the insured substantially more coverage, for liability attributable to any particular single occurrence, than the insured bargained or paid for. [Citations.] [¶] Insurers sometimes include `anti-stacking' provisions in their policies to avoid just this kind of result. Where, as in this action, there is no anti-stacking provision, there is precedent, characteristically in asbestos cases, for judicial intervention." ( FMC, supra, 61 Cal.App.4th at p. 1189 .)

The court cited Ins. Co. North America v. Forty-Eight Insulations (6th Cir. 1980) 633 F.2d 1212 , an asbestos case, which had relied on the policy language, reasoning, "`The initial exposure to asbestos fibers in any given year triggers coverage. However, under the terms of the policies, additional exposure to asbestos fibers is treated as arising out of the same occurrence. Thus, on its face, the liability of each insurer is limited to maximum amount "per occurrence" provided by each policy. We have no problem with the district court's extending the policy language so that each insurer would face no more liability per claim than the maximum limit it wrote during any applicable year of coverage.' [Citation.]" ( FMC, supra, 61 Cal.App.4th at p. 1189 , quoting Ins. Co. North America, at p. 1226, fn. 28.)

The court also cited Keene, supra, 667 F.2d 1034 , another asbestos case, which had stated: "`The principle of indemnity implicit in the policies requires that successive policies cover single asbestos-related injuries. That principle, however, does not require that Keene be entitled to "stack" applicable policies' limits of liability. To the extent possible, we have tried to construe the policies in such a way that the insurers' contractual obligations for asbestos-related diseases are the same as their obligations for other injuries. Keene is entitled to nothing more. Therefore, we hold that only one policy's limits can apply to each injury. Keene may select the policy under which it is to be indemnified.' [Citation.]" ( FMC, supra, 61 Cal.App.4th at pp. 1189-1190, quoting Keene, supra, 667 F.2d at pp. 1049-1050.)

The court concluded: " Keene 's statement of such a rule is appropriate in the circumstances of this case for reasons well articulated in Keene and cognate cases." ( FMC, supra, 61 Cal.App.4th at pp. 1190-1191.) The court dismissed Stonewall , stating: " Stonewall does not analyze the issue and appears to base its conclusion at least in part on a stipulation between the parties." ( FMC, at p. 1190.)

The court therefore directed: "[O]nly the policy limits of . . . policies in effect as of July 1 in one of the policy periods in which coverage is triggered for a single occurrence can apply to property damage attributable to that occurrence, but . . . if coverage for that occurrence is triggered in more than one policy period [the insured] may select the policy period in which the policy limits are to be fixed." ( FMC, supra, 61 Cal.App.4th at p. 1191 .)

FMC refused to follow Stonewall because, in its view, " Stonewall does not analyze the issue and appears to base its conclusion at least in part on a stipulation between the parties." ( FMC, supra, 61 Cal.App.4th at p. 1190 .) The analysis in Stonewall certainly could have been more detailed, but it is at least clearthe policy language is susceptible of the meaning that stacking is allowed, and it must be construed against the insurer. Stonewall relied on the insurer's stipulation strictly as an alternative rationale. [10]

FMC failed to identify any flaw in Stonewall 's analysis of the policy language. Instead, it disregarded the policy language entirely. It did observe that policies sometimes include specific "anti-stacking" provisions, but the policies before it did not. It then concluded that the only way to avoid stacking was by "judicial intervention." ( FMC, supra, 61 Cal.App.4th at p. 1189 .) Thus, it at least implied that, absent antistacking provisions or "judicial intervention," standard policy language permits stacking.

The FMC court resorted to judicial intervention because it felt that stacking "afford[s] the insured substantially more coverage, for liability attributable to any particular single occurrence, than the insured bargained or paid for." ( FMC, supra, 61 Cal.App.4th at p. 1189 .) This argument, however, is circular. It assumes what it is meant to provethat the policies do not provide for stacking. In our view, standard policy language does provide for stacking, and therefore that is exactly what the insured has bargained and paid for.

If an occurrence happens entirely within one policy period, the insured has paid one premium and can recover up to one policy limit; however, if an occurrence is continuous across two policy periods, the insured has paid two premiums, and can recover up to the combined total of two policy limits. We see nothing unfair or unexpected in this.

The more subtle argument against stacking of limits is that the insured should not be better off when an occurrence is continuous across multiple years than when it is instantaneous. This is viewed as a windfall to the insured. For example, FMC quoted the following language from Keene: "To the extent possible, we have tried to construe the policies in such a way that the insurers' contractual obligations for asbestos-related diseases are the same as their obligations for other injuries. [The insured] is entitled to nothing more." ( Keene, supra, 667 F.2d at p. 1049.)

Once again, however, a continuous loss spanning two or more policy periods is fundamentally different from an instantaneous loss, such that it is appropriate to place a greater contractual obligation on the insurers. Indeed, an antistacking rule would give a windfall to the insurers. This should be apparent from the remedy with which FMC came up. It allowed the insured to choose any one policy period (i.e., the one with the highest limits) and to recover up to the limits in effect for that one period. Thus, it implicitly recognized that, under the all-sums approach, the insured was entitled to recover the entire loss under any one applicable policy, subject to that policy's limits. Once the insured did so, however, it was barred from recovering the excess under any other policy. Nevertheless, any one insurer that did pay up to its policy limits would still be entitled to contribution from all the other insurers. Thus, in the end, it would not actually have to pay its full policy limits. Accordingly, the insurers would benefit from the fact that the insured purchased multiple policies covering multiple periods. The insured, who made this prudent choice, would not. [11]

The fact that the insured is allowed to choose the policy period for which it wants to recover exposes the fallacy at the heart of an antistacking rule. If the insured is entitled to choose, it necessarily follows that each of the policies involved affords coverage, up to the limits of that policy. There is no contractual basis, found in the policies themselves, for requiring the insured to forego recovery under any other applicable policies, up to the limits of those policies. The only basis for such a result is indeed "judicial intervention." Forcing the insured to elect a single policy under which to recover also seems inconsistent with the spirit, if not the letter, of the all-sums approach adopted in Aerojet-General.

As in any line-drawing situation, it is possible to construct a horrible hypothetical by using facts that fall close to the line. Antistacking advocates posit a two-day occurrence, which starts on the last day of one policy period and ends on the first day of the next. If the same occurrence had occurred entirely within one policy period, the insured could recover only up to the limits of one policy. However, because of a fluke in timing, the insured can recover up to the limits of two policies.

Similar flukiness, however, is inherent in many cases of overlapping insuranceespecially as overlapping insurance can be obtained inadvertently, as well as intentionally. ( Martin Marietta Corp. v. Insurance Co. of North America (1995) 40 Cal.App.4th 1113, 1135 [ 47 Cal.Rptr.2d 670 ].) For example, under peculiar concatenations of fact, an insured may be covered under both a homeowner's liability policy and an automobile liability policy. In State Farm Mut. Auto. Ins. Co. v. Partridge, supra, 10 Cal.3d 94 , the insured and some friends were hunting jackrabbits from his car, using a gun that he had previously modified, giving it a hair trigger. When the car hit a bump, the gun went off, striking one of the friends. ( Id. at pp. 97-98.) The insured had both an automobile policy, with a limit of $15,000, and a homeowner's policy, with a limit of $25,000, issued by the same insurer. ( Id. at p. 98.) It was undisputed that the insured was covered under his automobile policy, because the injury arose out of the use of the insured vehicle. ( Id. at pp. 97, 100-101.) However, the California Supreme Court held that he was also covered under his homeowner's policyeven though that policy excluded injuries arising out of the use of the insured vehicle because the modification of the gun was a concurrent cause of the injury, and that cause was covered under the homeowner's policy. ( Id. at pp. 101-107.) Accordingly, the insurer was liable up to the limit of both policies.

FMC failed to recognize that, in all other instances of multiple coverage, stacking is allowed. As a result, it failed to provide any principled basis for distinguishing stacking when there is multiple coverage for an occurrence spanning multiple policy periods from stacking when there is multiple coverage for any other reason. As already discussed, we do not perceive any relevant distinction. Neither does FMC support the Insurers' contention that excess policies are somehow different.

(10) Finally, even if stacking somehow resulted in a windfall to the insured or unfairness to the insurer, we would not be authorized to cure it through "judicial intervention." Our Supreme Court has repeatedly declared that "we do not rewrite any provision of any contract, including the standard policy underlying any individual policy, for any purpose. [Citation.]" ( Certain Underwriters at Lloyd's of London v. Superior Court (2001) 24 Cal.4th 945, 960 [ 103 Cal.Rptr.2d 672 , 16 P.3d 94 ]; accord, Powerine Oil Co., Inc. v. Superior Court (2005) 37 Cal.4th 377, 401 [ 33 Cal.Rptr.3d 562 , 118 P.3d 589 ]; Haynes v. Farmers Ins. Exchange (2004) 32 Cal.4th 1198, 1212, fn. 9 [ 13 Cal.Rptr.3d 68 , 89 P.3d 381 ]; Rosen v. State Farm General Ins. Co. (2003) 30 Cal.4th 1070, 1073, 1077-1078 [ 135 Cal.Rptr.2d 361 , 70 P.3d 351 ].) This includes general purposes of public policy.

As the court stated in Aerojet-General Corp. v. Transport Indemnity Co., supra, 17 Cal.4th 38 , "[b]eneath the Court of Appeal's concern about `fairness' and `justice' is, apparently, a belief that, without an approach like the one it adopted, Aerojet might get a windfall from the insurers. That is not the case. We shall assume for argument's sake that Aerojet has enjoyed great good luck over against the insurers. But the pertinent policies provide what they provide. Aerojet and the insurers were generally free to contract as they pleased. [Citation.] They evidently did so. They thereby established what was `fair' and `just' inter se. We may not rewrite what they themselves wrote. [Citation.] We must certainly resist the temptation to do so here simply in order to adjust for chancefor the benefits it has bestowed on one party without merit and for the burdens it has laid on others without desert. [Citations.] As a general matter at least, we do not add to, take away from, or otherwise modify a contract for `public policy considerations.' [Citation.] . . . We shall therefore allow whatever `gains' and `losses' there may be to lie where they have fallen." ( Id. at pp. 75-76, fns. omitted, quoting AIU Ins. Co. v. Superior Court (1990) 51 Cal.3d 807, 818 [ 274 Cal.Rptr. 820 , 799 P.2d 1253 ].)

The Insurers rely on California Pacific Homes, Inc. v. Scottsdale Ins. Co. (1999) 70 Cal.App.4th 1187 [ 83 Cal.Rptr.2d 328 ] ( California Pacific ). However, as they acknowledge, the issue in California Pacific was not stacking of limits; rather, it was stacking of deductibles.

The insured had paid just under $2 million to settle a lawsuit involving continuous property damage from 1984 through 1995. ( California Pacific, supra, 70 Cal.App.4th at pp. 1189-1190.) Two liability insurers had issued a total of five annual policies covering a period from 1990 through 1995. ( Id. at p. 1190 & fn. 1.) Each policy had a self-insured retention [12] of $250,000 per occurrence and a policy limit of $1,750,000 per occurrence. The insured made a demand for indemnity under just one of the policies. ( 70 Cal.App.4th at p. 1190 .) Thus, it was the insurers who were in favor of stackingthey took the position that the insured was required to bear the $250,000 self-insured retention under all five policies, or $1,250,000, before they had any duty to indemnify. ( Ibid. )

The appellate court held that only one self-insured retention applied. ( California Pacific, supra, 70 Cal.App.4th at pp. 1191-1195.) It relied primarily on the policy language, "The policy language . . . provides for the insured's retained limit to be $250,000 of the `ultimate net loss as the result of any one occurrence because of . . . property damage.' Occurrence is defined as `an accident, event or happening, including injurious exposure to conditions, which result, during the policy period, in . . . property damage . . . .' [¶] The parties stipulated . . . that the . . . claims arose from a single occurrence involving continuous or progressive property damage occurring during the time in which the first Scottsdale policy was in force. Thus, once the ultimate net loss of [the insured] had exceeded $250,000 the . . . policy provided coverage up to $1,750,000 `as the result of any one occurrence because of . . . property damage.'" ( Id. at p. 1192, fn. omitted.)

The court also relied on the fact that the insured was seeking to recover under only a single policy: "In the present case the insured made a demand under a single policy and the amount of the insured's ultimate net loss . . . was within the policy limits for one occurrence under that single policy. How these insurers choose to proceed as between themselves is not before us. [¶] This distinction is crucial. " ( California Pacific, supra, 70 Cal.App.4th at p. 1195 , italics added.)

California Pacific does not support the Insurers' position. First of all, because it dealt with stacking of deductibles, rather than stacking of limits, it focused on different policy language. Even more important, the court found it "crucial" that a single policy was sufficient to cover the entire loss; hence, it was appropriate to require the insured to pay only a single self-insured retention. It appears that, if the loss had been greater than the total limits of all five policies, and if the insured therefore had been seeking to recover under all five, it would have had to pay five self-insured retentions. That is perfectly consistent with our approach to stacking of limits.

Wausau additionally relies on a more recent case, Safeco Ins. Co. of America v. Fireman's Fund Ins. Co. (2007) 148 Cal.App.4th 620 [55 Cal.Rptr.3d 844]. There, a single primary insurer had issued a homeowner's liability policy, with policy limits of $500,000 per occurrence, and had then renewed it annually, so as to cover policy periods from 1997 through 2001. ( Id. at pp. 626, 631.) In 1998, a landslide from the insured's property damaged several downhill properties. ( Id. at p. 625.) Over the succeeding years, there was further damage as a result of the original landslide; for example, when it rained, "[m]ud, debris, vegetation, and water . . . continued to flow [downhill]." ( Id. at p. 626.) Eventually, the insured was held liable for approximately $4 million in property damage. ( Id. at p. 627.)

The appellate court held that the primary insurer was liable for only $500,000. ( Safeco Ins. Co. of America v. Fireman's Fund Ins. Co., supra, 148 Cal.App.4th at pp. 632-639.) It explained that there had been only one occurrence. ( Id. at pp. 632-634.) It concluded: "Finally, we note that many comprehensive general liability policies obligate the insurer to `"pay on behalf of the insured all sums which the insured shall become legally obligated to pay as damages because of . . . property damage . . . ."' [Citation.] Because the insurer agrees to pay `all sums,' the policy language may support an argument that the insured is entitled to the policy limits under successive policies even though there is only one occurrence. [Citations.] This argument, which could result in `stacking' policy limits, has no application to the homeowner's policy here, under which the insurer agreed to `[p]ay up to our limit of insurance [, $500,000 per occurrence, ] for the damages for which the "insured" is legally liable.' (Italics added.) As we have discussed, there was only one occurrence in this case, entitling the insured to $500,000 in indemnity under the first policy." ( Id. at p. 638, quoting Montrose, supra, 10 Cal.4th at p. 656 .)

Essentially, then, the court held that the all-sums rule does not apply to a homeowner's policy. Here, of course, the all-sums rule does apply; indeed, this is precisely why stacking is an issue in this case. [13] Safeco simply has no bearing on whether stacking is allowed under a standard liability policy.

(11) To summarize, then, we find all antistacking arguments and authorities unpersuasive. Based on the policy language, and also based on a consistent line of authority in California that allows stacking in other multiple coverage situations, we conclude that the State was entitled to stack the policy limits of all applicable policies across all applicable policy periods. The trial court's no-stacking ruling was therefore erroneous.

The State contends that the trial court erred by ruling that, for purposes of the Insurers' limits of liability, there had been only a single "occurrence."

In discovery responses, the State took the position that there had been the following four occurrences:

1. The escape of contaminants through the fractured rock, caused by the State's failure to discover the fractures;

2. The escape of contaminants through the barrier dam, caused by the State's use of some natural materials, rather than all concrete, in the construction of the dam;

3. The escape of contaminants through the underground streambed, caused by the State's failure to discover the streambed; and

4. The 1969 overflow of the dam, caused by the State's failure to provide adequate diversion channels and otherwise to protect against overflow. [14]

The State admitted, however, that it could not allocate any particular property damage to any particular one of the claimed occurrences.

By agreement, the parties filed briefs asking the trial court to decide the following issue: "What are the parties' respective burdens with respect to proving the cause and amount of property damage during the policy period?" The trial court ruled: "With respect to each occurrence, the State has the burden of showing the cause and the amount of property damage that resulted from that occurrence" (burden of proof ruling).

The Insurers then filed a motion for summary judgment, arguing that, once the trial court's burden of proof ruling was combined with the State's concession that it could not allocate the property damage among the claimed occurrences, the State could not recover. Alternatively, the Insurers sought an order in limine barring the State from asserting that there had been more than one occurrence.

The trial court denied the motion for summary judgment. However, it ruled that there were, at most, two occurrences. It explained:

"[T]he court consider[s] the failure of the State to discover the fractured granite, underground streambed, the inadequate barrier dam and negligent site construction to be an accumulation of factors creating a leaking sieve. These four conditions existed simultaneously once the site was open. . . . [¶] . . . [¶]

"The 1969 storm event presents a different issue. . . . [¶] . . . [¶] The waste leaving the site through the sieve is a different occurrence than the overflow caused by the heavy rains in 1969. These are independent sources of contamination and one can clearly occur without the other.

"[T]he 1969 overflow could be considered a distinct and separate occurrence from the continuous waste flow caused by the flaws in the site's geology. Therefore, the issue is one for the jury."

The Insurers then filed a motion in limine to bar the State from offering evidence that the 1969 overflow caused any different or additional property damage. They noted again that the State had conceded that it could not allocate the property damage among the claimed occurrences.

The trial court granted the motion and barred the State from offering any evidence that any portion of the claimed property damage was due to the 1969 overflow.

(12) "For the purpose of determining whether there was coverage within the policy period, it is well established that the time of the relevant `occurrence' or `accident' is not when the wrongful act was committed but when the complaining party was actually damaged. [Citations.]" ( Whittaker Corp. v. Allianz Underwriters, Inc. (1992) 11 Cal.App.4th 1236, 1241 [ 14 Cal.Rptr.2d 659 ].) However, "for the purpose of an insurer's limitation of liability to a certain amount for each covered occurrence . . ., occurrence has generally been held to mean the underlying cause of the injury, rather than the injury or claim itself; otherwise, the insurer's effort to limit its liability per occurrence would be substantially weakened." ( Id. at pp. 1241-1242, fns. omitted.)

"`Under the cause test, there is a single occurrence when "there was but one proximate, uninterrupted, and continuing cause which resulted in all the injuries and damage." When all injuries emanate from a common source or process, there is only a single occurrence for purposes of policy coverage. It is irrelevant that there are multiple injuries or injuries of different magnitudes, or that the injuries extend over a period of time. Conversely, when a cause is interrupted, or when there are several autonomous causes, there are multiple "occurrences" for purposes of determining policy limits and assessing deductibles.' [Citations.]" ( Caldo Oil Co. v. State Water Resources Control Bd. (1996) 44 Cal.App.4th 1821, 1828 [ 52 Cal.Rptr.2d 609 ] ( Caldo ), quoting 3 Cal. Insurance Law & Practice (1995) General Liability Policies, § 49.18[3][b], pp. 49:54 to 49:55.)

In Caldo, the court stated: "One treatise posits the following three hypotheticals to illustrate the `continuous or repeated exposure to conditions' clause: First, an above-ground storage tank leaks undetected for a period of time. Any damage `can be said to be continuous from the initial event.' Second, assume the leakage resulted from a transfer line which operates only intermittently. Whenever the transfer line is used, a tank valve leaks. `In this situation, the damage is not caused by a continuous exposure to conditions, i.e., a steady leak from the faulty tank, but rather by a repeated exposure to the same conditions, i.e., the [transfer] line and the faulty valve which intermittently [leaks].' Third, `Change the foregoing hypothetical to include both the faulty tank and the [transfer] line with the faulty valve. If both the tank failure and the valve leakage were the result of . . . over-pressurization, it is arguable that only one occurrence has taken place, i.e., the over-pressurization which resulted in damage. On the other hand, if the tank failure was caused, for example, by a negligent forklift driver who inadvertently punctured the tank without realizing it, while the valve malfunction was caused by the over-pressurization of the transfer line, it would seem that two occurrences have taken place, because each one resulted from different causal conditions.' [Citation.] We agree with this causal analysis." ( Caldo, supra, 44 Cal.App.4th at pp. 1828-1829, quoting Lathrop, Insurance Coverage for Environmental Claims (1995) Environmental Statutes, § 3.05[3], pp. 3-52 to 3-54.)

Here, the State focuses on three conditions: the fractured granite, the barrier dam, and the underground streambed. It ignores, however, a fourth condition, without which none of these conditions, either alone or together, could have caused any property damage: the deposit of waste at the site. It was only at that point that the three other preexisting conditions, operating concurrently, caused the escape of contaminants.

This situation does not fit precisely within any of the three hypotheticals in Caldo. However, it clearly is not analogous to the second hypothetical the intermittent use of the transfer line, resulting in intermittent leakage. Here, once the "leakage" began, it was continuous. Thus, our facts are most closely analogous to the first hypotheticala steady leak from a faulty tank. The only difference is that the "tank" was faulty for three reasons, rather than one. Nevertheless, the policies defined "occurrence" as "a continuous or repeated exposure to conditions which result in . . . damage to property . . . ." (Italics added.) Hence, there can be multiple contributing conditions, yet only a single occurrence.

Even assuming three different negligent forklift drivers poked three separate holes in the tank before it was installed, the relevant occurrence would be installing and filling the tank. Thus, the trial court aptly used the trope of a leaking sieve; it stated, "[T]he four events up until the time that you put waste material in the dump have no effect . . . . [I]nstead of building a bowl, you built a sieve. Does that mean that every little hole in the sieve is an occurrence? I don't think so." It would make no sense to treat each hole in a sieve as a distinct cause of the sieve's failure to hold water. This is true even if the sieve maker originally made it with one hole, then decided to add a second hole, and then a third hole. The relevant "occurrence" would be pouring water into the sieve.

Obviously, it would be different if the tank was installed and filled, and then three forklift drivers poked three holes in the tank. The State relies on cases like thatin which one event or condition had already begun causing damage before a second event or condition came along. For example, in Wiltshire v. Government of Virgin Islands (3d Cir. 1990) 893 F.2d 629 , a hospital treating a premature baby committed three negligent acts: (1) negligent use of an umbilical catheter, which caused an infection; (2) negligent administration of cardiopulmonary resuscitation, which was "`destructive to breathing'"; and (3) negligent placement of an intravenous feeding line, which scarred the baby's face. ( Id. at pp. 631, 634.) The baby was left "severely retarded." ( Id. at p. 631.) The court held that these were three occurrences; it explained that the baby's "injuries did not come about as a result of extended exposure to the same basic condition." ( Id. at p. 634.)

A case that we find more closely on point is a different decision by the Third Circuit, namely Flemming ex rel. Flemming v. Air Sunshine (3d Cir. 2002) 311 F.3d 282 . There, the plaintiff's husband had died after a plane crash. ( Id. at p. 284.) The plaintiff argued that there had been multiple occurrences, including (1) the plane crash itself, (2) the failure to provide a preflight safety briefing, and (3) the failure to notify passengers of the impending crash and to provide emergency safety instructions. ( Id. at p. 295.) The appellate court disagreed: "[The plaintiff]'s allegations of pre-crash negligence, including failure to provide a safety briefing and failure to provide warning of the crash, do not meet the policy definition of `occurrence' because they simply cannot be seen as `accidents' independent from the crash itself. Any pre-crash acts of negligence cannot be termed proximate causes of [the decedent]'s death because the crash intervened and the pre-crash negligence would not have caused any injury absent the crash." ( Id. at p. 295.)

The State argues that the trial court effectively inserted "one occurrence" provisions into policies that did not have them. This is unfair. The trial court ruled that the State had not shown multiple occurrences; it did not rule that the State could not do so. Indeed, it was willing to recognize the 1969 overflow as a distinct occurrence, if only the State could show that it had resulted in any distinct damages.

Finally, the State argues that the policy definition of "occurrence" was at least ambiguous. We do not believe, however, that it was reasonably susceptible of a construction under which the fractured rock, the barrier dam, and the underground streambed were separate occurrences. Arguably, each of these was a separate condition ; however, once they coincided with the deposit of waste at the site, they resulted in a single instance of "continuous . . . exposure" to all three conditions collectively.

In a related contention, the State argues that the trial court erred by ruling that it had the burden of proving the amount of property damage that resulted from each occurrence. Even assuming this was error, it was harmless. In this appeal, the State has taken the position that, "as a matter of law," there were three occurrencesthe fractured granite, the barrier dam, and the underground streambed. And, as we have just held, the trial court correctly ruled that these three occurrences were actually just one. Accordingly, the State never actually had to allocate the property damage as between more than one occurrence. Admittedly, the trial court's burden of proof ruling did factor into its later ruling that the 1969 overflow was not an occurrence, because the State had conceded that it could not allocate any particular property damage to the 1969 overflow. Nevertheless, inasmuch as the State is no longer claiming that the 1969 overflow was an occurrence, it does not appear that this prejudiced the State's substantial rights. (Cal. Const., art. VI, § 13; Code Civ. Proc., § 475.)

We therefore conclude that the trial court correctly ruled that the State had shown only a single occurrence.

The State contends that the trial court erred by allowing the Insurers a setoff, or credit, for settlements paid by other insurers.

By the end of trial, the State had entered into settlements with other defendant insurers totaling approximately $120 million. In all of these settlement agreements, the State released, at a minimum, the claims that it was asserting in this action for coverage for its liability arising out of the Stringfellow site. However, it typically also released any claims for bad faith. Sometimes it also released any other claims under the settling insurer's policies. Finally, in a few of the settlement agreements, the State also released other specified coverage claims, such as for liability arising out of other hazardous waste sites or out of exposure to other contaminants, such as asbestos.

After the entry of the jury's verdicts, pursuant to the trial court's no-stacking ruling, the State elected to recover under the 1970 to 1973 policy period. The total policy limits in effect for that period were $48 million.

The Insurers then filed a motion asking the trial court to offset the $120 million that the State had received from other insurers against the Insurers' liability. The trial court granted the motion. Because the settlements exceeded the State's maximum possible recovery, the trial court entered judgment in favor of the State in the amount of "$0."

Because we are reversing the no-stacking ruling, on remand, the State's recovery will not be limited to any one policy period. Accordingly, we need not discuss the State's contention that the setoff was improper because the settlement payments may have related to other policy periods, or because the State had not yet been made whole for its entire loss.

The State also contends that the setoff was improper because the settlement payments were made (at least in part) in exchange for the release of claims that were not at issue in this action, such as claims arising out of other sites and other contaminants, claims for bad faith, unknown claims, etc. In its reply brief, however, in arguing that it should not have the burden of allocating the settlement payments, it states that its loss (i.e., when calculated properly, with stacking allowed) "greatly exceeds all recoverable insurance." It asserts that, under these circumstances, an allocation of the settlement amounts between claims asserted against the Insurers in this action and other claims is "irrelevant[,] . . . unwarranted and unnecessary." We deem this to be a concession that, on remand, following reversal of the no-stacking ruling, the total loss will be so large that it will not matter whether the Insurers are allowed a setoff or not.

We conclude that, in light of our reversal of the no-stacking ruling, the State's challenge to the setoff ruling is moot.

The Insurers contend that the trial court erred by ruling that the State could recover portions of the loss that were caused by its own unreasonable delay.

In the underlying federal action, there were findings that, between 1975 and 1980, the State negligently delayed remediation of the site.

In particular, in 1975, Dr. Franks, a state geologist, and James Anderson, the state employee in charge of the site, both recommended that the State should cap the site, install a hydraulic barrier, and dig new monitoring wells. At least until 1980, none of this was done. As a result of these omissions, in 1978, during heavy rains, the State was forced to allow a "controlled discharge." Moreover, the infiltration of rainwater "increase[d], by multiples, the volume of toxic waste in the groundwater, thus increasing the damage to the environment and the difficulty and cost in cleaning up the site." Thus, "[t]his negligence exacerbated the damage and increased the cost of clean-up."

In 1975, the estimated cost of cleaning up the site was between $225,000 and $410,000; [17] now, the estimated cost of doing so, at least according to the State, is over $700 million.

In a pretrial motion, the Insurers sought a ruling that the State was collaterally estopped by the findings in the federal action, including the findings that it negligently delayed remediation at the site. In addition, they sought a ruling that this negligent delay violated the State's "duty to mitigate the damage for which it seeks coverage." The State responded that it had no such duty.

The trial court ruled, "[T]he State is bound by the underlying federal court's findings that the State negligently delayed and failed to remediate the Stringfellow site. However, the Court does not agree that the State breached a duty owed to the Insurers to mitigate damages. The Court does not believe that there is a duty to mitigate. The policies do not contain an express duty to mitigate and . . . such a duty to the Insurers is inconsistent with third-party liability policies."

(13) "The doctrine of mitigation of damages holds that `[a] plaintiff who suffers damage as a result of either a breach of contract or a tort has a duty to take reasonable steps to mitigate those damages and will not be able to recover for any losses which could have been thus avoided.' [Citations.] A plaintiff may not recover for damages avoidable through ordinary care and reasonable exertion. [Citation.]" ( Valle de Oro Bank v. Gamboa (1994) 26 Cal.App.4th 1686, 1691 [ 32 Cal.Rptr.2d 329 ], quoting Shaffer v. Debbas (1993) 17 Cal.App.4th 33, 41 [ 21 Cal.Rptr.2d 110 ].)

A duty to mitigate damages normally applies to damages resulting from the defendant's wrongful conduct. Accordingly, in recent years, the Supreme Court has referred to the duty to mitigate as "the avoidable consequences doctrine." ( State Dept. of Health Services v. Superior Court (2003) 31 Cal.4th 1026, 1043 [ 6 Cal.Rptr.3d 441 , 79 P.3d 556 ].) It has explained that, under this doctrine, "a person injured by another's wrongful conduct will not be compensated for damages that the injured person could have avoided by reasonable effort or expenditure. [Citations.]" ( Ibid. )

Here, however, the amounts that the Insurers argue the State failed to mitigate were not consequences of the Insurers' breach. The Insurers did not even arguably breach the contract of insurance until September 1998, when the State was held liable in the underlying federal action. Moreover, the Insurers' breach in no way caused the remediation costs for which the State was held liable. Thus, the doctrine of mitigation of damages, as such, simply did not apply.

The Insurers requested a jury instruction on mitigation of damages. (CACI No. 358.) [18] They claim that the trial court refused to give it. However, they have not supported this claim by any citation to the record. (See Cal. Rules of Court, rule 8.204(a)(1)(C).) At oral argument, the Insurers called our attention to an instructions conference reflected in the reporter's transcript. In it, however, we find no mention of the mitigation instruction (Defendants' Instruction No. 56). It does appear that the trial court did not give the instruction, but we cannot tell whether this was because it refused it, because the Insurers withdrew it, or for some other reason. Accordingly, they have forfeited any claim of error relating to this instruction. ( Faulk v. Soberanes (1961) 56 Cal.2d 466, 471 [ 14 Cal.Rptr. 545 , 363 P.2d 593 ] ["appellant[] has the burden to present a record sufficiently complete to establish that the claimed [instructional] errors were not invited by her, and in the absence of such a showing she may not properly complain"]; In re S.C. (2006) 138 Cal.App.4th 396 , 406-407 [ 41 Cal.Rptr.3d 453 ].) We note, however, if only out of an excess of caution, that because the doctrine of mitigation of damages, as such, did not apply, any refusal to give this instruction would not have been error.

(14) The principle on which the Insurers actually appear to be relying has been called "mitigation of loss." According to Couch, "the initial concept of mitigation of loss for the benefit of the insurer arose in the context of marine insurance, where it appears to reside primarily in two doctrines: (1) that insureds should be able to recover the costs of repairs undertaken to prevent a more serious loss, and (2) insureds should be able to recover the costs of efforts . . . to recover or protect insured property. The modern concept has somewhat slowly spread beyond marine insurance to such other contexts as property insurance, business interruption coverage, no-fault automobile insurance, and even disability insurance." (11 Couch on Insurance (3d ed. 2000) § 168:9, p. 168-16, fns. omitted.)

There are two versions of the doctrine of mitigation of loss the carrot and the stick. In the carrot version, the insured is entitled to recover any expenses it has incurred to prevent additional losses, so that the insured is rewarded for preventing them. In the stick version, the insured is not allowed to recover any additional losses that it negligently failed to prevent, so that the insured is punished for failing to prevent them.

The discussion by Couch, quoted above, deals exclusively with the carrot version of the doctrine. Elsewhere, however, Couch also deals with the stick version, stating, "[B]reach of the duty to mitigate may prevent the insured from recovering the portion of the loss which compliance with the mitigation duty could have prevented. . . . [T]he fact that the loss could have been less had the insured or the plaintiff taken greater care after the loss was sustained will bar the insured from recovery, at least for such excess as could have been avoided." (11 Couch on Insurance, supra, § 168:13, pp. 168-23 to 168-24, fns. omitted.) "Failure to take reasonable care to avoid an increase of the loss may defeat a recovery, for the mere existence of an insurance policy does not justify carelessness in failing to protect the insured property after loss." ( Id. at p. 168-24, fn. omitted.)

California does recognize the carrot version of the doctrine, at least in the first party context. Thus, Insurance Code section 531 provides: "An insurer is liable: [¶] . . . [¶] (b) If a loss is caused by efforts to rescue the thing insured from a peril insured against." However, we have not found any California cases squarely applying the carrot version of the doctrine in the third party context. Admittedly, in City of Laguna Beach v. Mead Reinsurance Corp. (1990) 226 Cal.App.3d 822 [ 276 Cal.Rptr. 438 ], this court stated, "The [insured] has accurately cited authority for the general proposition that an insured can recover from an insurer costs incurred by the insured for the primary benefit of the insurer to mitigate against the further occurrence of an insured loss." ( Id. at p. 833.) This was dictum, however, as we went on to hold that the mitigation expenses that the insured was seeking were not recoverable because they had not actually gone to mitigate "an insured loss," i.e., a loss for which the insured was liable. ( Ibid. )

Similarly, in Globe Indem. Co. v. State of California (1974) 43 Cal.App.3d 745 [ 118 Cal.Rptr. 75 ] ( Globe ), the court stated, "A rule, reasonably applied, permitting expenses incurred in the mitigation of damages to tangible property to be recoverable under policies insuring against liability incurred because of damages to tangible property would seem to require universal application as it encourages a most salutary course of conduct." ( Id. at p. 752.) Once again, however, this was dictum. The state had sued the insureds for the costs of fighting a fire they had allegedly started. ( Id. at p. 748.) The court's actual holding was that these costs were within the insuring clause of the policies, which "extend[ed] coverage `to all sums which the insureds become legally obligated to pay as damages because of' property damage. . . ." ( Id. at p. 751.)

Finally, in Aerojet-General Corp. v. Superior Court (1989) 211 Cal.App.3d 216 [ 258 Cal.Rptr. 684 ], the state and the United States had sued the insureds to recover environmental response costs they had incurred. ( Id. at p. 221.) The court held that at least some of the response costs were covered under third party liability policies, because they were "damages" within the meaning of the insuring clauses of the policies. ( Id. at pp. 225-232.) In the course of its discussion, it cited and quoted Globe. ( Aerojet-General Corp., at pp. 226-227.) However, the court also specifically held that response costs incurred "to prevent future pollution of a type which has not yet occurred, or to prevent pollution from a source which has not yet caused pollution" would not be covered. ( Id. at p. 237.) It explained, "These costs would not be causally related to property damage and would therefore not be covered as `damages' under the policies." ( Ibid. ) Thus, it essentially held that at least some costs of mitigating a loss are not recoverable under a third party liability policy.

We likewise have not found any California cases squarely applying the stick version of mitigation of loss, either in the first party or the third party context. [19] Couch does not cite any California cases supporting his views on this point. Two cases dealing with a "sue and labor" clause [20] have stated that this clause imposes an express requirement that would in any event be implied; yet again, however, this was dictum. ( Young's Market Co. v. American Home Assur. Co. (1971) 4 Cal.3d 309, 313 [ 93 Cal.Rptr. 449 , 481 P.2d 817 ]; Southern Cal. Edison Co. v. Harbor Ins. Co. (1978) 83 Cal.App.3d 747, 757 [ 148 Cal.Rptr. 106 ].) Considering how often insureds must negligently fail to mitigate a loss, the very absence of authority seems telling.

(15) Finally, even assuming the stick version might apply to first party coverage, we conclude that it does not apply to third party liability coverage. "First party coverage for damage to the insured's own property is not the same as third party liability insurance and should be treated differently. [Citations.]" ( Shell Oil Co. v. Winterthur Swiss Ins. Co. (1993) 12 Cal.App.4th 715, 765 [ 15 Cal.Rptr.2d 815 ]; see also Montrose, supra, 10 Cal.4th at p. 692 .)

The stick version of mitigation of loss is antithetical to the whole purpose of third party liability insurance, which is to provide coverage for the insured's own negligence. Such coverage should apply not only to the insured's initial negligence that causes the loss, but also to any negligence that either aggravates or fails to mitigate the loss. Any distinction is a distinction without a difference. In both instances, the insured fails to act with due care, causing a loss for which the insurer is ultimately liable. Suppose that during an operation, one hospital employee leaves a sponge inside the patient; later, a different hospital employee negligently fails to discover the sponge, which could be viewed as a failure to mitigate damages. We can see no principled basis for holding that the first employee's negligence makes the hospital's insurer liable, yet the second employee's negligence relieves the insurer of further liability.

We find some support for our conclusion in Downey Savings & Loan Assn. v. Ohio Casualty Ins. Co. (1987) 189 Cal.App.3d 1072 [ 234 Cal.Rptr. 835 ]. There, defendant Ohio had issued a fidelity bond to plaintiff Downey. "The bond provided coverage for losses caused by the dishonest act of any of Downey's employees . . . . The bond also promised to indemnify Downey for court costs and attorney's fees incurred by Downey in defending any lawsuit brought by a third party arising out of the dishonest acts of Downey's employees." ( Id. at pp. 1080-1081.) One of Downey's employees then entered into a conspiracy, involving the deposit of a bad check for $100,000 into a Downey account, and the use of that account as collateral for a $100,000 loan from another lender. ( Id. at pp. 1081-1082.) When the other lender discovered the fraud, it sued Downey. ( Id. at p. 1082.) Ohio denied any coverage under the bond. ( Id. at pp. 1082, 1084-1085.)

At one point, one of the conspirators had offered to buy the bad check back from Downey for $100,000 in cash. Upon being told that the bad check was not immediately available, he withdrew his offer. ( Downey Savings & Loan Assn. v. Ohio Casualty Ins. Co., supra, 189 Cal.App.3d at pp. 1090, 1095.) Based on this evidence, Ohio asked the trial court to give a jury instruction on mitigation of damages; it refused. ( Id. at pp. 1094-1095.) The appellate court held that this was not error, in part because Ohio had conceded that the offer was not bona fide. ( Id. at p. 1095.) It added, however: "[E]ven if Downey's conduct constituted negligence, it would not have exonerated Ohio. Insurance Code section 533 provides that an insurer `is not exonerated by the negligence of the insured, or of the insured's agents or others.' [Citation.]" ( Ibid. ) [21]

Commercial Union Assurance Companies v. Safeway Stores, Inc. (1980) 26 Cal.3d 912 [ 164 Cal.Rptr. 709 , 610 P.2d 1038 ] ( Commercial Union ), while not precisely on point, also supports our conclusion. There, a judgment was entered against Safeway in the amount of $125,000. Two insurers (collectively called Commercial) had issued excess liability policies to Safeway, covering liability in excess of $100,000; thus, Commercial had to pay $25,000 of the judgment. Commercial then sued Safeway, alleging that Safeway had had an opportunity to settle the case for $50,000 or $60,000 but had negligently failed to do so. The trial court sustained a demurrer to this claim. ( Id. at p. 916.)

The California Supreme Court affirmed, holding that an excess policy "imposes no implied duty upon the insured to accept a settlement offer which would avoid exposing the insurer to liability. Moreover such a duty cannot be predicated upon an insured's implied covenant of good faith and fair dealing." ( Commercial Union, supra, 26 Cal.3d at p. 921 .) It explained: "We have no quarrel with the proposition that a duty of good faith and fair dealing in an insurance policy is a two-way street, running from the insured to his insurer as well as vice versa [citations]. However, what that duty embraces is dependent upon the nature of the bargain struck between the insurer and the insured and the legitimate expectations of the parties which arise from the contract." ( Id. at p. 918.) "The object of the excess insurance policy is to provide additional resources should the insured's liability surpass a specified sum. The insured owes no duty to defend or indemnify the excess carrier; hence, the carrier can possess no reasonable expectation that the insured will accept a settlement offer as a means of `protecting' the carrier from exposure. The protection of the insurer's pecuniary interests is simply not the object of the bargain." ( Id. at p. 919.) It concluded: "If an excess carrier wishes to insulate itself from liability for an insured's failure to accept what it deems to be a reasonable settlement offer, it may do so by appropriate language in the policy." ( Id. at p. 921.)

Admittedly, the discussion in Commercial Union was not framed in terms of mitigation of loss. Nevertheless, Commercial essentially was seeking to recover against Safeway for negligent failure to mitigate the loss; the Supreme Court ruled that Safeway had no duty in this regard. We can only conclude that the California Supreme Court has refused to recognize a duty to mitigate the loss under a third party liability policy.

(16) This is not to say that a liability insurer is wholly at the mercy of an improvident insured. The insured cannot recover any portion of the loss that it has caused willfully. (Ins. Code, § 533; see generally Shell Oil Co. v. Winterthur Swiss Ins. Co., supra, 12 Cal.App.4th at pp. 739-743.) Thus, the insurer's liability is appropriately confined to losses caused by the negligence of its insured. In this case, the jury specifically found that "the contamination arising from the selection, design and construction of the Stringfellow Site [did not] result from a `willful act' of the State within the meaning of Insurance Code section 533 . . . ."

In their reply brief, the Insurers contend for the first time that the State's failure to mitigate was, in fact, willful. They forfeited this contention by failing to raise it in their opening brief. ( People v. Zamudio (2008) 43 Cal.4th 327, 353-354 [ 75 Cal.Rptr.3d 289 , 181 P.3d 105 ]; Paulus v. Bob Lynch Ford, Inc. (2006) 139 Cal.App.4th 659, 685 [ 43 Cal.Rptr.3d 148 ].) Moreover, they had already forfeited it by failing to raise it below, at least in connection with the issue of failure to mitigate. The jury was instructed on Insurance Code section 533; the Insurers' briefs fall short of showing that they were, in fact, precluded from asking the jury to find that the State's failure to mitigate the loss was willful.

We therefore conclude that the trial court correctly ruled that the State had no duty to the Insurers to mitigate the loss, by undertaking reasonable remediation measures or otherwise.

Because we are reversing the judgment and remanding for further proceedings, we must consider Stonebridge's protective cross-appeal.

The State claimed to be insured under a lost policy issued by Beneficial Fire & Casualty Insurance Company (Beneficial), Stonebridge's predecessor in interest. It filed a motion in limine seeking the admission of two particular documents to prove the existence and terms of the lost policy:

1. The Colvin memo: A memo dated June 17, 1965, from Dorothy Colvin, an employee of one of the State's insurance brokers. The memo pertained to flood damage that had occurred in 1964. It was sent to the State's brokers as well as to a list of 12 "Interested Companies," including "Beneficial Fire & Casualty, #11694," Wausau, "#XXXXXXXXXXXX," and "North Star . . ., #XSX5094."

Two earlier letters were attached to the Colvin memo. However, they were not particularly relevant to the lost policy issue or to any other issue. Accordingly, we need not analyze them separately.

2. The schedule of policies: A letter dated May 24, 1966, from a consortium of insurance brokers who had been retained by the State. It listed the same 12 insurance companies, with the same policy numbers; it indicated that collectively, they were providing the State with a total of $35 million in excess liability insurance. Thus, the list included Beneficial, which was listed as having issued policy No. 11694, in the amount of $2.05 million; Wausau, which was listed as having issued policy No. 063700030896, with limits of $2 million; and North Star, which was listed as having issued policy No. NSX 5094, with limits of $5 million.

The State argued that the documents were admissible under the business records exception to the hearsay rule. (Evid. Code, § 1271.) It relied largely on the deposition testimony of John Dattner, an officer of General Reinsurance Corporation (General Re), to the effect that these two documents had been found in General Re's underwriting file.

Stonebridge objected that the State had failed to lay a sufficient foundation for the admission of the documents under the business records exception. The trial court sustained the objection. However, it also granted the State's request to reopen Dattner's deposition.

At the reopened deposition, Dattner identified the documents at issue as records from General Re's underwriting files. He also testified that, when the State first made a claim against General Re, he reviewed General Re's underwriting files to determine what coverage it had provided. At that time, he "relied on the accuracy of the underwriting files . . . ." "[O]n [a] routine basis [as] part of [his] job [he] depend[ed] on the reliability of the underlying records maintained by General Re . . . ." On cross-examination, however, he also testified that he did not rely on these particular documents in determining the scope of General Re's coverage"they really had nothing to do with the claim in question."

Dattner further testified that it was his "understanding" that the documents at issue were "prepared or received in the regular course of . . . General [Re's] . . . business." He admitted, however, that he had no personal knowledge concerning the preparation or receipt of the documents.

Stonebridge renewed its hearsay objection. This time, it argued that the State had failed to lay a sufficient foundation under either the business records exception or the ancient documents exception to the hearsay rule. (Evid. Code, § 1331.) The trial court then admitted the documents under the ancient documents exception.

The jury found, by special verdict, that Beneficial had issued a liability policy to the State with policy limits of $2.05 million, covering the period from September 20, 1964, through September 20, 1966.

Stonebridge contends that the trial court erred by admitting hearsay evidence of its lost policy.

(17) Preliminarily, the State claims that Stonebridge failed to object to the admission of the challenged documents. Not so. As we have already discussed, Stonebridge did object, repeatedly and in writing. It is true that, after the trial court had already overruled these objections, the State sought leave to introduce the documents as a single exhibit; counsel for Stonebridge responded, "No objection, your Honor." However, this particular failure to object was clearly limited to the form of the exhibit, not its content. The trial court had already ruled that the documents themselves were admissible; Stonebridge was not required to object to them again. "`. . . "`An attorney who submits to the authority of an erroneous, adverse ruling after making appropriate objections or motions, does not waive the error in the ruling by proceeding in accordance therewith and endeavoring to make the best of a bad situation for which he was not responsible.'" [Citation.]' [Citation.]" ( Park City Services, Inc. v. Ford Motor Co., Inc. (2006) 144 Cal.App.4th 295, 311 [ 50 Cal.Rptr.3d 373 ] [Fourth Dist., Div. Two], quoting Mary M. v. City of Los Angeles (1991) 54 Cal.3d 202, 212-213 [ 285 Cal.Rptr. 99 , 814 P.2d 1341 ], quoting People v. Calio (1986) 42 Cal.3d 639 [ 230 Cal.Rptr. 137 , 724 P.2d 1162 ], quoting Leibman v. Curtis (1955) 138 Cal.App.2d 222, 225 [ 291 P.2d 542 ].)

We turn, then, to the merits.

(18) "In construing statutes, `our fundamental task is "to ascertain the intent of the lawmakers so as to effectuate the purpose of the statute." [Citations.] We begin by examining the statutory language because it generally is the most reliable indicator of legislative intent. [Citation.] We give the language its usual and ordinary meaning, and "[i]f there is no ambiguity, then we presume the lawmakers meant what they said, and the plain meaning of the language governs." [Citation.] If, however, the statutory language is ambiguous, "we may resort to extrinsic sources, including the ostensible objects to be achieved and the legislative history." [Citation.] Ultimately we choose the construction that comports most closely with the apparent intent of the lawmakers, with a view to promoting rather than defeating the general purpose of the statute. [Citations.]' [Citation.]" ( Mays v. City of Los Angeles (2008) 43 Cal.4th 313, 321 [ 74 Cal.Rptr.3d 891 , 180 P.3d 935 ].)

(19) Somewhat surprisingly, since Evidence Code section 1331 became effective, in 1967, it does not appear to have been construed in any published decisions. It provides: "Evidence of a statement is not made inadmissible by the hearsay rule if the statement is contained in a writing more than 30 years old and the statement has been since generally acted upon as true by persons having an interest in the matter." (Evid. Code, § 1331.) Thus, the "statement" is whatever the proponent of the evidence is seeking to introduce. (See also Evid. Code, § 225 [defining "statement" as including oral or written verbal expression].) It must have been generally acted upon as true. It also must be contained in a writing that is more than 30 years old. (See also Evid. Code, § 250 [defining "writing" as any means of recording communication upon a tangible thing].)

Here, what the State was seeking to introduce were the Colvin memo and the schedule of policies. Thus, these were the relevant "statements" that had to have been generally acted upon as true. There was no evidence, however, that this was the case. Dattner testified that he considered the underwriting file, in general, to be reliable. He admitted, however, that he had never been called upon to rely on either the Colvin memo or the schedule of policies. Thus, the State proved, at most, that if Dattner had been called upon to rely on these particular statements, he probably would have done so. This falls short of showing that anyone had ever actually acted upon them as true.

Moreover, the State had to prove that Dattner (or General Re) was a "person having an interest in the matter." The "matter" refers back to the statement and its truth or falsity. In other words, the persons who must have acted on the statement as true must have been persons who had an interest in whether it was true or not. Here, there was no evidence that the truth of the statements in either the Colvin memo or the schedule of policies mattered in any way to either Dattner or General Re. To the contrary, Dattner testified that he was concerned with the scope of General Re's own coverage, and these particular documents were irrelevant to that question.

The State argues that the State itself acted on the statements as true, i.e., by "suing Stonebridge under the B[eneficial] policy." However, there was no evidence that the State did, in fact, rely on the statements in filing suit. A more fundamental problem is that this is bootstrapping. Virtually every time a party offers a statement into evidence under the ancient documents exception, it is, in some sense, acting on the statement as true. This cannot be enough, standing alone, to satisfy the exception.

The State also argues that its brokers acted on the statements as true. Once again, there was no evidence of this. What the State appears to mean is that the authors of the documents inferably believed that their own statements were true. But this is just another attempt at bootstrapping. The maker of a false statement is just as likely to communicate it to others as the maker of a true statement, if not more so.

Because the statute is not ambiguous, we are not called upon to consider either its purpose or its legislative history. If we were to consider them, they arguably might point toward a different outcome. For example, Evidence Code section 1331 was derived from Code of Civil Procedure former section 1963, subdivision 34, which dealt with the authentication of ancient documents; as a result, a plausible argument could be made that, using similar language, the drafters of Evidence Code section 1331 did not intend to require any more than that the document as a whole have been treated as reliable. We also have some concern that our interpretation of the statute could render the ancient documents exception unavailable in many cases in which the evidence is needed and there is no real threat of falsehood. [22] For example, the State argues that our interpretation gives insurers an incentive to not keep (or even to destroy) copies of older policies. Nevertheless, we are bound by the Legislature's choice of language. If it wants the exception to have broader scope, it is free to amend Evidence Code section 1331.

The State's fallback argument is that the documents were admissible under the business records exception. That exception provides: "Evidence of a writing made as a record of an act, condition, or event is not made inadmissible by the hearsay rule when offered to prove the act, condition, or event if:

There was no evidence, however, that the documents were, in fact, made in the regular course of a business. Dattner testified that they were received and kept in the regular course of General Re 's business, but this was not sufficient to satisfy the statutory requirement. There was also no evidence that the documents were made at or near the time of the conditions that they purported to record. Dattner was not qualified to testify concerning the mode of preparation of the documents, nor did he attempt to do so. Thus, there was likewise no evidence that the sources of information, method of preparation, and time of preparation of the documents indicated that they were trustworthy. (See People v. Matthews (1991) 229 Cal.App.3d 930, 939-940 [ 280 Cal.Rptr. 134 ] [rap sheets were not admissible under business records exception where custodian could not testify to their sources of information or mode of preparation].) As the State argues, a person other than the creator of a document may be qualified to lay the foundation for the admission of the document as a business record. (See, e.g., People v. Remiro (1979) 89 Cal.App.3d 809, 846 [ 153 Cal.Rptr. 89 ].) Dattner, however, was not such a person.

Finally, we cannot say that the error in admitting these documents was harmless. (See Evid. Code, § 353, subd. (b).) They constituted essentially the only evidence of the existence and of the terms of the Stonebridge policy. Thus, it is at least reasonably probable that, if these documents had been excluded, the result would have been more favorable to Stonebridge. (See Cassim v. Allstate Ins. Co. (2004) 33 Cal.4th 780, 800 [ 16 Cal.Rptr.3d 374 , 94 P.3d 513 ].) Indeed, the State does not argue otherwise. Accordingly, for this reason, too, we would have to reverse the judgment as between the State and Stonebridge.

The judgment is reversed and the matter is remanded for further proceedings. Stonebridge is awarded costs on appeal against the State. The State is awarded costs on appeal against the other Insurers.

Ramirez, P. J., and Hollenhorst, J., concurred.

[*] Pursuant to California Rules of Court, rules 8.1105(b) and 8.1110, this opinion is certified for publication with the exception of parts V. and VIII.C.

[1] Stonebridge disputes the existence of the alleged policy.

[2] Actually, at the time, the trial court left it open to the State to prove that the 1969 storm overflow constituted a separate occurrence. Later, however, the parties stipulated that the 1969 storm overflow did not constitute an occurrence.

[3] Once again, Wausau has not cross-appealed. Accordingly, all references to the "Insurers" in the context of the cross-appeal exclude Wausau.

[4] The Insurers proceed to cite out-of-state cases rejecting the all-sums approach. However, they do not tell us how many out-of-state cases accept it. Thus, they fail to support their assertion that a "majority of jurisdictions" reject it. In 2004, one journal article reported that "states making the all-sums determination are currently in the majority." (Smith, Environmental Cleanup and the Interpretation of Comprehensive General Liability Insurance Policies: A Lesson from the Oregon Legislature (2004) 31 J. Legis. 217 , 219.)

[5] Sybron Transition Corp. v. Security Ins. (7th Cir. 2001) 258 F.3d 595 is sometimes cited as rejecting stacking. There, however, the parties agreed that, under New York law ( id. at p. 597), the successive insurers' liability had to be prorated based on time on the risk. The court referred briefly to "[s]tacking (aka joint and several liability)" then stated, "No matter what the right name of this possibility, it is antithetical to a time-on-the-risk approach." ( Id. at pp. 600-601, italics omitted.) Thus, it failed to recognize that it is possible to have "joint and several liability" (i.e., the all-sums approach) without stacking. At most, Sybron stands for the banal proposition that stacking is not allowed in a pro rata jurisdiction. This sheds no light on the question before uswhether stacking is allowed in an all-sums jurisdiction.

[6] In the following table, totals may appear to be off by plus or minus one, due to rounding.

[7] Wausau issued four policies, covering policy periods 1964 to 1967, 1967 to 1970, 1970 to 1973, and 1973 to 1976, respectively; two of these policies had the same policy number. Nevertheless, Wausau does not argue that these policies were subject to just a single policy limit because they constituted only a single continuous contract that was repeatedly renewed. (See generally A.B.S. Clothing Collection, Inc. v. Home Ins. Co. (1995) 34 Cal.App.4th 1470 [ 41 Cal.Rptr.2d 166 ].) We deem any such contention forfeited.

[8] In the trial court, the Insurers argued that, because the State required them to use its own master policy form, the State should be regarded as the drafter of the policies, and therefore any ambiguities should be resolved against the State. However, they have not reiterated this argument on appeal. In any event, it seems clear that the State's master policy form incorporated standard policy language originally drafted by insurance industry representatives. (See Montrose, supra, 10 Cal.4th at pp. 671-673.)

[9] The Insurers claim that an earlier case, Armstrong, supra, 45 Cal.App.4th 1 , "rejected stacking." Not so. There, it was the trial court that had rejected stacking. The appellate court noted, accurately, that this "decision . . . is supported by Keene . . . ." ( Id. at p. 50, fn. 15.) However, it also noted that the appellants had not challenged this ruling. ( Ibid. ) Hence, it had no need to address stacking.

[10] Wausau cites Employers Ins. of Wausau v. Granite State Ins. Co. (9th Cir. 2003) 330 F.3d 1214 ( Wausau ) for the proposition that stacking is not allowed in the absence of a stipulation. In Wausau, the issue was not stacking of limits across policy periods; rather, it was whether the limits under the primary insurer's policies had to be stacked before the excess insurer could be held liable. ( Id. at pp. 1218-1219.) Accordingly, the court should simply have applied California's horizontal exhaustion rule. Instead, it launched into an extensive discussion of Stonewall and FMC . ( Wausau, at pp. 1219-1221.) Ultimately, it held that Stonewall , not FMC , was controlling, because the primary insurer had entered into a stipulation much like the one in Stonewall . ( Wausau, at p. 1220.) A dissenting justice would have held that FMC was controlling. ( Wausau, at pp. 1221-1222 (dis. opn. of Thompson, J.).)

Obviously, as a decision of an intermediate federal court, Wausau is not binding on us. ( People v. Weeks (2008) 165 Cal.App.4th 882, 888 [ 71 Cal.Rptr.3d 257 ].) Because we believe that it misconceived the issue before it, and it therefore incorrectly regarded Stonewall and FMC as the controlling authorities, we do not find it helpful.

[11] Keene forthrightly acknowledged that this is a problem with an antistacking rule: "[B]ecause we hold . . . that only one policy's limits of liability may apply to one injury, an insured who has purchased several policies that cover an injury will only be able to collect under one of those policies, even though he paid for several. Therefore, it is the insurernot the insuredwho reaps the most benefit . . . ." ( Keene, supra, 667 F.2d at p. 1049, fn. omitted.)

[12] A self-insured retention is simply a variation on a deductible. "[Self-insured retentions (SIR's)] are more common in commercial liability insurance than deductibles (which are more common in personal liability and first party property insurance). The difference between them is that the policy limits apply on top of an SIR (e.g., if the policy limit is $500,000 and there is a $50,000 SIR, after the SIR is exhausted the insured has the full $500,000 coverage). A deductible, however, reduces policy limits (e.g., if there is a $500,000 policy limit and a $50,000 deductible, the insured has only $450,000 coverage)." (Croskey et al., Cal. Practice Guide: Insurance Litigation, supra, ¶ 7:384 (rev. #1, 2005).)

[13] The Safeco court seems to have implicitly assumed that the four annual policies actually constituted a single policy, annually renewed. Hence, there was only a single promise to pay up to $500,000 per occurrence, rather than four separate promises to pay up to $500,000 per occurrence. (Cf. A.B.S. Clothing Collection, Inc. v. Home Ins. Co., supra, 34 Cal.App.4th at pp. 1475-1479.)

Here, each policy was issued by a separate insurer, except for the series of policies issued by Wausau; as we noted in footnote 7, ante, Wausau has not argued that its policies constituted a single continuous contract. Accordingly, we are justified in concluding that each policy constituted a separate promise to pay up to the policy limits per occurrence.

[14] The State also listed, as a fifth occurrence, the 1978 overflow of the dam. Later, however, the trial court, acting on a motion for summary adjudication, ruled that the 1978 overflow was not an occurrence, because it was expected and intended rather than accidental. The State has not challenged that ruling.

[*] See footnote, ante, page 1114.

[17] The State claims that this estimate was wildly mistaken, because at the time no one knew how grievously flawed the site really was. For present purposes, however, all that matters is that the Insurers have a colorable claim that the cost increased by some amount. The State thus goes on to concede that "there is little point in fully addressing this factual issue" in this appeal.

[18] The requested instruction provided: "If Defendants breached their contracts and the breach caused harm, the State is not entitled to recover damages for harm that Defendants prove the State could have avoided with reasonable efforts or expenditures. You should consider the reasonableness of the State's efforts in light of the circumstances facing it at the time, including its ability to make the efforts or expenditures, without undue risk of hardship.

"If the State made reasonable efforts to avoid harm, then your award should include reasonable amounts that it spent for this purpose."

[19] The Insurers themselves assert that cases dealing with the carrot version are "[i]napplicable" here.

[20] A "sue and labor" clause requires the insured to protect, rescue, and repair the insured property in the event of an insured peril and requires the insurer to pay the costs of doing so. Such clauses are most common in marine insurance policies. (12 Couch on Insurance, supra, § 183:162, pp. 183-114 to 183-116.)

[21] Although this was an alternative holding, it was not dictum. "`[I]t is well settled that where two independent reasons are given for a decision, neither one is to be considered mere dictum, since there is no more reason for calling one ground the real basis of the decision than the other. The ruling on both grounds is the judgment of the court and is of equal validity. [Citations.]' [Citations.]" ( Southern Cal. Ch. of Associated Builders etc. Com. v. California Apprenticeship Council (1992) 4 Cal.4th 422, 431, fn. 3 [ 14 Cal.Rptr.2d 491 , 841 P.2d 1011 ], quoting Bank of Italy etc. Assn. v. Bentley (1933) 217 Cal. 644, 650 [ 20 P.2d 940 ].)

[22] Interestingly, the analogous federal rule does not require that the statement (or even the writing) must have been acted on as true. (Fed. Rules Evid., rule 803(16), 28 U.S.C.)

[*] See footnote, ante, page 1114.