*122 KISTLER, J.
This case arises out of a dispute over insurance coverage for plaintiffs’ ship dismantling business. Before this court, the parties have raised primarily two issues. The first is whether plaintiffs or defendants had the burden to prove that environmental damages resulting from the operation of plaintiffs’ business were neither expected nor intended. On that issue, the Court of Appeals held that, when the insurance policies that defendants issued expressly granted coverage only for unexpected and unintended damages, plaintiffs had the burden of proving that their loss came within that coverage. ZRZ Realty v. Beneficial Fire and Casualty Ins., 222 Or App 453, 472-73, 194 P3d 167 (2008), modified on recons, 225 Or App 257, 201 P3d 912 (2009). However, the court also held that, when the insurance policies granted broad coverage subject to an implied limitation for expected or intended damages, the implied limitation functioned as an exclusion that defendants had the burden to prove. 222 Or App at 474-75.
The second issue that the parties have raised involves protection and indemnity policies in which defendants agreed to indemnify plaintiffs for liability they incurred for “damages to any harbor, dock, * * * buoy, telegraph cable or other fixed or moveable thing whatsoever.” The Court of Appeals held that the promise to indemnify plaintiffs for “damages to any * * * other fixed or moveable thing whatsoever” did not include a promise to indemnify plaintiffs for damages to the riverbed. Id. at 488-91. Rather, according to the Court of Appeals, defendants agreed to indemnify plaintiffs only for damage to artificial structures. Id. We allowed plaintiffs’ petition for review and now affirm the Court of Appeals decision allocating the burden of proof but reverse its decision regarding coverage for damage to the riverbed.
A. Facts
Plaintiffs (collectively Zidell) are a group of related companies that began acquiring and dismantling decommissioned navy and merchant marine ships after World War II. 1 *123 Zidell towed the decommissioned ships to a site along the Willamette River in Portland (the Moody Avenue site) where it dismantled the ships for scrap, generally while the ships were tied to a dock at the site. Dismantling the ships resulted in the release of pollutants found in, among other places, the ships’ paint, batteries, motors, and fuel tanks. The pollutants included polychlorinated biphenyls, petroleum products in the form of fuel, lubricating, and hydraulic oils, and a variety of metals, such as arsenic, cadmium, chromium, lead, mercury, and zinc. Some of the pollutants were released directly into the river; others contaminated the land on which Zidell operated its business and, from there, leached into the groundwater and the river.
Over the years, Zidell purchased different types of insurance from a variety of insurers. Only two insurers, Certain Underwriters at Lloyd’s of London and Certain London Market Insurance Companies, remain in this litigation; the others have settled. The interests of the two remaining insurers are, for the purposes of this appeal, identical, and we refer to them collectively as London. From 1956 to 1983, Zidell purchased three types of insurance policies from London that are at issue in this litigation: comprehensive general liability policies, a form of marine excess coverage known as bumbershoot policies, and another form of marine insurance known as protection and indemnity policies. In setting out the facts, we discuss the comprehensive general liability policies and refer to the other policies where appropriate.
The comprehensive general liability policies consisted of primary and excess policies that provided coverage for damage to property but contained an owned property exclusion. See Schnitzer Investment Corp. v. Certain Underwriters, 341 Or 128, 138-39, 137 P3d 1282 (2006) (discussing effect of owned property exclusion on coverage for environmental contamination). From 1956 to 1965, London issued a series of comprehensive general liability policies that the parties refer to as the “implied fortuity policies.” By their terms, those policies provided coverage without regard to whether the property damage that resulted from Zidell’s business activities was either expected or intended. One policy, for example, provided coverage for “any and all liability imposed *124 by law against the Assured for loss of or damage to or destruction of the property of others * * * arising from any cause whatsoever out of the operation” of Zidell’s business.
Despite that broad grant of coverage, the trial court read a limitation into the implied fortuity policies. Specifically, the trial court ruled that “the general principle that only ‘fortuitous’ losses will be covered will be deemed a condition of each such policy, consistent with the holding in A-1 Sandblasting & Steamcleaning Co. v. Baiden, 293 Or 17, 643 P2d 1260 (1982).” As the trial court explained, that principle
“limits coverage to those losses that were not expected or intended for policies between 1957 and January 1, 1968. From January 1, 1968 until July 1, 1983, the implied fortuity doctrine limits coverage to those losses that were not intended.” 2
Both parties accept the limitation that the trial court read into the agreements that they entered into before January 1, 1968. 3
Starting in 1966, London began issuing comprehensive general liability policies to Zidell that the parties refer to as the “express fortuity policies.” Those policies provided coverage only for damages that were both unexpected and unintended. For example, in one of those policies, London agreed to pay Zidell “all sums which the Assured shall be obligated to pay by reason of the liability [ijmposed upon the Assured by law * * * for damages * * * on account of * * * [p]roperty damage * * * caused by or arising out of [an] occurrence.” The policy defined “occurrence” as an “accident or a happening or *125 event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally results in * * * property damage * * * during the policy period.” 4
In 1994, the Oregon Department of Environmental Quality (DEQ) issued a notice to Zidell stating that it was potentially responsible for cleaning up the environmental contamination resulting from its business at the Moody Avenue site. Zidell chose to participate in a voluntary clean-up program that DEQ offered and also requested that London, pursuant to the various policies that it had issued over the years, defend Zidell and indemnify it for the costs of remedi-ating any environmental damage that had resulted from Zidell’s business. When London denied that it had any obligation to defend or indemnify Zidell, Zidell brought this action, alleging breach of contract and also seeking a declaration of London’s obligations under its policies. 5
Before trial, the court ruled on summary judgment that London had a duty to defend Zidell in response to DEQ’s notice that Zidell was potentially responsible for cleaning up the environmental damage resulting from its business. The court also ruled, on summary judgment, that London bore the burden of proving that Zidell either expected or intended the environmental damages that resulted from the operation of its business. 6 The court explained that its ruling applied to both the express and the implied fortuity policies.
With those ground rules in place, the case proceeded to trial on the question of indemnity. Among other things, the parties and the trial court sought to identify the specific contaminants that, as a result of Zidell’s business, had damaged the environment over the period of time in which London had insured Zidell. The trial court also sought to determine, *126 regarding each pollutant, whether Zidell had either expected or intended the resulting damage. 7 For example, the trial court found that Zidell had never intended any damage to the environment from removing anti-fouling paint (paint used to discourage the growth of organisms on the ships’ hulls) but that “Zidell management expected damage to sediments from [arsenic found in] anti-fouling paint beginning in 1972.” Because 90 percent of the arsenic found at the site came from removing the anti-fouling paint, the trial court sought to allocate London and Zidell’s respective responsibility for the cost of remediating the damage accordingly. 8
At the conclusion of the trial, the court issued lengthy findings of fact and conclusions of law, which it incorporated in the judgment. In essence, the court found that, for a number of years, Zidell neither expected nor intended that its business operations would result in environmental damage; however, the court found that, beginning in the 1970s, Zidell expected that some of the substances released as a result of dismantling the ships would damage the soil and also the sediment in the river. 9 As a result of those and other findings, the trial court held that London’s policies required it to indemnify Zidell for part of the remediation costs. 10
As noted, the Court of Appeals affirmed the trial court’s ruling regarding the burden of proof on the implied *127 fortuity policies but reversed its ruling regarding the burden of proof on the express fortuity policies. The Court of Appeals reasoned that the express fortuity policies offered coverage only for unexpected and unintended damages and that Zidell had the burden to prove that the damages resulting from its business came within the scope of that coverage. The Court of Appeals also reversed the trial court’s ruling that the protection and indemnity policies that London had issued covered damages to the riverbed. We allowed Zidell’s petition for review to consider those rulings and begin with the question of the burden of proof.
B. Burden of Proof
On appeal, and again on review, the parties debate whether the trial court correctly allocated the burden of proof. Both parties start from the proposition that the insured (Zidell) has the burden to prove coverage while the insurer (London) has the burden to prove an exclusion from coverage. Compare Stanford v. American Guaranty Life Ins. Co., 280 Or 525, 527, 571 P2d 909 (1977) (insurer has the burden to prove an exclusion), with Lewis v. Aetna Insurance Co., 264 Or 314, 316, 505 P2d 914 (1973) (insured has the burden to prove coverage). They disagree, however, whether the requirement that any damage to the environment be neither intended nor expected is part of a limited grant of coverage to which the insured would have to prove entitlement or an exclusion from a broad grant of coverage for which the insurer would have to prove justification. In analyzing that issue, we begin with the express fortuity policies and then turn to the implied fortuity policies.
1. Express Fortuity Policies
The express fortuity policies provide coverage for “[property damage * * * caused by or arising out of [an] occurrence.” They define an occurrence as an “accident or a happening or event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally results in * * * property damage * * * during the policy period.” Reading those two provisions together, London reasons that the express fortuity policies provide coverage only for unexpected and unintended property damage. London argues that the *128 parties to an insurance agreement may structure their agreement in different ways. They may choose to grant limited coverage (coverage only for unexpected and unintended losses), or they may choose to grant broad coverage subject to specific exclusions. London contends that, when, as in this case, the parties choose the former course, the insured has the burden to prove that the loss comes within the limited coverage that the policy grants. Conversely, if the parties chose the latter course, London does not dispute that the insurer would have the burden to prove that the exclusion applies.
Zidell takes a different tack. It contends that whether the parties choose to grant limited coverage or broad coverage subject to an exclusion is not dispositive. It contends that some provisions are, in essence, exclusions while others are, in essence, grants of coverage. Zidell reasons that courts should designate provisions according to their essence and without regard to the way that the parties structure their agreement. Zidell acknowledged at oral argument that sometimes limited coverage is just that — coverage for bodily injury, for example — and that the insured would bear the burden to prove that the loss came within the scope of that limited coverage. Zidell contends, however, that this court’s cases support its position that limitations on unexpected and unintended damages are, in essence, exclusions and should be treated as such.
Zidell’s position is difficult to square with established principles of Oregon contract interpretation. In drafting the express fortuity policies, the parties chose to provide coverage only for unexpected and unintended damages. They did not list expected or intended damages as an exclusion. One of the express fortuity policies, for example, sets out a list of specific exclusions under the heading, “THIS INSURANCE IS SUBJECT TO THE FOLLOWING EXCLUSIONS.” (Capitalization in original.) Neither expected nor intended damages is included on that list of exclusions. To accept Zidell’s argument, we would have to convert what the parties chose to describe as a limitation on coverage into an exclusion from coverage, even though the parties did not include “expected or intended damages” as one of the listed exclusions. In other words, we would have to rewrite the *129 terms of the parties’ agreement, something that courts ordinarily lack authority to do. See Usinger v. Campbell, 280 Or 751, 755, 572 P2d 1018 (1977) (courts may not rewrite contracts); Wikstrom v. Davis, 211 Or 254, 268, 315 P2d 597 (1957) (same); cf. Frick v. Hoag, 277 Or 135, 138, 559 P2d 827 (1977) (noting that courts have equitable authority to reform contracts in limited circumstances).
Zidell identifies no reason why the parties to an insurance contract may not choose either to grant coverage only for unexpected and unintended damages or to grant broad coverage subject to an exclusion for expected or intended damages. Cf. Robert E. Keeton, Insurance Law § 5.2(a) 275 (1971) (describing that decision as a “draft[er’s] choice”). Put differently, there is no reason, either logically or semantically, why a limitation on unexpected and unintended damages may not be stated either as a grant of limited coverage or separately as an exclusion from a broad grant of coverage. Nor are we able to identify some neutral principle that would allow us to discern which provisions are essentially grants of limited coverage and which are essentially exclusions from a broad grant of coverage. If the parties structure their agreement to provide limited coverage, as they did here, then our cases place the burden on the insured to prove that the event that triggered the loss came within the scope of that coverage.
Zidell contends, however, that three of this court’s cases demonstrate that the limitation on coverage in the express fortuity policies constitutes an exclusion from coverage. Zidell relies initially on Smith v. Ind. Hosp. Assn., 194 Or 525, 242 P2d 592 (1952), for the proposition that insurers have the burden to prove policy provisions that are “defensive in character.” We agree with Zidell that Smith stands for the proposition that insurers have the burden to prove exclusions from coverage. See Stanford, 280 Or at 527 (citing Smith for the proposition that “[t]he insurer has the burden to prove that the loss is excluded”). However, we do not read Smith more broadly than that.
In Smith, the employer’s health insurance policy provided coverage to employees for acute sickness and chronic conditions, subject to separate exemptions. The two *130 provisions at issue in Smith were an exemption for preexisting conditions and one for new employees whose chronic condition “acutely manifested]” itself within the first six months of work. 11 And the specific question in Smith was whether the trial court erred in denying the insurer’s motion for a directed verdict on the ground that no reasonable juror could find that the exemptions did not apply. In the course of resolving that issue, the court noted that the burden was on the insurer to prove that the exemptions applied. 12 The policy in Smith did not provide limited coverage, as the express fortuity policies do in this case. Rather, the policy in Smith provided broad coverage subject to specific exemptions or exclusions, and the court’s statement regarding the burden of proof cannot be separated from the issue before the court in Smith.
Zidell also cites two cases in which the policies were similar to the express fortuity policies in this case. See Ledford v. Gutoski, 319 Or 397, 877 P2d 80 (1994); Nielsen v. St. Paul Companies, 283 Or 277, 583 P2d 545 (1978). In both Ledford and Nielsen, the insurance policies provided coverage for bodily and property damage that resulted from an occurrence, which the policies defined as an “accident, including continuous or repeated exposure to conditions, which results in bodily injury or property damage neither expected *131 nor intended.” Ledford, 319 Or at 399; Nielsen, 283 Or at 279. Neither case, however, addressed who had the burden to prove that the damage was expected or intended; that is, neither case sought to determine whether the requirement that damages be neither expected nor intended was part of a limited grant of coverage or an exclusion from a broad grant of coverage. Rather, in both cases, the question was whether the insurer had a duty to defend the insured, which entailed determining only whether the conduct alleged in the complaint necessarily required an inference that the insured had intended the harm.
It is true that, in Nielsen, the court referred to the definition of occurrence as a “policy provision excluding intentional harm.” 283 Or at 281. However, we hesitate to read too much into that passing reference, when the issue before the court had nothing to do with the burden of proof but focused instead on the substantive question whether the allegations in the complaint filed against the insured established a duty to defend. The same is true of Ledford. In that case, the court referred to “various policy provisions excluding coverage for intentionally-caused injuries” in the course of explaining that the provision applied only “when the insured intended to cause the particular injury or harm, as opposed to merely intending the act.” 319 Or at 401. The court made that comment in explaining that, given the allegations in the complaint, the insurer did not have a duty to defend. The court neither addressed nor expressed any opinion on whether the limitation on coverage should be viewed as an exclusion on which the insurer had the burden of proof or a limitation of coverage on which the insured had the burden of proof. In short, this court has neither considered nor resolved the issue that this case presents, and we decline to give off-hand references in our prior cases controlling significance. 13
*132 Zidell presses a broader argument. It contends that “the location of a provision should not control the assignment of the burden of proof.” It reasons that there is no practical difference between a policy that grants coverage for unexpected and unintended damages and one that grants broad coverage subject to an exclusion for expected or intended damages. We agree with Zidell that both policies insure against the same risk. However, our cases allocate the burden of production and persuasion based on whether the policy grants limited coverage or broad coverage subject to an exclusion. We must classify the limitation one way or another, and we conclude that respecting the parties’ choice is the more appropriate method. Not only does it permit the parties to structure their agreements in a way that allocates the burden of proof, but it also avoids putting courts in the difficult position of divining the “essence” of contractual provisions that logically may serve either as a grant of limited coverage or an exclusion from a broad grant of coverage. Accordingly, we hold that, for the express fortuity policies, Zidell had the burden to prove that the damages were neither expected nor intended.
2. Implied Fortuity Policies
The implied fortuity policies present the same question but in a different posture. As noted, the implied fortuity policies contain a broad grant of coverage without any requirement that the damages be unexpected and unintended. For instance, one of the first comprehensive general liability policies that London issued to Zidell provided coverage:
“From and against all loss which the Assured may sustain or incur by reason of or in consequence of:
“Any and all liability imposed by law against the Assured for loss of or damage to or destruction of property of others * * * sustained or alleged to have been sustained during the currency of this certificate and arising from any cause whatsoever out of the operations, activities, work and/or business of the Assured * *
*133 By its terms, the policy covered “[a]ny and all liability imposed by law” without exception. Despite that broad grant of coverage, the trial court read into that policy, and others like it, a limitation that the damages that gave rise to the insured’s liability be neither expected nor intended, at least for those policies issued before 1968. (The court read a less restrictive limitation into the implied fortuity policies issued in 1968 and afterwards.)
According to the trial court, the limitation that it read into the implied fortuity policies derived from what this court described in A-1 Sandblasting and earlier in Isenhart v. General Casualty Co., 233 Or 49, 377 P2d 26 (1962), as a “public policy” limitation. 14 As this court explained in Isenhart, a “public policy” limitation excludes from coverage an event that otherwise would come within the policy’s terms. See Isenhart, 233 Or at 51-52 (“An action for assault and battery would fall within this coverage unless, as [the insurer] asserts, coverage is precluded upon the ground that it would be contrary to public policy.”). For instance, in Isenhart, after reasoning that “a clause in a contract of insurance purporting to indemnify the insured for damages recovered against him as a consequence of his intentional conduct in inflicting injury upon another [wa]s unenforceable” as against public policy, the court explained that “the insurance policy in the present case must, in effect, be regarded as excluding such coverage.” Id. at 53.
It follows, we think, both from the terms of the policies themselves and from the reasoning in Isenhart, that the limitation that the trial court read into the implied fortuity policies functions as an exclusion. It excludes an event that, under the literal terms of the policy, otherwise would be covered. 15 See Cimarron Ins. Co. v. Travelers Ins. Co., 224 Or 57, *134 61, 355 P2d 742 (1960) (explaining, in the context of automobile insurance, that “the effect of an exclusion clause is to deny the protection of the policy to some one who, but for the denial, would be an insured”). Literally and functionally, the limitation that the trial court read into the implied fortuity policies serves as an exclusion from a broad grant of coverage rather than as a grant of limited coverage.
London, however, advances two reasons for saying that Zidell bears the burden of proving that the damages were unexpected and unintended. London argues initially that a statute in effect until 1967 became part of all the pre1968 policies and constituted a limitation on coverage. Alternatively, London contends that, in bringing a declaratory judgment claim, Zidell assumed the burden of proving that its damages were unexpected and unintended. We consider each argument in turn.
Until the legislature repealed former ORS 736.005(1) (1965) in 1967, that statute defined “insurance” as a:
“contract whereby one undertakes to indemnify another against loss, damage or liability arising from an unknown or contingent event, whereby the insured or his beneficiary suffers loss or injury.”
See Or Laws 1967, ch 359, § 704 (repealing former ORS 736.005(1)). London argues that that statute became part of each insurance contract it issued to Zidell before 1968 and that the statute functioned as a limitation on coverage, not as an exclusion. We read the former statute differently.
Former ORS 736.005(1) (1965), the statute on which London relies, was enacted as part of a comprehensive act that the legislature passed in 1917 to regulate the business of insurance. See Or Laws 1917, ch 203, § 1 (defining insurance); Lovejoy v. Portland, 95 Or 459, 460-61, 188 P 207 (1920) (describing the 1917 act that was later codified as ORS chapter 736). The 1917 act “regulate[d] the conditions under which [the] business [of insurance] may be commenced and the manner in which it may be conducted.” Lovejoy, 95 Or at 461. Among other things,
“[that act] regulated] the organization of the insurance department, and it prescribed] the jurisdiction and *135 define[d] the powers of the state insurance commissioner. It deal[t] with the mode of organizing local companies, the admission of foreign companies, the nature of the investments of local and foreign companies, and examinations and reports of companies. It contained] various provisions to insure the solvency of insurance companies, and it afford[ed the] means for excluding companies of impaired or doubtful solvency.”
Id. Only companies that were engaged in the business of insurance were regulated and subject to the act’s provisions. Former ORS 736.010 (1965).
Former ORS 736.005(1) (1965) was central to that regulatory scheme. It defined “insurance” and, by extension, identified those companies that were in the business of insurance and thus subject to both the act’s requirements and the Insurance Department’s jurisdiction. See former ORS 736.005(3)(b) (1965) (defining “company,” “corporation,” and “insurance company,” as used in the act, as all individuals and entities engaged “in the business of insurance”); cf. Hall v. Metropolitan Co., 146 Or 32, 28 P2d 875 (1934) (holding that an annuity did not constitute “insurance” within the meaning of former ORS 736.005(1) (1965) and, for that reason, did not require the Insurance Commissioner to pre-approve the annuity).
Former ORS 736.005(1) (1965) provided a basis for determining which agreements and companies were subject to the 1917 insurance act. However, nothing in that definition of insurance, or the larger act of which it was a part, suggested that the legislature intended either to require that the definition be part of every insurance agreement or to limit the scope of available coverage, as London argues. At most, the definition provided a legislative source from which a substantive “public policy” limitation could be implied. See A-1 Sandblasting, 293 Or at 22 (noting that public policy may be found in legislative enactments). 16 We thus disagree with *136 London’s argument that former ORS 736.005(1) (1965) automatically became part of the insurance policies that London issued and that it served as a limitation on coverage. 17
London advances a second argument. London reasons that, even if it ordinarily would have the burden to prove that the damages were expected or intended, Zidell assumed that burden when it included a declaratory judgment claim in its action against London. Relying on First National Bank v. Malady, 242 Or 353, 408 P2d 724 (1966), and State Farm Fire and Cas. v. Reuter, 299 Or 155, 700 P2d 236 (1985), London contends that, when Zidell alleged in its declaratory judgment claim that the damages caused by its business operations were unexpected and unintended, Zidell assumed the burden of proving those facts.
London reads Malady and State Farm too broadly. More specifically, London’s argument fails to distinguish two related but separate categories of declaratory judgment actions. In the first category of declaratory judgment actions, the party that ordinarily would have the burden of proof on a claim or defense brings a declaratory judgment action to establish its claim or defense. In that situation, the authorities are unanimous that the plaintiff in the declaratory judgment action has the same burden of production and persuasion that it ordinarily would have. See James W. Moore, 12 Moore’s Federal Practice § 57.62[2][c] at 57-134 (3d ed 2009) (stating proposition); Edwin Borchard, Declaratory Judgments 404 (2d ed 1941) (same); Developments in the Law -Declaratory Judgments, 62 Harv L Rev 787, 836 (1949) (same).
In the second category of declaratory judgment actions, the party that ordinarily would not have the burden of proof on a claim or affirmative defense brings a declaratory judgment action either to prove nonliability on a claim or to *137 establish that an affirmative defense does not apply. In considering that category of declaratory judgment actions, the authorities have divided over who bears the burden of production and persuasion. See 12 Moore’s Federal Practice § 57.62[2][c] at 57-134 (recognizing division of authority). Some authorities have reasoned that the party that ordinarily would have the burden of proof retains it, even though that party is now the defendant in the declaratory judgment action. See, e.g., Borchard, Declaratory Judgments at 404-09 (in a declaratory judgment action, the burden of proving operative and affirmative facts rests upon the party that relies upon them); Travelers Ins. Co. v. Greenough, 88 NH 391, 190 A 129 (1937) (same). Other authorities, including this court, have reasoned that, when the parties are transposed, the plaintiff in the declaratory judgment action has the burden to prove its affirmative allegations, even though it ordinarily would not have the burden of doing so. See Charles Allen Wright, Arthur R. Miller, and Mary Kay Kane, 10B Federal Practice & Procedure § 2770 at 677-80 (discussing cases) (1998); Malady, 242 Or at 358 (generally adopting that position but leaving room for some variation when the circumstances warrant it).
London’s argument — that, in alleging a claim for declaratory relief, Zidell assumed the burden to prove that it neither expected nor intended to cause damage to the environment — rests on the assumption that this case comes within the second category of declaratory judgment actions to which Malady applies. That assumption is not correct, however. In this case, Zidell alleged two claims for relief against London. The first claim was for breach of contract; the second sought a declaration that London had a contractual duty to defend and indemnify Zidell. As Zidell’s parallel claims for relief make clear, this case falls in the first category of declaratory judgment cases, in which the parties are not transposed and the burden of proof remains unchanged. For that reason, Malady provides no help to London. 18
*138 Similarly, State Farm, on which London also relies, does not advance its argument. In that case, the insurer brought a declaratory judgment action to establish an affirmative defense (issue preclusion) on which it ordinarily would have had the burden of production and persuasion. 299 Or at 158-60. This court recognized that the insurer had the burden of proof on that defense, without regard to whether it either brought an action seeking a declaration that the defense precluded coverage or waited until the insured sued on the policy and raised the defense as a bar to the insured’s claim. Id. at 166-67 n 9. Put differently, State Farm fell into the first category of declaratory judgment actions in which the use of a declaratory judgment action did not affect the ordinary allocation of the burden of production and persuasion.
It follows from the foregoing discussion that Zidell’s addition of a declaratory judgment claim did not alter the ordinary allocation of the burden of proof and production. Zidell retained its usual burden to prove coverage, and London retained its burden to prove exclusions to that coverage, which, for the implied fortuity policies, meant that London had the burden to prove that the damages were expected or intended. Having considered London’s arguments, we conclude that the trial court and the Court of Appeals correctly held that, for the implied fortuity policies, London had the burden of proving that Zidell either expected or intended the environmental damages resulting from its business operations.
C. Protection and Indemnity Coverage
In 1959, London issued to Zidell the first of several “open cover” marine insurance policies that provided protection and indemnity (P & I) coverage for ships that Zidell purchased for scrap and resale. 19 The policy included a standard form for port risks, which provided that London agreed to indemnify the Assured:
*139 “if by reason of interest in the Vessel the Assured shall become liable to pay and shall pay any sum or sums in respect of any liability, claim, demand, damages, and/or expenses arising from or occasioned by any of the following matters or things during the currency of this Policy, that is to say:
“Loss of or damage to any harbour, dock (graving or otherwise), slipway, way, gridiron, pontoon, pier, quay, jetty, stage, buoy, telegraph cable or other fixed or moveable thing whatsoever, or to any goods or property in or on the same howsoever causedf]”
The trial court found that Zidell’s ship dismantling and ship repair activities contaminated the sediment in the river. It concluded that London’s promise, in the standard port risk form, to indemnify Zidell for liability arising from “damage to any * * * other fixed or moveable thing whatsoever” included a promise to indemnify Zidell for liability for damage to the sediment in the Willamette River that Zidell “incurred by reason of its interest in [the insured ships].” The court explained that
“[t]he condition that liability arise ‘by reason of interest’ in a vessel to establish coverage under Zidell’s ship dismantling coverage is satisfied where materials are released from a covered vessel onto the river such that they end up in the sediments. * * * Once materials are disassociated from a vessel being dismantled, they come under the crane barge coverage, and once they hit land, they come under general liability coverage.”
The Court of Appeals reversed the trial court’s ruling, holding that the phrase “damage to any * * * other fixed or moveable thing whatsoever” did not include damage to soil or sediment in the river. The court’s reasoning consisted of three steps. First, it noted that almost all the specific items that precede the phrase “any * * * other fixed or moveable thing whatsoever” are artificial structures or objects. ZRZ Realty, 222 Or App at 490. Second, it observed that, although “harbour” usually refers to a naturally occurring feature, it also can refer to an artificial structure and concluded that, in this context, the word “most likely refers to something that *140 has been built or constructed.” Id. Finally, having determined that all the specific items that precede the general phrase referred to artificial structures, it concluded that the general phrase “any * * * other fixed or moveable thing whatsoever” was similarly limited. Id. at 490-91. The court reasoned that, at a minimum, river sediment was so dissimilar from the specific listed items that the general phrase “any * * * other fixed or moveable thing whatsoever” did not include it. Id.
On review, Zidell argues that the Court of Appeals read both the specific items and the general phrase too narrowly. London, for its part, urges us to follow the Court of Appeals’ interpretation of the P & I policy. It also advances an alternative argument. It contends that, even if Zidell’s interpretation is correct, it is still not liable to indemnify Zidell unless Zidell incurred the liability in its capacity as the owner of the vessel and in the course of operating the vessel. In considering those issues, we begin with the question whether “damage to any harbour, dock (graving or otherwise), slipway, way, gridiron, pontoon, pier, quay, jetty, stage, buoy, telegraph cable or other fixed or moveable thing whatsoever” includes damage to the riverbed.
On that question, we start from the proposition that the Court of Appeals’ interpretation of the provision is plausible. That is, the word “harbour” can refer to both natural and artificial features, and it is plausible to read the term “harbour,” in context, to refer to artificial features and to read a similar limitation into the general phrase “any * * * other fixed or moveable thing whatsoever.” The question, however, that Zidell’s argument raises is whether its reading of the provision is also plausible and, if it is, whether the remainder of the policy resolves the resulting ambiguity. See Hoffman Construction Co. v. Fred S. James & Co., 313 Or 464, 470, 836 P2d 703 (1992) (describing methodology for interpreting provisions in insurance policies). If, after considering the remainder of the policy, two plausible interpretations remain, then we construe the phrase against the drafter and in favor of the insured. Id.
Ordinarily, we assume that a nonspecific term in a series, such as “any * * * other fixed or moveable thing whatsoever,” shares the same qualities as the specific terms that *141 precede it. See Baker v. City of Lakeside, 343 Or 70, 76, 164 P3d 259 (2007) (discussing ejusdem generis rule). 20 In this case, Zidell starts from the proposition that the list of specific items includes natural features (harbors), artificial structures (docks, slipways, jetties), and specific types of property found in ports (buoys and telegraph cables). 21 One plausible reading of the listed items is that London undertook to indemnify Zidell for liability arising from damage to various types of property found in ports. Viewed in that context, the phrase “any * * * other fixed or moveable thing whatsoever” includes damage to the river sediment. At a minimum, the broad wording of the general phrase makes Zidell’s reading of the provision a plausible one.
We note that, one year before London issued the first P & I policy involved in this case, the United States District Court for the Southern District of Texas held that essentially the same provision in a P & I policy required the insurer to indemnify a barge owner for damage to “a number of shore-side properties” caused by oil spilled from the barge. See Gulf States Marine & Min. Co. v. Norwich Union Fire Ins. Soc., 168 F Supp 863, 865 (SD Tex 1958) (discussed in Alex L. Parks, 2 The Law and Practice of Marine Insurance and Average 957-58 (1987), as an example of the coverage under this clause). The Fifth Circuit reversed but not because it disagreed with the district court’s holding. U. S. Fire Ins. Co. v. Gulf States Marine & Min. Co., 262 F2d 565 (5th Cir 1959). Rather, the Fifth Circuit noted that, according to the terms of the policy, the P & I coverage would apply only if there were no other coverage for the loss. Id. at 568. In that case, however, there was coverage for the damages under a *142 policy that covered the tug towing the barge. Id. That opinion provides further support for Zidell’s reading of the policy.
In sum, we think each party’s interpretation of the insurance provision is plausible, and we look to the remainder of the P & I policy that London issued Zidell to attempt to resolve that ambiguity. In the remainder of the policy, London undertook to indemnify Zidell for liability that it incurred for a series of different types of property loss and harm to persons. The port risks form provides that the underwriters agreed to indemnify Zidell, subject to certain exceptions, for liability that it incurred for: damage to other persons’ vessels and cargo, whether caused directly by a collision or indirectly; damage to other vessels and cargo owned by Zidell; the cost of raising or removing the insured vessel and its cargo should it sink; and damages resulting from the loss of life and personal injury. Nothing in those additional covered risks sheds any light on the nature of the specific risk that the parties chose to cover in the provision at issue here. Cf. St. Paul Fire & Marine Ins. v. Vest Transp., 666 F2d 932, 940-44 (5th Cir 1982) (explaining that the P & I policy in that case covered a series of discrete risks). Because the remainder of the policy does not remove any ambiguity in the meaning of the provision, we interpret the phrase “any * * * other fixed or moveable thing whatsoever” in favor of the insured Zidell to include damage to the sediment in the river.
London advances an alternative argument. It contends that, even if the phrase “any * * * other fixed or moveable thing whatsoever” includes damage to the riverbed, the terms of the policy and general principles of marine insurance make clear that London agreed to indemnify Zidell only for liabilities that Zidell incurred (1) as the owner of the insured vessel and (2) in the course of operating the vessel. The argument that London advances raises two issues, and we consider them separately.
London contends that, under the terms of the policy, it agreed to indemnify Zidell only to the extent that Zidell incurred liability as the owner of the insured vessel. London argues, however, that “Zidell’s liability in this case does not arise by reason of its interest in any vessel; it arises by reason of its interest in the Moody Avenue site.” As we understand *143 London’s argument, London does not dispute that Zidell owned or had an interest in the vessels it dismantled. Rather, London’s first argument rests on the proposition that the trial court imposed liability on London under the P & I policies for damages that Zidell incurred in some capacity other than its capacity as the owner of the vessels. We do not read the trial court’s ruling as broadly as London does.
The third paragraph of the port risk form provides that London has a duty to indemnify Zidell “if by reason of interest in the [insured] Vessel the Assured shall become liable to pay and shall pay any sum or sums in respect of any liability * * * arising from or occasioned by” a series of specified “matters or things,” one of which is “damage to any harbour, dock, * * * or other fixed or moveable thing whatsoever.” In discussing London’s obligation under that provision, the trial court explained that “[t]he condition that liability arise ‘by reason of interest’ in a vessel to establish coverage under Zidell’s ship dismantling coverage is satisfied where materials are released from a covered vessel onto the river such that they end up in the sediments.” The trial court also explained that, once the materials were “disassociated” from a vessel being dismantled, different insurance policies covered the damages.
London has not explained why the trial court erred in drawing the distinction that it did or why Zidell’s liability for materials released onto the river from its vessels would not constitute liability that arose by reason of its interest in those vessels. Nothing that London has offered in its briefs or at oral argument persuades us that the trial court imposed greater liability on London under the P & I policies than the terms of those policies warranted. 22
*144 London raises a related but separate argument. It contends that, “[w]ith respect to P & I insurance, a well-established and uniform rule of maritime law disposes of Zidell’s claim — a rule that requires a causal nexus between a vessel’s operation and the liability for which coverage is sought.” London reasons that, because Zidell was not operating vessels but was instead scrapping them, Zidell did not incur liability by reason of its operation of a vessel and therefore cannot come within the terms of the P & I coverage.
London’s argument construes P & I coverage too narrowly, both generally and in this case. P & I coverage is essentially marine liability coverage. It arose initially to provide coverage for liability that an insured vessel incurred as a result of a collision with another vessel. Parks, 2 Marine Insurance and Average at 832-33. Since then, coverage has been extended to liability incurred as a result of a variety of different risks, including liability resulting from ship building and ship repair. Id. at 839. As Parks notes in his treatise on marine insurance:
“There are, of course, many different types of protection and indemnity policies other than the classic form originally intended to cover collision liabilities which were not then provided by the hull policies. There are freight, demurrage, and defense associations, and associations covering shipbuilders’ liabilities, ship repairers’ liabilities, and the like.”
Id. (footnote omitted). Contrary to London’s argument, P & I policies are not limited to liability arising from the operation of a vessel; rather, they can provide coverage for such things as liability arising from building and repairing ships.
14. In this case, the P & I policy that London issued Zidell covers liability arising from scrapping ships. That policy provides: “Vessels for scrapping insured subject to — Hull F.P.A. Cover conditions as attached.” (Emphasis in original.) The attached Hull F.P.A. cover conditions incorporate an attached standard form for Port Risks, which includes the provision at issue on review. 23 As Zidell correctly argues, the P & I policy that London issued in this case was not limited to *145 liability that Zidell incurred in operating vessels; it expressly covered liability that Zidell incurred in scrapping ships.
Although London argues that two cases support its position, they go to the first issue that London raises, not the second. In the first case, a crane operator on a fixed offshore platform negligently raised a piece of machinery from a vessel, injuring one of the vessel’s crew members. Lanasse v. Travelers Insurance Company, 450 F2d 580, 582 (5th Cir 1971). The crane operator was negligent; the vessel was not. Id. at 583. Although the same entity (Chevron) owned both the platform and the vessel, it was liable for the crew member’s injuries as the owner of the platform, not as the owner of the vessel. Id. at 584. The P & I insurance in that case, however, provided that the insurer would pay only those sums that the assured, as owner of the vessel, was liable to pay. Id. at 583 n 7 & 584. Because Chevron did not incur any liability as the owner of the vessel, the insurer had no obligation to indemnify Chevron under the P & I policy in that case.
The second case is a variation on the first. In that case, a vessel towing a barge negligently caused the barge to hit a bridge and sink. St. Paul Fire & Marine Ins., 666 F2d at 935. The federal government removed the sunken barge and was entitled to recover the cost of doing so from the entity that negligently caused the barge to sink. Id. at 940. Only the towing vessel’s owner was liable to the government for the cost of removing the barge. Id. The barge’s owner incurred no liability to the government. Id. The court assumed that the same entity (Vest Transportation) owned the towing vessel and the barge. Id. at 941. When Vest sought to recover under a P & I policy, the court held that the insurer had agreed to indemnify only the owner of the barge for the liability it had incurred for the cost of the barge’s removal. Id. Vest, however, had incurred no liability as the owner of the barge and accordingly could not recover under the P & I policy. Id. at 941, 945.
Neither of the cases on which London relies stands for the proposition that an insurer is obligated to indemnify the assured under P & I coverage only if the assured incurred liability in the course of operating an insured vessel. Each *146 case does stand for the proposition that the insurer’s obligation in those cases was limited to indemnifying the assured only to the extent that the assured incurred liability as the owner of the insured vessel. In this case, the P & I policy provides that London has a duty to indemnify Zidell to the extent that Zidell is liable by reason of its interest in the insured vessels. As explained above, the trial court’s findings do not impose any greater obligation on London than the terms of the policy permit. We find no error in the trial court’s ruling.
D. Proceedings on Remand
Having found that the trial court erred in allocating the burden of proof to London on the express fortuity policies, the Court of Appeals reversed the trial court’s judgment, vacated the supplemental judgment awarding Zidell attorney fees, and remanded the case for a new trial. ZRZ Realty, 222 Or App at 457-58. On review, Zidell argues that, even if we uphold the Court of Appeals’ allocation of the burden of proof, the Court of Appeals erred in remanding for a new trial on both the express and implied fortuity polices. 24 Zidell contends that a new trial is necessary only on the express fortuity policies. Zidell also argues that the Court of Appeals should have affirmed the “attorney-fee portion of the judgment as well as the supplemental judgment for attorney fees.” London responds that, under Maxwell v. Port. Terminal RR. Co., 253 Or 573, 456 P2d 484 (1969), the Court of Appeals was correct in “requiring a retrial as to all policies, and in reversing the fee awards.”
We do not read Maxwell as broadly as London does. The court explained in Brown v. Bonesteele, 218 Or 312, 335, 344 P2d 928 (1959), that “[u]nder appropriate circumstances this court may remand a case for a new trial on a part only of *147 the issues raised in the original proceeding.” The court consistently has followed that principle. See Estate of Michelle Schwarz v. Philip Morris Inc., 348 Or 442, 460, 235 P3d 668 (2010) (remanding for a retrial only on punitive damages); Western Feed Co. v. Heidloff, 230 Or 324, 349, 370 P2d 612 (1962) (remanding for a retrial only on the defendant’s counterclaim); Dunn v. Henderson, 122 Or 331, 336, 258 P 183 (1927) (upholding determination of liability and remanding for a trial on damages only).
In Maxwell, this court recognized an exception to that general rule. The court explained that, “[i]n the ordinary two-party personal injury case, * * * evidence of fault can influence the jury’s measurement of damages; and the kind and degree of injuries may influence some jurors in their evaluation of the evidence on liability.” Maxwell, 253 Or at 577. Accordingly, the court held that a “new trial in a personal-injury case ordinarily should be a new trial on all contested factual issues.” Id.
This case differs from Maxwell in two respects. Not only was the court careful to limit its holding in Maxwell to cases tried to a jury, id. at 575, but the court’s reasoning in Maxwell turned on the relationship between liability and damages in personal injury cases, id. at 577. This case, by contrast, was tried to the court, and nothing that London has identified persuades us that the trial court cannot, on remand, fairly limit any retrial to the express fortuity policies; that is, we see no reason why the trial court needs to retry London’s responsibility to indemnify Zidell under the implied fortuity policies to determine whether London is also required to indemnify Zidell under the express fortuity policies. The remand should have been limited, at least initially, to the question whether, for the purposes of the express fortuity policies, Zidell either expected or intended that a series of contaminants resulting from the operation of its business would damage the environment. 25
*148 In retrying that issue, the trial court must determine initially whether it is necessary to supplement the record; that is, if neither party can establish a specific basis for saying that the record would have been different if the trial court had placed the burden of production and persuasion initially on Zidell, then the trial court may find, based on the existing record, what Zidell expected or intended for the purposes of the relevant express fortuity policies. 26 If, on either the existing or a supplemented record, the trial court makes the same findings on remand that the trial court did initially, then it presumably can reenter the judgment, with any appropriate adjustment for attorney fees. Conversely, if the trial court reaches a different conclusion on remand as to when Zidell expected or intended property damage for the purposes of the express fortuity policies, then the court also will presumably have to adjust the findings allocating responsibility for reme-diating the damage between London and Zidell.
Zidell also argues that the Court of Appeals erred in reversing the fee award in the judgment and in vacating the fee award in the supplemental judgment. As we understand the trial court’s rulings on attorney fees, it found that London had a duty to defend Zidell in response to DEQ’s 1994 notice. 27 The trial court also found that, when Zidell transmitted the 1994 DEQ letter to London, that letter served as a proof of loss for the purposes of ORS 742.061. Finally, the *149 trial court found that, because London did not offer to pay Zidell’s defense costs within the period set out in ORS 742.061, that statute authorized Zidell to recover the attorney fees that it incurred in bringing this action to establish and enforce London’s duty to defend, but not its duty to indemnify. 28
In the Court of Appeals, London assigned error to four of the trial court’s rulings regarding attorney fees. The Court of Appeals considered and rejected London’s first assignment of error — that the 1994 letter was not a sufficient proof of loss. ZRZ Realty, 222 Or App at 493-95. The Court of Appeals, however, did not consider the other three assignments of error that London raised regarding the attorney fees or the two assignments of error concerning attorney fees that Zidell raised on cross-appeal. Id. at 495, 497. The Court of Appeals reasoned that it was not necessary to reach those issues. It explained that, “[b]ecause we reverse and remand for trial on the breach of contract and declaratory judgment claims, we necessarily reverse the award of attorney fees in favor of Zidell. ORS 20.220(3).” Id. 29
On review, Zidell argues that the Court of Appeals erred in reversing the attorney fee award in the judgment and in vacating the attorney fee award in the supplemental judgment. It contends that the Court of Appeals’ ruling reversing the trial court’s allocation of the burden of proof on the express fortuity policies had no effect on the trial court’s award of attorney fees. We agree with Zidell. According to the trial court, it awarded Zidell only those fees and costs that were attributable to London’s breach of its duty to defend. 30 *150 By contrast, the issue that the trial court must retry on remand — whether Zidell expected or intended the damage resulting from its business activity — is relevant only to the question whether London has a duty to indemnify Zidell under the various express fortuity policies that London issued.
This court has long recognized that “[t]he duty to indemnify is independent of the duty to defend.” Ledford, 319 Or at 403. An insurer may have a duty to defend its insured, as the trial court found here, but no duty to indemnify. It follows that, even if the trial court finds on remand that London had no duty to indemnify Zidell under the express fortuity policies, that ruling would have no effect on either the trial court’s ruling that London had a duty to defend Zidell in DEQ’s enforcement action or the trial court’s conclusion that ORS 742.061 authorized Zidell to recover attorney fees it incurred in this action in establishing and enforcing London’s duty to defend. The Court of Appeals erred in reversing and vacating the fee awards. 31
That said, we note that London’s eighth assignment of error, which the Court of Appeals did not reach, contended that trial court erroneously had “awarded Zidell fees its attorneys incurred on issues other than LPGL Underwriters’ duty to defend.” 32 London also assigned error to the trial court’s rulings on the appropriate hourly rate and a discrete category of activity for which the trial court allowed fees. Zidell, for its part, raised two assignments of error on cross-appeal regarding attorney fees. Because we conclude that the Court of Appeals erred in reversing and vacating the fee awards, it follows that we must remand this case initially to the Court of Appeals to permit it to consider those assignments of error. On remand before the Court of Appeals, the *151 parties are also free to ask the court to consider, if appropriate, those assignments of error that the court did not reach because it remanded the case for a new trial on both the express and implied fortuity policies. Because at least some of those issues must go back to the Court of Appeals for its resolution before the case returns to the trial court, we remand this case initially to the Court of Appeals.
The decision of the Court of Appeals is affirmed in part and reversed in part. The case is remanded to the Court of Appeals.
Footnotes
[1] Some plaintiffs are either the successors in interest to or assignees of those companies. That status is not material to the issues on review.
[2] The trial court found that, in addition to the pre-1966 comprehensive general liability policies, London issued protection and indemnity policies from 1957 through 1983 and bumbershoot policies from 1970 through 1977, none of which required that property damage be unexpected and unintended. The court read a fortuity requirement into those policies, as well as into the pre-1966 comprehensive general liability policies.
[3] London argued in the Court of Appeals that the same limitation — expected or intended damages — should apply to policies issued after January 1, 1968. The Court of Appeals initially rejected that argument, upholding the trial court’s conclusion that only intended damages were excluded. ZRZ Realty, 222 Or App at 481. On reconsideration, it withdrew its ruling on that issue and left the issue for the trial court on remand. ZRZ Realty, 225 Or App at 265. Zidell has not challenged that ruling on review.
[4] The wording in the express fortuity policies is not identical. However, the parties have not argued that any difference in wording in those policies affects how the burden of proof should be allocated.
[5] Zidell’s third amended complaint included other claims for relief. The trial court, however, dismissed those claims with prejudice, and Zidell has not challenged that ruling. Accordingly, we discuss only Zidell’s breach of contract and declaratory judgment claims.
[6] As we understand the trial court’s ruling, it used the phrase “burden of proof’ as a shorthand reference to the burden of production and the burden of persuasion.
[7] The trial court explained that, “[t]o establish that [Zidell] intended third party-property damage [i.e., damage to the river or the groundwater], the trier of fact will have to find that [Zidell] polluted for the purpose of causing damage to the Willamette River or groundwater.” (Emphasis in original.) It also explained that “[a] person expects a result if they act with awareness that the result is substantially certain to follow.” (Emphasis in original.) Neither party has challenged those definitions in this court, and we express no opinion on them.
[8] For example, the trial court found that the vast majority of the arsenic in the sediment came from anti-fouling paint used before 1950. It allocated responsibility for remediating 90 percent of the arsenic based on the number of pre-1950 ships dismantled before 1972 (allocated to London) and those dismantled after that date (allocated to Zidell). The court also articulated a principle to allocate the cost of remediating the environmental damages for the arsenic when the amount of damage for which one party was responsible marginally increased the remediation costs. The court employed a similar type of analysis for other contaminants.
[9] The court found that Zidell did not expect that its activities would damage the groundwater.
[10] At the time of trial, Zidell had not incurred any recoverable indemnity costs. The court’s findings accordingly established formulas for allocating responsibility between London and Zidell for those future costs.
[11] One paragraph, which was captioned “exemptions,” provided that “[t]he services provided for herein shall not apply to any * * * condition * * * which definitely existed at the time the afflicted employee came under the protection of this contract * * 194 Or 530 (emphasis and ellipses in original). It also provided a limited exemption for “chronic conditions.” It provided that employees “will not be entitled to services for chronic conditions * * * until they have been under the coverage of [this policy] for six consecutive months * * Id. A complementary paragraph, captioned “new employees,” provided that new employees “shall not be entitled to services for chronic conditions * * * which may acutely manifest themselves during the first six months of employment * * *.” Id. (emphasis in original).
[12] Technically, the reference in Smith to the burden of proof was dictum. As noted, the issue in Smith arose on a motion for a directed verdict. Because there was sufficient evidence from which the jury could have found either for the plaintiff or the defendant, there was no issue regarding who bore the burden of production. Moreover, because the only issue on review of a motion for directed verdict was whether the evidence required the jury to find that the plaintiffs condition came within the exemptions, the burden of persuasion was not implicated either. In that regard, there was no claim that the only conclusion a reasonable juror could reach was that the evidence was in equipoise. Cf. State v. James, 339 Or 476, 483, 123 P3d 251 (2005) (if, in resolving a motion subject to proof by a preponderance of the evidence, a court finds the evidence in equipoise, the party bearing the burden of persuasion loses).
[13] We note that the jurisdictions that have considered this issue have split on it. Compare Travelers Cas. v. Ribi Immunochem Research, 108 P3d 469, 474-75 (Mont 2005) (viewing requirement that damages be neither expected nor intended as limitation on coverage that the insured had to prove); Consolidated Edison v. Allstate, 774 NE2d 687, 690-92 (NY 2002) (same); Queen City Farms v. Central Nat. Ins. Co., 882 P2d 703, 715-16 (Wash 1994) (same), with State v. CNA Ins. Companies, 779 A2d 662, 671-72 (Vt 2001) (limitation was an exclusion that insurer had to prove); Carter-Wallace v. Admiral Ins., 712 A2d 1116, 1125-26 (NJ *132 1998) (same); Cameron Mut. Ins. Co. v. Moll, 50 SW3d 329, 332 (Mo App 2001) (same).
[14] Consistent with its practice in other areas, the court in A-l Sandblasting sought to ground the public policy limitation either in legislative enactments or, failing that, in case law. See 293 Or at 22-26 (exploring those sources and noting the difficulty in courts independently determining public policy).
[15] As we understand the trial court’s ruling, the scope of the public policy exception derives, from 1956 to 1967, from a statute that the legislature repealed in 1967 and, from 1968 on, from the cases discussed in A-l Sandblasting. That is why, beginning in 1968, the trial court read a less restrictive limitation into the implied fortuity policies.
[16] That is how, as we understand the trial court’s rulings, it regarded the former statute. In that respect, we note that neither party has challenged the scope of the public policy limitation that the trial court drew from former ORS 736.005(1) (1965), and we express no opinion on it.
[17] London also argues that the trial court ruled that the former statute was a limitation on coverage and that, because Zidell has not challenged that ruling on appeal, that ruling binds Zidell. The difficulty with London’s argument is that it focuses on one of the trial court’s rulings without considering the court’s other rulings regarding the scope and nature of the implied fortuity principle that the trial court read into the pre-1968 insurance policies.
[18] Since this court’s decision in Malady, other courts have considered additional factors in allocating the burden of production and persuasion. See 12 Moore’s Federal Practice § 57.62[2][c] (describing a more flexible approach); Fireman’s Fund Ins. Co. v. Videfreeze Corp., 540 F2d 1171, 1174-76 (3d Cir 1976) (illustrating analysis). Given our resolution of this issue, we need not decide whether those factors would apply and, if they did, whether they would affect the analysis.
[19] The 1959 policy provided coverage for each ship that Zidell purchased during the period in which the policy was in force and for which Zidell provided notice to London. In discussing the P & I policies, the parties discuss only the 1959 policy and treat it as representative of the P & I policies that followed. We follow the parties’ lead, although we note that the later policies are not identical.
[20] We say “ordinarily because the relationship between the specific and general terms can vary depending on the particular wording that the legislature uses. See Schmidt v. Mt. Angel Abbey, 347 Or 389, 402-07, 223 P3d 399 (2009) (discussing ejusdem generis rule); id. at 408-10 (Walters, J., concurring) (same).
[21] The word “harbour” refers to both naturally occurring and artificially created harbors. See Webster’s Third New Int’l Dictionary 1031 (unabridged ed 2002) (defining harbor as “a small bay or other sheltered part of a considerable body of water usu. well protected either naturally or artificially (as by jetties) from high waves * * * and deep enough to furnish anchorage for ships”); Robert H. Brown, A Dictionary of Marine Insurance Terms 171 (4th ed 1975) (defining harbor as “[a] place either natural or artificial where ships may anchor or lay and be protected from the open sea”). It is permissible to read that word, as Zidell does, to refer to natural features as well as artificial ones.
[22] The P & I policy includes a “special clause” that applies to vessels intended for scrapping. It provides:
“In respect of vessel (S) insured hereunder, it is agreed that this policy also covers the Assured and Affiliated Companies of the Assured be they Owners, subsidiaries or inter-related Companies and/or as bareboat Charterers and/or Charterers and/or Subcharterers and/or Operators and/or in whatever capacity * *
The trial court did not have occasion to consider how this clause affected, if at all, London’s obligation to indemnify the assured for liability occasioned “by reason of interest in the Vessel.”
[23] The attached Hull F.P.A Cover Conditions is subtitled “Applicable vessels intended for scrapping.”
[24] The Court of Appeals decision does not explicitly require a retrial on both the express and implied fortuity policies. However, that appears to have been its intent. For example, the Court of Appeals declined to address London’s third assignment of error, which was directed at both the implied and express fortuity policies, on the ground that it was reversing and remanding for a new trial on those policies. ZRZ Realty, 222 Or App at 480. Moreover, Zidell petitioned for reconsideration on the ground that the Court of Appeals should have remanded for a retrial only on the express fortuity policies, and the Court of Appeals denied Zidell’s petition. ZRZ Realty, 225 Or App at 262.
[25] Zidell concedes that, if the trial court erred in placing the burden of proof on London regarding the express fortuity policies, then the case must he remanded for a limited retrial. Zidell’s concession is well taken. For example, the trial court found by a preponderance of the evidence that, in 1972, Zidell expected that the arsenic in the anti-fouling paint would cause property damage. The trial court, however, made no finding as to what Zidell expected or did not expect for the *148 preceding years. For all that the record reveals, the trial court could have found that, for 1971 and for each of the preceding years in which the trial court incorrectly allocated the burden of production and persuasion, London simply failed to persuade the trial court that Zidell had expected property damage from the arsenic without also finding that Zidell in fact had not expected property damage. Given this record, we agree that the trial court’s misallocation of the burden of production and persuasion requires at least a limited retrial regarding the express fortuity policies.
[26] At trial, the parties discussed what would happen on remand if the appellate courts reversed the trial court’s allocation of the burden of production and persuasion. Initially, the trial court explained that, in its view, it could simply retry the issue on the existing record. When London’s counsel suggested that Zidell might have tried its case differently if the trial court had placed the burden of production and persuasion on Zidell, the trial court tentatively agreed. Zidell, however, did not identify how it would have tried its case differently, and we think that it is incumbent on the party that wants to supplement the record to identify a specific, legitimate reason for doing so.
[27] London argues, and Zidell does not dispute, that only some of the policies imposed a duty to defend on London and that only some of the London defendants, the LPGL Underwriters, bore that duty. For ease of reference, we refer generally to London’s duty to defend.
[28] On review, neither party challenges the trial court’s findings regarding attorney fees, and we express no opinion on them.
[29] ORS 20.220(3) provides:
“When an appeal is taken from a judgment under ORS 19.205 to which an award of attorney fees or costs and disbursements relates:
“(a) If the appellate court reverses the judgment, the award of attorney fees or costs and disbursements shall be deemed reversed; or
“(b) If the appellate court modifies the judgment such that the party who was awarded attorney fees or costs and disbursements is no longer entitled to the award, the party against whom attorney fees or costs and disbursements were awarded may move for relief under ORCP 71 B(l)(e).”
[30] We express no opinion on the question whether the fee award may have included fees that Zidell’s “attorneys incurred on issues other than LPGL Underwriters’ duty to defend,” as London argued before the Court of Appeals. As *150 explained below, the Court of Appeals did not reach that issue, and we leave that issue to the Court of Appeals on remand.
[31] Because the part of the trial court’s judgment that the Court of Appeals reversed does not relate to the awards of attorney fees, OES 20.220(3) did not require the Court of Appeals to reverse and vacate the fee awards.
[32] What the trial court included in its fee award is not always clear, in part because the parties appear to have negotiated and settled on a rolling basis some, if not many, of the specific charges that give rise to this issue.
