Opinion

In this cause, we resolve two issues relating to standard commercial general liability insurance policies, which were formerly called comprehensive general liability insurance policies. The first question is whether site investigation expenses—broadly, expenses for determining the existence, nature, extent, effect, etc., of the discharge of hazardous substances at a location—may constitute defense costs that the insurer must incur in fulfilling its duty to defend. The second is whether defense costs may be allocated to the insured. As we shall explain, we conclude that, as to each, the answer is qualifiedly affirmative.

I

This is still another chapter in the yet-to-be-completed volume relating the story of Aerojet-General Corporation in Sacramento County. (See, e.g., *46 Mangini v. Aerojet-General Corp. (1996) 12 Cal.4th 1087 [ 51 Cal.Rptr.2d 272 , 912 P.2d 1220 ]; Aerojet-General Corp. v. Transport Indemnity Insurance (1993) 18 Cal.App.4th 996 [ 22 Cal.Rptr.2d 862 ]; Mangini v. Aerojet-General Corp. (1991) 230 Cal.App.3d 1125 [ 281 Cal.Rptr. 827 ]; Aerojet-General Corp. v. Superior Court (1989) 211 Cal.App.3d 216 [ 257 Cal.Rptr. 621 ].) Aerojet-General Corporation is, and has been, a leading manufacturer in the aerospace and defense markets. Throughout the course of its operations from the early 1950’s into the 1980’s, it discharged hazardous substances, including trichloroethylene, in an ongoing fashion at its Sacramento site and thereby caused pollution in and around that location as such substances spread onto the ground, into the groundwater, and beyond toward the American River.

In 1982, Transport Indemnity Company and Associated International Insurance Company (hereafter collectively Transport Indemnity) filed a complaint for declaratory relief in the Superior Court of San Mateo County, which was docketed under No. 262425, against, inter alios, numerous other insurers and their common insureds, Aerojet-General Corporation and its wholly owned subsidiary Cordova Chemical Company (hereafter collectively Aerojet), regarding the parties’ rights and duties under various comprehensive general liability and other insurance policies. 1 It appears that what was stated above was already known or believed—that, throughout the course of its operations from the early 1950’s into the 1980’s, Aerojet had discharged hazardous substances in an ongoing fashion at its Sacramento site and had thereby caused pollution in and around that location resulting in continuous and/or progressively deteriorating bodily injury and/or property damage. In its complaint, Transport Indemnity sought declarations including that it was not obligated to provide, and Aerojet was not entitled to receive, either indemnification or defense.

Aerojet, which had been represented by independent counsel since about 1979, filed a cross-complaint for declaratory and other relief against, inter alios, Transport Indemnity and other of its insurers—which, for convenience’s sake, will generally be referred to without differentiation as “the insurers.”

Aerojet later filed an amended cross-complaint—the one operative here— against 54 insurers, under 245 comprehensive general liability and other *47 insurance policies with periods incepting as early as 1950 and. expiring as late as 1984, as to 3 actions brought by either the United States or the State of California and 35 actions brought by private parties, each of which was based on facts, alleged or otherwise disclosed, to the effect that, throughout the course of its operations from the early 1950’s into the 1980’s, Aerojet discharged hazardous substances in an ongoing fashion at its Sacramento site and thereby caused pollution in and around that location resulting in continuous and/or progressively deteriorating bodily injury and/or property damage. The private actions were various. The governmental ones were these: (1) an action instituted by the State of California against Aerojet in the Superior Court of Sacramento County in 1979 under authority of, inter alia, the Porter-Cologne Water Quality Control Act (Wat. Code, § 13000 et seq.), seeking relief including an injunction directing the company to undertake cleanup, abatement, and remedial work as to its pollution, and an order requiring the company to reimburse the state for the costs that the latter had, and would, incur in its own cleanup, abatement, and remedial work; (2) an action instituted by the United States against Aerojet in the United States District Court for the Eastern District of California in 1986 under authority of, inter alia, the Comprehensive Environmental Response, Compensation, and Liability Act of 1980 (hereafter CERCLA) ( 42 U.S.C. § 9601 et seq.), seeking relief including an injunction directing the company to abate and remedy its pollution and its effects, and an order requiring the company to reimburse the United States for so-called “response costs,” viz., the costs of removal and/or remediation (see 42 U.S.C. § 9601 (25)), 2 which the United States had, and would, incur with regard thereto; and (3) a parallel CERCLA action instituted by the State of California against Aerojet on the same day and in the same court, seeking relief including an order requiring the company to reimburse the state for the “response costs” that the latter had, and would, incur; the federal and state CERCLA actions were consolidated. In its amended cross-complaint, Aerojet sought, among other things, a declaration that it was entitled to receive, and the insurers were obligated to provide, both indemnification and defense. In pertinent part, it alleged to the *48 effect that it had tendered the defense, but that the insurers had either refused or had accepted only under “unreasonable” reservations of rights.

In an opinion certified for publication, the Court of Appeal, First Appellate District, Division Five, granted a petition for writ of mandate submitted by Aerojet to compel the superior court to vacate an order granting a motion by the insurers for summary adjudication of certain issues and to enter a new and different order denying that motion. (Aerojet-General Corp. v. Superior Court, supra, 211 Cal.App.3d at pp. 220-238.) In effect, the superior court had summarily adjudicated that “response costs” under CERCLA, and similar costs under the Porter-Cologne Water Quality Control Act, could not constitute indemnification costs, i.e., expenses to resolve liability, that the insurers had to incur in fulfilling their duty to indemnify. The Court of Appeal concluded to the contrary. It therefore caused issuance of a peremptory writ of mandate as prayed. It did not consider whether costs of this sort could constitute defense costs, i.e., expenses to avoid or at least minimize liability.

Following entry of the dismissal without prejudice of the complaint by Transport Indemnity, the superior court effectively transformed Aerojet’s cross-action against the insurers into an action in and of itself, ordering Aerojet to be designated “plaintiff’ and not “cross-complainant” and the insurers to be designated “defendants” and not “cross-defendants.”

The superior court subsequently ordered the action to be tried in phases. Phase I would be tried to the court, and would concern issues such as: (1) the authenticity of certain policies purportedly issued to Aerojet by some of the insurers; (2) the existence and wording of certain other policies missing in whole or in part; (3) the meaning of all such policies; and (4) related questions. Phase II would be tried to a jury, and would concern issues such as the application of the policies proved to the evidence presented in order to determine whether the insurers had a duty to indemnify Aerojet. Phase III would be tried to the same jury, and would concern issues such as: (1) the sum that Aerojet was entitled to receive from the insurers as a result of any duty to indemnify; (2) inasmuch as Aerojet’s tender of the defense had by now been accepted by the insurers under certain reservations of rights, whether, and apparently in what amount, Aerojet was entitled to receive any additional payment for defense costs, and whether, and apparently in what amount, the insurers were entitled to obtain reimbursement for any such payment already provided; and (3) whether, and in what amount, the insurers were liable to Aerojet for so-called “bad faith” damages. Phase IV, if necessary, would be tried to the court, and would concern issues such as the allocation among the insurers of any sum determined to be owing to Aerojet.

*49 At phase I, the superior court resolved issues concerning the existence, authenticity, wording, and meaning of various policies issued or purportedly issued to Aerojet by the insurers.

Specifically, in phase IA, the superior court made determinations as to existence, authenticity, and wording. From 1956 to 1984, it appears, Aerojet had various comprehensive general liability insurance policies and similar instruments that are pertinent here.

In phase IB, the superior court made determinations as to meaning. By way of background: Prior to 1966, in its insuring clause the standard comprehensive general liability insurance policy covered specified harm, such as bodily injury or property damage, caused by “ ‘accident.’ ” (Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ^ 7:26, p. 7A-8 italics omitted.) In 1966, the form was revised in its insuring clause to cover specified harm caused by an “ ‘occurrence,’ ” which was defined as an “ ‘accident, including injurious exposure to conditions, which results during the policy period in’ ” harm of this sort “ ‘neither expected nor intended from the standpoint of the insured.’” (Id., H 7:28, p. 7A-9, italics omitted.) In 1973, the form was further revised: In its insuring clause, it continued to cover specified harm caused by an “ ‘occurrence,’ ” but now defined that term as an “ ‘accident, including continuous or repeated exposure to conditions, which results in’ ” harm of this sort “ ‘neither expected nor intended from the standpoint of the insured.’ ” (Id., ^ 7:29, p. 7A-9, italics omitted.) To return to the superior court’s determinations: Aerojet’s comprehensive general liability insurance policies and similar instruments largely conformed to the standard ones in applicable' aspects. The phrase “neither expected nor intended” was express in the insuring clause of some, reflecting Insurance Code section 533, which states that “[a]n insurer is not liable for a loss caused by the wilful act of the insured . . . .” The phrase was implied in the insuring clause of the rest, through operation of the same provision. The phrase incorporated a subjective standard as to “intent” but, under City of Carter Lake v. Aetna Cas. and Sur. (8th Cir. 1979) 604 F.2d 1052, 1058-1059 , an objective standard as to “expectation.” If specified harm is “expected” or “intended” by the insured, it is effectively caused by a “wilful act” within the meaning of Insurance Code section 533, and hence outside of coverage. From 1976 to 1984, Aerojet’s policies, which were issued by the Insurance Company of North America (hereafter INA), essentially took a form similar to that of a so-called “fronting” policy, 3 to the following effect: Although, in the body, it was stated that INA had a duty to indemnify Aerojet, by endorsement it was provided that (1) *50 INA had a duty to make payments only beyond stated deductible amounts, which matched or approached indemnification limits, and (2) in case of Aerojet’s default, INA had a duty to make payments within the stated deductible amounts and a corresponding right to obtain reimbursement 4 and although, in the body, it was stated that INA had a duty to defend Aerojet, by endorsement it was provided that Aerojet should pay its own defense costs—under which provision it was understood by Aerojet that it should defend itself. 5

At phase II, Aerojet and the insurers presented evidence relevant to the issue of the duty to indemnify. In substance, it was established that, throughout the course of its operations from the early 1950’s into the 1980’s, Aerojet discharged hazardous substances in an ongoing fashion at its Sacramento site and thereby caused pollution in and around that location. The superior court charged the jury on the duty to indemnify. In the course of passing on various motions, it had previously determined that there was no such duty for periods before 1956 or after 1979. (specifically, after July 14, 1979). It submitted the question to the jury whether there was any such duty for any period from 1956 to 1979. In pertinent part, it instructed that “Aerojet has no insurance coverage if’ specified harm “is expected or intended from the standpoint of Aerojet.” It defined “expected” objectively to “denote [] that the actor knew or should have known that there was a substantial probability that certain consequences would result from his or her acts or omissions.” By contrast, it defined “intended” subjectively to “denote[] that the actor desires the consequences of his act or believes that the consequences are substantially certain to follow.” The jury proceeded to return a unanimous verdict determining that there was no duty to indemnify for any period from 1956 to 1979.

At phase III, the issues to be tried to the jury were limited to a single one pursuant to a stipulation between Aerojet and certain of the insurers other than INA: “What sums, if any, expended by Aerojet for [site] investigation are defense costs?” In the stipulation, Aerojet and the insurers agreed, in pertinent part, to the following effect: In shares determined among themselves, the insurers had paid, or would pay, as defense costs the expenses that Aerojet had incurred, or would incur, under the categories of “legal” and “legal support” through the final disposition of any appeal Aerojet would *51 take from any ensuing judgment in the action; Aerojet and the insurers would litigate whether site investigation expenses were defense costs; Aerojet and the insurers would not litigate whether, or in what amount, the insurers were entitled to obtain reimbursement for Aerojet’s “legal” or “legal support” expenses, or whether, or in what amount, the insurers were liable to Aerojet for bad faith damages; but, in the event of retrial following reversal of the judgment on appeal, the insurers could litigate the question of reimbursement and Aerojet could litigate the question of bad faith. In another stipulation, Aerojet and the insurers agreed as follows: Since 1979, Aerojet had incurred as defense costs “legal” expenses of $5,283,568, and “legal support” expenses of $5,634,149, for a total of $10,917,717; and, toward that amount, the insurers had paid $5,680,367, plus interest.

At the outset of phase III, the superior court ruled to the effect that, generally, site investigation expenses were not defense costs. It also ruled that defense costs could be allocated to the insured. Specifically, “Aerojet has held itself out to the world as insured by I.N.A.” under its “fronting” comprehensive general liability insurance policies. “Aerojet has elected not to buy insurance for defense costs from 1976 on.” “Under the court’s equitable powers, Aerojet is responsible for a co-equal allocation of the defense cost.”

Before the presentation of evidence at phase III, the superior court preinstructed the jury, in accordance with the ruling described above, as follows: “In this phase of the trial Aerojet seeks the costs it incurred to investigate the pollution at the Sacramento site as costs of defending the litigation filed against them by the governments and others. [^] All sums that Aerojet paid, one, because of government orders or requests to investigate, clean up or remediate, or, [1] two, because of Aerojet’s agreement or commitment to perform the Aerojet investigation, cleanup or remediation are indemnity expenses, and therefore not recoverable as defense costs. flO Investigation costs, such as investigating the extent of the contamination or the viability of cleanup options and monitoring the spread of the wastes from the site, which were incurred as part of Aerojet’s effort to clean up or remediate the site, are not considered defense .costs, ffl] If the costs involved here were necessary to or part of Aerojet’s effort to clean up and remediate the site, such costs are [not] defense costs, even if Aerojet’s lawyers used information developed during the investigation to assist them in negotiating with the governments to limit Aerojet’s obligations to clean up and to remediate. fi[] If any investigation costs did not relate to any of the purposes described above, but were incurred because they were specifically requested by a lawyer and were reasonable for the purpose of defending Aerojet in litigation, such investigation costs would be defense costs.”

*52 At phase III, Aerojet and the insurers presented evidence concerning Aerojet’s site investigation expenses. By then, Aerojet had incurred such expenses in the amount of about $26,655,787.01. In part, it had conducted its site investigation against the private actions, which were ultimately about 38 in number. By around 1986, it had resolved all, or at least almost all, of these proceedings. After prevailing in a test case comprising three such matters, it settled all, or almost all, of the outstanding claims for about $450,000. In other part, it had conducted its site investigation against the state’s action under the Porter-Cologne Water Quality Control Act, or more precisely, against various administrative orders or requests that had been made antecedent thereto. In yet other part, it had conducted its site investigation against the consolidated federal and state CERCLA actions. In 1989, it had become subject to a partial consent decree entered therein, which incorporated an agreement negotiated by the parties. 6 Although it denied any and all liability, it was required by the decree to “complete a Remedial Investigation/Feasibility Study” or “RI/FS,” which was subject to oversight and review by the federal and state governments, in order “to determine the nature and extent of public health and environmental problems, if any, presented by the release or threat of release of hazardous substances at or from [its Sacramento site] and to develop and evaluate remedial alternatives so it can subsequently be determined which, if any, is necessary to remedy public health or environmental problems identified.” 7 It expressly admitted that it had incurred its site investigation expenses, which it conceded included “costs ... to investigate the extent of the contamination or the viability of cleanup option [sic] or the monitoring of the spread of waste from the site,” “because of government orders or requests to investigate, cleanup or remediate” or “because of [its own] agreement or commitment to perform the investigation, cleanup, or remediation . . . .” But it also impliedly claimed that it had, or would have, incurred such expenses in order to avoid or at least minimize liability. In addition, it had assertedly incurred remediation expenses, more broadly defined, in the amount of about $35,240,495.44 and, more narrowly defined, in the amount of $30,454,495.44.

After the presentation of evidence at phase III, the superior court instructed the jury with language virtually identical to that of the preinstruction quoted above.

*53 Following deliberations at phase III, the jury returned a unanimous verdict determining that Aerojet’s site investigation expenses were not defense costs in any part.

Because the jury at phases II and III had determined that no sum was owing to Aerojet by the insurers, the superior court had no need to, and did not, proceed to phase IV, at which it would have resolved issues such as the allocation of any sum of this sort among the insurers.

The superior court then rendered what it denominated its “final judgment.” Previously, it had entered various judgments and orders related thereto. Subsequently, it would enter an order on the taxing of costs.

Aerojet filed a notice of appeal from the final judgment. The appeal was docketed in the Court of Appeal, First Appellate District, under No. A057812, and was assigned to Division Five thereof. Previously, Aerojet had filed a notice of appeal from the various judgments and orders related to the final judgment. This had been docketed in the First Appellate District under No. A053808, and had been assigned to Division Five. Subsequently, Aerojet would file a notice of appeal from the order on the taxing of costs. This would be docketed in the First Appellate District under No. A059976, and would be assigned to Division Five.

In an opinion certified for partial publication, the Court of Appeal, having effectively consolidated all three appeals for consideration and decision, affirmed the “final judgment,” the various judgments and orders related thereto, and the order on the taxing of costs, except as indicated below.

In part I of its discussion, which it did not certify for publication, the Court of Appeal reviewed phase I of the trial. Following Shell Oil Co. v. Winterthur Swiss Ins. Co. (1993) 12 Cal.App.4th 715, 743-748 [ 15 Cal.Rptr.2d 815 ], which it deemed to have been approved on this point in Montrose Chemical Corp. v. Superior Court (1993) 6 Cal.4th 287, 304-305 [ 24 Cal.Rptr.2d 467 , 861 P.2d 1153 ], it concluded, inter alia, that the superior court erred in its determination, based on City of Carter Lake v. Aetna Cas. and Sur., supra, 604 F.2d 1052 , that the phrase “neither expected nor intended,” which it held to be express in the insuring clause of some of Aerojet’s comprehensive general liability insurance policies and similar instruments and implied in the insuring clause of the rest, incorporated an objective standard as to “expectation,” rather than a subjective one.

In part II of its discussion, which it also did not certify for publication, the Court of Appeal reviewed phase II of the trial. In conformity with its holding *54 that the phrase “neither expected nor intended” incorporated a subjective standard as to “expectation,” it concluded, inter alia, that the superior court erred by instructing the jury to the effect that “Aerojet has no insurance coverage if’ it “expected” specified harm, and that it “expected” such harm if it “knew or should have known that there was a substantial probability that certain consequences would result from [its] acts or omissions.” (Italics added.) But applying the harmless error rule of section 13 of article VI of the California Constitution, as construed in Soule v. General Motors Corp. (1994) 8 Cal.4th 548, 573-581 [ 34 Cal.Rptr.2d 607 , 882 P.2d 298 ], it concluded that the superior court did not prejudice Aerojet. In substance, it determined: The issue whether Aerojet “expected” that it would cause pollution at and around its Sacramento site focused on whether it “knew,” from the very commencement of its operations, that “there was a substantial probability” that adverse consequences would result from its acts or omissions; that Aerojet did in fact possess knowledge of this sort was supported by “overwhelming” evidence.

In part III of its discussion, which alone it certified for publication, the Court of Appeal reviewed phase III of the trial.

Specifically, in part III, the Court of Appeal concluded, inter alia, that the superior court erred by instructing the jury to the effect that Aerojet’s site investigation expenses were not defense costs to the extent that they were linked to orders or requests by the United States or the State of California. It held in substance as follows: Generally, site investigation expenses by an insured are defense costs to the extent that they are reasonable and necessary to avoid or at least minimize liability; AIU Ins. Co. v. Superior Court (1990) 51 Cal.3d 807, 818-843 [ 274 Cal.Rptr. 820 , 799 P.2d 1253 ] is not to the contrary, standing as it does only for the proposition that certain site investigation expenses by a third party—such as those included within “response costs” by the United States under CERCLA—may be indemnification costs to the extent that they must be reimbursed by the insured and resolve liability for specified harm; nevertheless, under the approach of the United States District Court in Fireman’s Fund Ins. Companies v. Ex-Cell-O Corp. (E.D.Mich. 1992) 790 F.Supp. 1318, 1338 (hereafter sometimes Ex-Cell-O I), and Fireman’s Fund Ins. Companies v. Ex-Cell-O Corp. (E.D.Mich. 1992) 790 F.Supp. 1339, 1346 (hereafter sometimes Ex-Cell-O II), there is a presumption that an insured’s site investigation expenses in connection with a Remedial Investigation/Feasibility Study or RI/FS responding to an order or request by the United States for specific information under CERCLA are not defense costs, but that presumption may be rebutted by the insured by showing that the site investigation expenses in question were reasonable and necessary to avoid or at least minimize liability and *55 would have been incurred even in the absence of such order or request; the general rule applies to Aerojet’s site investigation expenses against (1) the private actions, (2) the state’s action under the Porter-Cologne Water Quality Control Act and the various administrative orders or requests antecedent thereto, which were determined not to have sought site investigation, and (3) as qualified below, against the federal and state CERCLA actions; the exception applies to Aerojet’s site investigation expenses in connection with the Remedial Investigation/Feasibility Study responding to any order or request by the federal or apparently the state government for specific information under CERCLA. Applying the harmless error rule, it concluded that the superior court did indeed prejudice Aerojet: The erroneous instruction, which was given at the commencement of this phase as well as at its conclusion, effectively prevented Aerojet from presenting available evidence that had a tendency in reason to prove that its site investigation expenses were reasonable and necessary to avoid or at least minimize liability; moreover, it provided the jury with the wrong law to apply to such evidence as Aerojet and the insurers had in fact presented.

In part III, the Court of Appeal also concluded that the superior court did not err in ruling that defense costs could be allocated to the insured. It held to this effect: because Aerojet agreed that it would pay its own defense costs under the “fronting” comprehensive general liability insurance policies issued by INA from 1976 to 1984, it “should now carry [its] fair share of the burden” pro rata based on the time of noninsurance within the time as a whole.

The Court of Appeal denied a petition for rehearing by Aerojet. On its own motion, it modified part II of its opinion, which was unpublished, simply to add a single sentence at the very end in order to “note the obvious—that [its] decision in [this part] is based on the specific facts of this case.” (Italics in original.)

On separate petitions by Aerojet and the insurers, we granted review. Pursuant to rule 29.2(b) of the California Rules of Court, we subsequently specified the issues to be argued as indicated in the introduction.

n

The issues to be resolved are whether, under standard comprehensive or commercial general liability insurance policies, site investigation expenses may constitute defense costs that the insurer must incur in fulfilling its duty to defend, and whether, under such policies, defense costs may be allocated *56 to the insured. We shall first speak generally about the policies in question and then turn to address each of the questions in 8

A

Standard comprehensive or commercial general liability insurance policies are contracts between an insurer and an insured: In each, the insurer makes promises, and the insured pays premiums, the one in consideration for the other, against the risk of loss. (E.g., Buss v. Superior Court (1997) 16 Cal.4th 35, 44-45 [ 65 Cal.Rptr.2d 366 , 939 P.2d 766 ].)

In pertinent part, standard comprehensive or commercial general liability insurance policies provide that the insurer has a duty to indemnify the insured for those sums that the insured becomes legally obligated to pay as damages for a covered claim. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 45.) By definition, this duty entails the payment of money (e.g., id. at p. 46), which is expressly limited in amount (see Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ¶ 7:354 , p. 7A-76), in order to resolve liability (e.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 46 ). It is not narrowly confined to money that the insured must give under law as compensation to third parties, but may also include money that the insured must itself expend in equity in order to provide relief of the same sort. (AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d at pp. 818-843.) It runs to claims that are actually covered, in light of the facts proved. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at pp. 45-46.) It arises only after liability is established and as a result thereof. (E.g., id. at p. 46; see Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 659, fn. 9.) It is triggered if specified harm is caused by an included occurrence, 9 so long as at least some such harm results within the policy period. (Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at pp. 669-673.) It extends to all *57 specified harm caused by an included occurrence, even if some such harm results beyond the policy period. (See id. at p. 686.) 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. 10 To illustrate by a hypothetical similar to the present case: 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.

Standard comprehensive or commercial general liability insurance policies also provide that the insurer has a duty to defend the insured in any *58 action brought against the insured seeking damages for a covered claim. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 45 .) By definition, the duty entails the rendering of a service, viz., the mounting and funding of a defense (e.g., id. at p. 46), which is not limited, expressly or otherwise (see Travelers Ins. Co. v. Lesher (1986) 187 Cal.App.3d 169, 191 [ 231 Cal.Rptr. 791 ], disapproved on other points, Buss v. Superior Court, supra, 16 Cal.4th at pp. 50, fn. 12, & 52, fn. 14; cf. Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ¶ 7:647 , p. 7B-32 [speaking generally and without specific reference to such policies]), in order to avoid or at least minimize liability (see Gray v. Zurich Insurance Co. (1966) 65 Cal.2d 263, 279 [ 54 Cal.Rptr. 104 , 419 P.2d 168 ]). As such, it requires the undertaking of reasonable and necessary efforts for that purpose (see ibid.), including investigation (see Pacific Indem. Co. v. Universal etc. Ins. Co. (1965) 232 Cal.App.2d 541, 543-544 [ 43 Cal.Rptr. 26 ]). It also requires the incurring of reasonable and necessary costs to that end (see Travelers Ins. Co. v. Lesher, supra, 187 Cal.App.3d at p. 191 ), including investigative expenses (see Pacific Indem. Co. v. Universal etc. Ins. Co., supra, 232 Cal.App.2d at pp. 543-544). It runs to claims that are merely potentially covered, in light of facts alleged or otherwise disclosed. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 46 .) It arises as soon as tender is made (e.g., ibid.), before liability is established and apart therefrom (e.g., Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 659, fn. 9). It is discharged when the action is concluded. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 46 .) It may be extinguished earlier, if it is shown that no claim can in fact be covered. (E.g., ibid.) If it is so extinguished, however, it is extinguished only prospectively and not retroactively: Before, the insurer had a duty to defend; after, it does not have a duty to defend further. (E.g., ibid.) It is triggered if specified harm may possibly have been caused by an included occurrence, so long as at least some such harm may possibly have resulted within the policy period. (Cf. Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at pp. 669-673 [holding to such effect as to the duty to indemnify].) It extends to all specified harm that may possibly have been caused by an included occurrence, even if some such harm may possibly have resulted beyond the policy period. (Cf. id. at p. 686 [holding to such effect as to the duty to indemnify].) In other words, if specified harm may possibly have been caused by an included occurrence and may possibly have resulted, at least in part, within the policy period, it perdures to all points of time at which some such harm may possibly have resulted thereafter. 11 To illustrate again by a hypothetical: Insurer has a duty to defend Insured as to a claim for damages for property damage caused by its discharge of hazardous substances brought by Neighbor. Insured may possibly have discharged such a substance. It thereby may possibly have caused property damage to *59 Neighbor’s land within the policy period of year one. It may possibly have caused further damage as the substance may possibly have spread under the surface in year two through year thirty. Insurer must defend Insured as to the claim in its entirety.

It is plain that the insurer’s duty to defend is broader than its duty to indemnify. (E.g., Buss v. Superior Court, supra, 16 Cal.4th at p. 46 .) But it is also plain that it is not unlimited. (E.g., ibid.) It extends beyond claims that are actually covered to those that are merely potentially so, but no further. (E.g., ibid.)

Thus, in an action wherein all the claims are at least potentially covered because they may possibly embrace some triggering harm of the specified sort within the policy period caused by an included occurrence, the insurer has a duty to defend. (Buss v. Superior Court, supra, 16 Cal.4th at pp. 46-47.) “This obligation is express in the policy’s language. It rests on the fact that the insurer has been paid premiums by the insured for a defense. ‘The rule is grounded in basic principles of contract law.’ [Citation.] The duty to defend is contractual. [Citations.] ‘An insurer contracts to pay the entire cost of defending . . . claimfs]’ that are at least potentially covered.” (Id. at p. 47.)

By contrast, in an action wherein none of the claims is even potentially covered because it does not even possibly embrace any triggering harm of the specified sort within the policy period caused by an included occurrence, the insurer does not have a duty to defend. (Buss v. Superior Court, supra, 16 Cal.4th at p. 47 .) “This freedom is implied in the policy’s language. It rests on the fact that the insurer has not been paid premiums by the insured for a defense. This ‘rule’ too ‘is grounded in basic principles of contract law.’ [Citation.] As stated, the duty to defend is contractual. ‘The insurer has not contracted to pay defense costs’ for claims that are not even potentially covered.” (Ibid.)

It follows that, in a “mixed” action, in which at least one of the claims is at least potentially covered and at least one of the claims is not, the insurer does not have a duty to defend the action in its entirety arising out of contract: It “has a duty to defend as to the claim[] that [is] at least potentially covered, having been paid premiums by the insured therefor, but does not have a duty to defend as to [the claim] that [is] not, having not been paid therefor.” (Buss v. Superior Court, supra, 16 Cal.4th at pp. 47-48.)

Nevertheless, the insurer has a duty to defend the entire “mixed” action imposed by law in support of the policy: “To defend meaningfully, [it] must *60 defend immediately. [Citation.] To defend immediately, it must defend entirely.” (Buss v. Superior Court, supra, 16 Cal.4th at pp. 48-49.)

It is manifest that this analysis applies, as it were, not only between claims but also between parts of a single claim. 12

Thus, when all the parts of a claim are at least potentially covered because each may possibly embrace some triggering harm of the specified sort within the policy period caused by an included occurrence, the insurer has a duty to defend. It has “ ‘contracted] to pay the entire cost of defending’ ” a claim of this sort. (Buss v. Superior Court, supra, 16 Cal.4th at p. 47 .)

By contrast, when none of the parts of a claim is even potentially covered because it does not even possibly embrace any triggering harm of the specified sort within the policy period caused by an included occurrence, the insurer does not have a duty to defend. It “ ‘has not contracted to pay defense costs’ ” for a claim of this sort. (Buss v. Superior Court, supra, 16 Cal.4th at p. 47 .)

It follows that, as to a “mixed” claim, in which at least one of the parts is at least potentially covered and at least one of the parts is not, the insurer does not have a contractual duty to defend the claim in its entirety.

Nevertheless, the insurer has a prophylactic duty to defend the entire “mixed” claim. That is because to defend meaningfully, it must defend immediately, and to defend immediately, it must defend entirely.

B

The first issue on review concerns whether, under standard comprehensive or commercial general liability insurance policies, site investigation expenses may constitute defense costs that the insurer must incur in fulfilling its duty to defend.

The insurer has a duty to defend. In fulfilling its duty, it must undertake reasonable and necessary efforts to avoid or at least minimize liability. To that end, it must incur reasonable and necessary costs. All this it must do from as early as tender of the defense through as late as conclusion of the action.

It follows that the insured’s site investigation expenses constitute defense costs that the insurer must incur in fulfilling its duty to defend if, *61 and only if, the following requirements are satisfied. First, the site investigation must be conducted within the temporal limits of the insurer’s duty to defend, i.e., between tender of the defense and conclusion of the action. Second, the site investigation must amount to a reasonable and necessary’ effort to avoid or at least minimize liability. Third and final, the site investigation expenses must be reasonable and necessary for that purpose.

Thus, if and to the extent that the insured’s site investigation is conducted within the temporal limits of the insurer’s duty to defend and amounts to a reasonable and necessary effort to avoid or at least minimize liability, the related site investigation expenses may possibly be defense costs that the insurer must incur in fulfilling its duty to defend. If and to the extent that these site investigation expenses are reasonable and necessary for that purpose, they are in fact defense costs that the insurer must incur in fulfilling its duty to defend; but if and to the extent that they are not, they are not. By contrast, if and to the extent that the site investigation is not conducted within the temporal limits of the insurer’s duty to defend or does not amount to a reasonable and necessary effort to avoid or at least minimize liability, the related site investigation expenses cannot even possibly be defense costs that the insurer must incur in fulfilling its duty to defend.

Not to the contrary is AW Ins. Co. v. Superior Court, supra, 51 Cal.3d 807 . As pertinent here, that decision stands only for the proposition that certain site investigation expenses by a third party—such as those included within “response costs” by the United States under CERCLA—may be indemnification costs to the extent that they must be reimbursed by the insured and resolve liability for specified harm. (Id. at pp. 824-843.) It simply does not hold or state that such expenses by an insured are not defense costs to the extent that they are reasonable and necessary to avoid or at least minimize liability. 13 In AIU Ins. Co. v. Superior Court, we construed standard policy language covering indemnification costs, i.e., “sums which [the. insured] becomes ‘legally obligated’ to pay as ‘damages’ . . . because of ‘[bodily injury or] property damage.’ ” (Id. *62 at p. 824.) We concluded that the phrase “legally obligated” means required of the insured, whether at law or in equity. (Id. at pp. 824-825.) We further concluded that the term “damages” comprehends the insured’s costs of providing the relief required, whether such costs are paid by the insured itself or reimbursed by it to a third party. (Id. at pp. 825-842.) We then concluded that the phrase “property damage” refers to specified harm that the insured has caused. (Id. at pp. 842-843.) It is manifest that site investigation expenses by the insured may not be indemnification costs, i.e., “sums which [the insured] becomes ‘legally obligated’ to pay as ‘damages’ . . . because of ‘[bodily injury or] property damaged ” For example, such expenses may be required of the insured under CERCLA even before it has been proved to have caused specified harm (see, e.g., 42 U.S.C. § 9604 (a)(1) [reaching “potentially responsible part[ies]” (italics added)])—and even if it is subsequently proved not to have done so (see American Bumper v. Hartford Ins. (1996) 452 Mich. 440, 443-447, 460-463 [ 550 N.W.2d 475, 477-479, 485-486 ]).

Whether the insured’s site investigation expenses are defense costs that the insurer must incur in fulfilling its duty to defend must be determined objectively, and not subjectively from the viewpoint of either the insurer or the insured.

Specifically, whether the site investigation is conducted by the insured within the temporal limits of the insurer’s duty to defend must be assessed under an objective standard. What matters is whether the site investigation actually occurs between tender of the defense and conclusion of the action, not whether it is honestly believed to occur.

Whether the insured’s site investigation amounts to a reasonable and necessary effort to avoid or at least minimize liability must also be assessed under an objective standard. What matters here is whether the site investigation would be conducted against liability by a reasonable insured under the same circumstances. Were it not, the question would require a discernment of motive. Why is the insured conducting the site investigation at issue? to resist liability? for that reason and some other? for a reason altogether different? “Motive, however, is ‘hard ... to discern.’ ” (Buss v. Superior Court, supra, 16 Cal.4th at p. 52, fn. 14 , quoting Della Penna v. Toyota Motor Sales, U.S.A., Inc. (1995) 11 Cal.4th 376, 405 [ 45 Cal.Rptr.2d 436 , 902 P.2d 740 ] (conc. opn. of Mosk, J.).) “That is true ... [as to] an individual: a person’s mind and heart typically reveal themselves and conceal themselves at one and the same time. It is truer still... [as to] a group of individuals: many minds and hearts are then involved, and they cannot simply be added up. And, of course, it is truest... [as to] a corporation or *63 similar entity”—like the typical commercial or governmental insured: “the ‘mind’ and ‘heart’ of such a one is purely fictive.” (Della Penna v. Toyota Motor Sales, U.S.A., Inc., supra, 11 Cal.4th at p. 405 (conc. opn. of Mosk, J.); accord, Buss v. Superior Court, supra, 16 Cal.4th at p. 52, fn. 14 .)

Lastly, whether the insured’s site investigation expenses are reasonable and necessary to avoid or at least minimize liability must be assessed under an objective standard as well. What matters here is whether the site investigation expenses would be incurred against liability by a reasonable insured under the same circumstances. Were it not, this question too would require a discernment of motive. Why is the insured incurring the site investigation expenses at issue? to resist liability? for that reason and some other? for a reason altogether different? “Motive,” again, “is ‘hard ... to discern.’ ” (Buss v. Superior Court, supra, 16 Cal.4th at p. 52, fn. 14 .)

All this is true even in the general context of a governmental request or order for the insured to conduct a site investigation and/or to incur site investigation expenses. Otherwise, the questions whether the insured’s site investigation is reasonable and necessary to avoid or at least minimize liability and whether its site investigation expenses are reasonable and necessary for that purpose would require a difficult discernment of motive. Why is the insured conducting the site investigation in question and why is it incurring the site investigation expenses at issue? to resist liability? to satisfy the government without regard to consequences? for an altogether different reason? or, most plausibly, both to resist liability and to satisfy the government?

All this is also true even in the specific context of an order or request for a Remedial Investigation/Feasibility Study or RI/FS under CERCLA by the United States. The federal government may require or ask for such a study. (See 42 U.S.C. § 9604 (a)(1).) It may conduct the study itself. (See ibid.) If it does, it may compel the insured to reimburse it for the cost thereof. ( 42 U.S.C. § 9607 (a).) The insured may be allowed to conduct the study if it is determined to be qualified to do so, if it is subjected to federal government oversight and review, and if it agrees to make reimbursement for the cost of such oversight and review. ( 42 U.S.C. § 9604 (a)(1).) It is well known that, by conducting the study itself, the insured may be able to avoid or at least minimize liability—both for the costs of the study and for any costs subsequent thereto (see, e.g., Aetna Cas. and Sur. Co., Inc. v. Pintlar Corp. (9th Cir. 1991) 948 F.2d 1507, 1517 ; see also AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d at p. 837 [stating that, “[a]s courts and commentators have recognized, government cleanup efforts [under CERCLA] are generally considerably more expensive than cleanups performed by” the insured]). *64 Here too, although the insured’s motives may be hard to discern, they are most plausibly both to resist liability and to satisfy the government.

Finally, on the question whether the insured’s site investigation expenses are defense costs that the insurer must incur in fulfilling its duty to defend, there arises the issue of the burden of proof.

In the general case, it is the insured that must carry the burden of proof on the existence, amount, and reasonableness and necessity of the site investigation expenses as defense costs, and it must do so by the preponderance of the evidence. “Evidence Code section 500 provides that, generally, a party desiring relief must carry the burden of proof thereon.” (Buss v. Superior Court, supra, 16 Cal.4th at p. 53 .) Further, “Evidence Code section 115 . . . provides that the burden of proof that is generally applicable is proof by a preponderance of the evidence. Of course, this burden is the ‘ordinary’ one for civil actions. [Citations.] It is applicable to contractual causes of action.” (Buss v. Superior Court, supra, 16 Cal.4th at pp. 53-54.)

By contrast, in the exceptional case, wherein the insurer has breached its duty to defend, it is the insured that must carry the burden of proof on the existence and amount of the site investigation expenses, which are then presumed to be reasonable and necessary as defense costs, and it is the insurer that must carry the burden of proof that they are in fact unreasonable or unnecessary. (Accord, Fireman’s Fund Ins. Companies v. Ex-Cell-O Corp. (Ex-Cell-O II), supra, 790 F.Supp. at p. 1346 [semble, but apparently intermingling the presumption affecting the burden of proof and the presumption affecting the burden of producing evidence]; cf. Isaacson v. California Ins. Guarantee Assn. (1988) 44 Cal.3d 775, 791 [ 244 Cal.Rptr. 655 , 750 P.2d 297 ] [concluding that, “if an insurer wrongfully fails to provide ... a defense, and the insured then settles the claim, . . . [i]n a later action against the insurer for reimbursement” “the insured is given the benefit of an evidentiary presumption,” later expressly held to affect the burden of proof (Xebec Development Partners, Ltd. v. National Union Fire-Ins. Co. (1993) 12 Cal.App.4th 501, 549 [ 15 Cal.Rptr.2d 726 ]) that the insured was indeed liable on, and in the amount of, the settled claim, so long as the settlement was reasonable].) As is ordinary in civil actions generally and for contractual claims in particular, the insurer and the insured must each carry its burden of proof by a preponderance of the evidence.

It follows that Aerojet’s site investigation expenses may constitute defense costs that the insurers must incur in fulfilling their duty to defend. From 1956 to about 1975, Aerojet had largely typical comprehensive general liability insurance policies and similar instruments, covering specified harm *65 including bodily injury and/or property damage, that were issued by various insurers. From 1976 to 1984, it had “fronting” policies of this sort that were issued by INA. The record on appeal shows that Aerojet’s site investigation may have been conducted within the temporal limits of each insurer’s duty to defend. It also shows that the site investigation, at least in part, may have amounted to a reasonable and necessary effort to avoid or at least minimize liability. It finally shows that the site investigation expenses, at least in part, may have been reasonable and necessary for that purpose. Whether and to what extent they actually were such are issues to be resolved on retrial.

The insurers argue against the foregoing conclusion. They fail to persuade.

For example, the insurers imply that site investigation and site investigation expenses are peculiar to hazardous substance discharge claims. That may be true. But it does not mean, as they apparently suppose, that site investigation cannot amount to a reasonable and necessary effort to avoid or at least minimize liability or that site investigation expenses cannot be reasonable and necessary for that purpose. Factually, a hazardous substance discharge claim may be sui generis. Legally, however, it is not unique. The insurers maintain that the governmental actions were different from others: They were not litigated, they did not seek a judgment; rather, they were negotiated, they aimed at settlement. The governmental actions may have been unlike others in degree. But not in kind. Many actions involve negotiation rather than litigation; many look toward settlement rather than judgment. What matters is the legal “essence,” as it were, of any site investigation and any site investigation expenses, and not their factual “accidents.”

In addition, the insurers assert that the insured will have carte blanche to pronounce any “site investigation expenses” to be defense costs, provided only that the insured entertains an honest belief—or persuasively says it entertains an honest belief—that they are. Not so. The standard is objective, not subjective.

Further, the insurers imply that site investigation expenses may be included within “response costs” under CERCLA. That may be true. But it does not mean, as they apparently suppose, that site investigation expenses cannot be defense costs. For instance, the insured’s site investigation expenses for identifying the hazardous substance discharged may be “response costs” insofar as they promote removal and remediation, by enabling it to determine how to neutralize the substance in question. They may also be defense costs insofar as they are reasonable and necessary to avoid or at least minimize liability, by making it possible for it to show that it was not in fact *66 the source of the discharge. Any assumption that site investigation expenses cannot do double duty is unsupported and hence must be rejected. So too any similar assumption about the site investigation itself. If site investigation expenses must be incurred by the insurer in fulfilling its duty to defend the insured, they must be incurred. The insurer gives, and the insured gets, what they bargained for. Even if the insured may happen to derive some added benefit, the insurer does not shoulder any added burden. The insurer may not be heard to complain. (Cf. Domtar, Inc. v. Niagara Fire Ins. Co. (Minn. 1997) 563 N.W.2d 724, 738-739 [applying Minnesota law, but speaking generally: That the same costs may do double duty as both indemnification costs and defense costs does not mean that they do not do duty as the latter as well as the former].) 14

Also, the insurers assume that site investigation expenses cannot be defense costs to the extent that the underlying site investigation is ordered or requested by the federal or state government or, it seems, by any other person or entity. Their premise is, apparently, that the effort undertaken by a party in defending itself must be voluntary, as for example initiated and controlled by the party itself. It is unsound. (See Hi-Mill Mfg. Co. v. Aetna Cas. & Sur. Co. (E.D.Mich. 1995) 884 F.Supp. 1109, 1116-1117 [applying Michigan law, but speaking generally]; American Bumper v. Hartford Ins, supra, 452 Mich, at pp. 460-463 [550 N.W.2d at pp. 485-486 [same]; General Acc. Ins. Co. v. State Dept. of Environ., supra, 143 N.J. at p. 476 [672 A.2d at pp. 1161-1162] [applying New Jersey law, but speaking generally].) What matters is what is done, not why. All the same: Typically, the effort undertaken by a defendant in responding to a plaintiff’s interrogatories is not voluntary in any real sense: It is initiated at the plaintiff’s request and is controlled by the plaintiff’s threat of a motion to compel further responses. But, typically, the expenses of responding to interrogatories may indeed be defense costs. The result is unaffected by the fact that a partial consent decree was entered in the consolidated federal and state CERCLA actions, under which Aerojet was required to “complete a Remedial Investigation/Feasibility Study” or “RI/FS,” which was subject to oversight and review by the federal and state governments. (See American Bumper v. Hartford Ins, supra, 452 Mich, at pp. 460-463 [550 N.W.2d at pp. 485-486] [applying Michigan law, but speaking generally].) “To be sure, consent decrees bear some of the earmarks of judgments entered after litigation. At the same time, because their terms are arrived at through *67 mutual agreement of the parties, consent decrees also closely resemble contracts.” (Firefighters v. Cleveland (1986) 478 U.S. 501, 519 [ 106 S.Ct. 3063, 3073 , 92 L.Ed.2d 405 ].) . Their “most fundamental characteristic,” however, is their “voluntary nature.” (Id. at pp. 521-522 [ 106 S.Ct. at p. 3075 ].) “Indeed, it is the parties’ agreement that serves as the source of the court’s authority to enter any judgment at all.” (Id. at p. 522 [ 106 S.Ct. at p. 3075 ].) “More importantly, it is the agreement of the parties, rather than the force of the law upon which the complaint was originally based, that creates the obligations embodied in a consent decree.” (Ibid.) 15

Contrariwise, the insurers assume that site investigation expenses cannot be defense costs to the extent that the underlying site investigation is “agreed” or “committed” to by the insured. Their premise is, apparently, that the effort undertaken by a party in defending itself must be involuntary. It is unsound. As stated, what matters is what is done, not why.

Next, the insurers assert that, if site investigation expenses may constitute defenses costs, so may settlement costs—a result they say is untenable. Again, not so. Site investigation expenses may be defense costs because they may be reasonable and necessary to avoid or at least minimize liability. Settlement costs cannot be defense costs because, instead, they resolve liability.

Finally, the insurers imply that the standard for determining whether the insured’s site investigation expenses constitute defense costs is not a “bright line” rule and hence will encourage needless litigation. To our mind, the test is clear. It is certainly as clear as the law allows. All must proceed as best they can.

By giving an affirmative answer to the question whether site investigation expenses may constitute defense costs that the insurer must incur in fulfilling its duty to defend, the Court of Appeal did not err. Surely, it was right to hold prejudicially erroneous the superior court’s instruction to the jury to the effect that Aerojet’s site investigation expenses did not constitute defense *68 costs to the extent that they were linked to orders or requests by the United States or the State of California. In its absence, there was a “reasonable chance” (College Hospital, Inc. v. Superior Court (1994) 8 Cal.4th 704, 715 [ 34 Cal.Rptr.2d 898 , 882 P.2d 894 ], italics omitted) that the jury would have found at least some of the site investigation expenses—which amounted to $26,655,787.01—to be such. The insurers’ claim to the contrary notwithstanding, evidence in support had been presented.

In answering the question of site investigation expenses as it did, the Court of Appeal erred to the extent that it chose to adopt the approach of the federal district court in Ex-Cell-O I (Fireman’s Fund Ins. Companies v. Ex-Cell-O Corp., supra, 790 F.Supp. at p. 1338) and Ex-Cell-O II (Fireman’s Fund Ins. Companies v. Ex-Cell-0 Corp., supra, 790 F.Supp. at p. 1346)— which is, in substance, that an insured’s site investigation expenses in connection with a Remedial Investigation/Feasibility Study or RI/FS responding to an order or request by the United States for specific information, under CERCLA are presumed, albeit only rebuttably, not to be defense costs. The federal district court did not present any reasoning, or cite any authority, in support. None is apparent. We shall not follow. 16

C

The second issue on review concerns whether, under standard comprehensive or commercial general liability insurance policies, defense costs may be allocated to the insured.

The insurer has a duty arising out of the policy as a contract to defend as to a claim, or a part of a claim, that is at least potentially covered because it may possibly embrace some triggering harm of the specified sort within the policy period caused by an included occurrence. It has been held *69 that, under principles of the law of contract, the insurer may not obtain reimbursement from the insured for defense costs that can be allocated to a claim that is at least potentially covered: “With regard to defense costs” of this sort, “the insurer has been paid premiums by the insured. It bargained to bear these costs. To attempt to shift them would upset the arrangement.” (Buss v. Superior Court, supra, 16 Cal.4th at pp. 49-50.) Implicit in this holding is the proposition that the insurer may not obtain reimbursement from the insured for defense costs that can be allocated to a part of a claim that is at least potentially covered. It follows that, pursuant to contract, defense costs that can be allocated to a claim, or a part of a claim, that is at least potentially covered cannot be allocated to the insured.

By contrast, the insurer does not have a duty arising out of the policy as a contract—but may have one imposed by law in support thereof—to defend as to a claim, or a part of a claim, that is not even potentially covered because it does not even possibly embrace any triggering harm of the specified sort within the policy period caused by an included occurrence. It has been held that, under principles of the law of restitution, the insurer may obtain reimbursement from the insured for defense costs that can be allocated solely to a claim that is not even potentially covered: “With regard to defense costs” of this sort, “the insurer has not been paid premiums by the insured. It did not bargain to bear these costs. To attempt to shift them would not upset the arrangement. [Citation.] The insurer therefore has a right of reimbursement that is implied in law as quasi-contractual . . . .” (Buss v. Superior Court, supra, 16 Cal.4th at pp. 50-51.) Implicit in this holding is the proposition that the insurer may obtain reimbursement from the insured for defense costs that can be allocated solely to a part of a claim that is not even potentially covered. It follows that, pursuant to quasi-contract, defense costs that can be allocated solely to a claim, or a part of a claim, that is not even potentially covered can be allocated to the insured.

On the allocation of defenses costs to the insured, it is the insurer that must carry the burden of proof, and it must do so by the preponderance of the evidence. What we said above we say here: “Evidence Code section 500 provides that, generally, a party desiring relief must carry the burden of proof thereon,” and “Evidence Code section 115 .. . provides that the burden of proof that is generally applicable is proof by a preponderance of the evidence.” (Buss v. Superior Court, supra, 16 Cal.4th at p. 53 .) We discern no reason to make an exception here.

In the case at bar, the record on appeal discloses this: From 1956 to about 1975, Aerojet had largely typical comprehensive general liability insurance policies and similar instruments, covering specified harm including bodily injury and/or property damage, that were issued by various *70 insurers. From 1976 to 1984, it had “fronting” policies of this sort that were issued by INA. With the exception of INA, as to which Aerojet agreed to pay its own defense costs and indeed to defend itself, 17 the insurers each had a duty to defend all the governmental and private actions in their entirety—to be precise, each had such a duty separate and independent from the others (Continental Cas. Co. v. Zurich Ins. Co. (1961) 57 Cal.2d 27, 37 [ 17 Cal.Rptr. 12 , 366 P.2d 455 ]). In each action, each insurer was presented with what was in substance a single broad “mixed” claim. That claim was predicated on facts, alleged or otherwise disclosed, to the effect that, throughout the course of its operations from the early 1950’s into the 1980’s, Aerojet discharged hazardous substances in an ongoing fashion at its Sacramento site and thereby caused pollution in and around that location resulting in continuous and/or progressively deteriorating bodily injury and/or property damage. That claim, therefore, was at least potentially covered in part under each of the policies pertinent here because it might possibly involve specified harm caused by an included occurrence, 18 with triggering harm of that sort within the policy period in question. Each insurer was “on the risk” for at least one such period. None was on the risk for all. Each, as stated, was presented with a “mixed” claim: At least one of the parts was at least potentially covered because it might possibly embrace some triggering harm of the specified sort within the policy period caused by an included occurrence, and at least one of the parts was not even potentially covered because it did not even possibly embrace any triggering harm of the specified sort within the policy period caused by an included occurrence. Nevertheless, each had a duty, prophylactic although not contractual, to defend all the parts. 19 It is true that INA had no duty to defend whatsoever. But that fact is immaterial so far as the other insurers are concerned. The “fronting” policies created rights and duties between Aerojet and INA. They did not even purport to create any right or duty in either Aerojet or INA as against the world, including the other insurers. Nor could they. As contracts, they were effective only between Aerojet and INA. (Chandler v. Roach (1957) 156 Cal.App.2d 435, 444 [ 319 P.2d 776 ].)

*71 It follows that, if it carries the burden of proof by a preponderance of the evidence, each insurer may allocate defense costs to Aerojet for any part of the single broad “mixed” claim presented in the governmental and private actions that was not even potentially covered because it did not even possibly embrace any triggering harm of the specified sort within its policy period or periods caused by an included occurrence. For example, on the requisite proof, it may allocate defense costs for a part involving acts or omissions that may possibly have caused bodily injury or property damage— whether continuous or progressively deteriorating, on the one side, or discrete, on the other side—only after its policy or policies expired. (Cf. Maples v. Aetna Cas. & Surety Co. (1978) 83 Cal.App.3d 641, 644-650 [ 148 Cal.Rptr. 80 ] [holding that there is no coverage under a comprehensive general liability policy as to acts or omissions that caused specified harm of any kind only after the policy expired].) On same proof, it may also allocate defense costs for a part involving acts or omissions that may possibly have caused bodily injury or property damage—specifically, discrete bodily injury or property damage—only before its policy or policies incepted. (Cf. Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 691 [implying that there is no coverage under a comprehensive general liability policy as to acts or omissions that caused specified, discrete harm only before the policy incepted].)

Aerojet argues against the foregoing conclusion. It fails to persuade. It does not take any account of the fact that the insurer’s duty to defend the entire “mixed” claim is prophylactic and not contractual.

The insurers also argue against the foregoing conclusion. They too fail to persuade.

Generally, the 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.

More specifically, the insurers imply that they may allocate defense costs to Aerojet more broadly, apparently based on the “fronting” policies issued by INA to Aerojet, or at least on the fact that Aerojet chose to avail itself of such policies. They are wrong here as well. Whatever duty to defend each may have had was provided for in its own policy or policies, and was not affected by Aerojet’s subsequent agreement to pay its own defense costs, and indeed defend itself, under INA’s. At bottom, their reasoning seems as *72 follows: If Aerojet had been issued policies by INA that were not of the “fronting” kind, they could have sought “equitable contribution” from INA as to the defense costs (Truck Ins. Exchange v. Amoco Corp. (1995) 35 Cal.App.4th 814, 827 [ 41 Cal.Rptr.2d 551 ]; see generally, Continental Cas. Co. v. Zurich Ins. Co., supra, 57 Cal.2d at pp. 35-38); because it did not, it should be required to contribute itself. Such an inference fails. At the outset, Aerojet was in fact issued “fronting” policies by INA. Had the situation been different, it might have led to different consequences. But it was not. Be that as it may, let us assume that, if Aerojet had been issued policies by INA that were not of the “fronting” kind, INA might have been required to make an equitable contribution to the defense costs. (See, e.g., CNA Casualty of California v. Seaboard Surety Co. (1986) 176 Cal.App.3d 598, 620 [ 222 Cal.Rptr. 276 ].) That assumption, however, does not compel the conclusion that, because it was issued “fronting” policies by INA, Aerojet should be required to make such a contribution itself. Although insurers may be required to make an equitable contribution to defense costs among themselves, that is all: An insured is not required to make such a contribution together with insurers. (Truck Ins. Exchange v. Amoco Corp., supra, 35 Cal.App.4th at pp. 827-828; County of San Bernardino v. Pacific Indemnity Co. (1997) 56 Cal.App.4th 666, 690-691 [ 65 Cal.Rptr.2d 657 ] [following Truck Ins. Exchange].) Equitable contribution applies only between insurers (Truck Ins. Exchange v. Amoco Corp., supra, 35 Cal.App.4th. at p. 828; County of San Bernardino v. Pacific Indemnity Co., supra, 56 Cal.App.4th at pp. 690-691 [following Truck Ins. Exchange]), and only in the absence of contract (see Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 687). It therefore has no place between insurer and insured, which have contracted the one with the other. Neither does it have any place between an insurer and an uninsured or “self-insured” 20 party. (Truck Ins. Exchange v. Amoco Corp., supra, 35 Cal.App.4th at pp. 827-828; County of San Bernardino v. Pacific Indemnity Co., supra, 56 Cal.App.4th at pp. 690-691 [following Truck Ins. Exchange].) 21

It bears repeating: If specified harm may possibly have been caused by an included occurrence and may possibly have resulted, at least in *73 part, within the policy period, the duty to defend perdures to all points of time at which some such harm may possibly have resulted thereafter.

By giving an affirmative answer to the question whether defense costs may be allocated to the insured, the Court of Appeal did not err.

In answering the question of allocation of defense costs as it did, the Court of Appeal erred to the extent that it strayed away from the contractual/quasi-contractual analysis set out above in the direction of vague “fairness” and rough “justice.” 22

The Court of Appeal expressed the view that the “logical conclusion” of such contractual/quasi-contractual analysis “would permit [Aerojet] to have been fully insured, for example, with Transport Indemnity ... for only 1 year out of the 30 years at issue here, then to have carried the subject INA policies for the remaining 29 years but claim a defense from Transport Indemnity for the entire 30-year period.”

The Court of Appeal stated that, under its illustration, Aerojet would not have a right to a defense by Transport Indemnity against an entire claim of bodily injury or property damage spanning all 30 years. It implied that 29/ao of the defense costs would have to be allocated to Aerojet, in the apparent belief that “ ‘. . . the insured must bear its share of those costs determined by the fraction of the time of injurious exposure in which it lacked coverage^] ’ ” (Quoting Gulf Chemical & Metallurgical v. Associated Metals, supra, 1 F.3d at p. 372.)

*74 But, as explained, even in the Court of Appeal’s illustration, Aerojet might indeed have a right to a defense by Transport Indemnity against the entire claim. Its right would be contractual if all the years of the claim were at least potentially covered because year one might possibly have embraced some triggering harm of the specified sort caused by an included occurrence, and year two through year thirty might possibly have embraced some such harm resulting therefrom. By contrast, its right would be prophylactic if some of the years of the claim were at least potentially covered and the others were not, on the ground that the first year might possibly have embraced some triggering harm of the specified sort caused by an included occurrence, but* some of the succeeding 29 years might possibly have embraced some such harm resulting therefrom and the others did not even possibly do so.

As also explained, even though in the Court of Appeal’s illustration Transport Indemnity might indeed be able to allocate defense costs to Aerojet, it would be able to do so only as to the part of the claim that was not even potentially covered. It would not be able to do so in accordance with any mechanical pro rata formula based on the time of noninsurance within the time as a whole. Especially so, because, contrary to the assumption, the duty to defend is not limited to the policy period. 23 If specified harm may possibly have been caused by an included occurrence and may possibly have resulted, at least in part, within the policy period, it perdures to all points of time at which some such harm may possibly have resulted thereafter. 24

Lastly, even though in the Court of Appeal’s illustration Aerojet may be said to have “assumed the risk of defending against claims” during year two through year thirty—strictly speaking, it simply did not shift the risk to another 25 —and hence “should now carry ... the burden,” it did not thereby affect Transport Indemnity’s duty to defend. It “assumed the risk,” and *75 “should now carry ... the burden,” of defending only against claims based on any harm that may possibly have been caused only after Transport Indemnity’s policy expired at the end of year one. Such a risk is surely not negligible, extending as it does from year two through year thirty. But it is Transport Indemnity that “assumed the risk,” and “should now carry ... the burden,” of defending against any claim that is based on specified harm that may possibly have been caused by an included occurrence and may possibly have resulted, albeit only in part, within the policy period in year one. Its duty to defend was triggered when specified harm was possibly caused by an included occurrence, because at least some such harm may possibly have resulted within the policy period in the first year. It extended to all specified harm that was possibly caused by an included occurrence, even if some such harm may possibly have resulted beyond the policy period in the succeeding 29 years. In a word, although the trigger of the duty to defend is limited to the policy period, the extent of the duty to defend is not. (Cf. Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 686 [holding that, although the trigger of the duty to indemnify is limited to the policy period, the extent of the duty to indemnify is not]; Armstrong World Industries, Inc. v. Aetna Casualty & Surety Co., supra, 45 Cal.App.4th at p. 105 [same].)

Beneath 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. (Linnastruth v. Mut. Benefit etc. Assn. (1943) 22 Cal.2d 216, 218 [ 137 P.2d 833 ].) They evidently did so. They thereby established what was “fair” and “just” inter se. We may not rewrite what they themselves wrote. (Phelps v. Allstate Ins. Co. (1980) 106 Cal.App.3d 752, 758 [ 165 Cal.Rptr. 263 ].) We must certainly resist the temptation to do so here simply in order to adjust for chance—for the benefits it has bestowed on one party without merit and for the burdens it has laid on others without desert. (See Third Story Music, Inc. v. Waits (1995) 41 Cal.App.4th 798, 809 [ 48 Cal.Rptr.2d 747 ]; Walnut Creek Pipe Distributors, Inc. v. Gates Rubber Co. (1964) 228 Cal.App.2d 810, 815 [ 39 Cal.Rptr. 767 ].) 26 As a general matter at least, we do not add to, take away from, or otherwise modify a contract for “public policy considerations.” *76 (AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d at p. 818 .) 27 We would certainly not do so here, where such considerations depend in large part on the amassing and analyzing of complex and extensive empirical data, which belong more appropriately to the executive and legislative branches than to the judicial. 28 We shall therefore allow whatever “gains” and “losses” there may be to lie where they have fallen. 29

*77 in

For the reasons stated above, we conclude that we must affirm in part and reverse in part the judgment of the Court of Appeál. Specifically, we must: (1) affirm the judgment to the extent that it holds that site investigation expenses may constitute defense costs that the insurer must incur in fulfilling its duty to defend under standard comprehensive or commercial general liability insurance policies; (2) affirm the judgment to the extent that it holds that defense costs may be allocated to the insured under such policies; (3) reverse the judgment to the extent that it holds that such costs should be allocated pro rata based on the time of noninsurance within the time as a whole; (4) otherwise affirm the judgment; and (5) direct the Court of Appeal to remand the cause to the superior court for proceedings not inconsistent with the views expressed in this opinion.

It is so ordered.

George, C. J., Werdegar, J., and Brown, J., concurred.

As would subsequently appear, Aerojet’s policies included “manuscript” as well as “standard” ones. Policies “are usually issued on standard forms containing terms and conditions drafted by the [insurer]. Often, the insurer is willing to modify or change the standard forms by ‘endorsements’ .... Sometimes, the policy issued is entirely nonstandard and drafted for the particular risk undertaken”—a so-called “manuscript” policy. (Croskey et al., Cal. Practice Guide: Insurance Litigation 1 (The Rutter Group 1997) 3:33, p. 3-6.)

Under CERCLA, the terms “removal” and “remediation” bear the following meanings. “Removal” refers to the “cleanup. or removal of released hazardous substances from the environment, such actions as may be necessary taken in the event of the threat of release of hazardous substances into the environment, such actions as may be necessary to monitor, assess, and evaluate the release or threat of release of hazardous substances, the disposal of removed material, or the taking of such other actions as may be necessary to prevent, minimize, or mitigate damage to the public health or welfare or to the environment, which may otherwise result from a release or threat of release.” ( 42 U.S.C. § 9601 (23).) “Remediation” refers to “those actions consistent with permanent remedy taken instead of or in addition to removal actions in the event of a release or threatened release of a hazardous substance into the environment, to prevent or minimize the release of hazardous substances so that they do not migrate to cause substantial danger to present or future public health or welfare or the environment.” {Id., § 9601(24).)

A “fronting” policy has been described as one “which does not indemnify” or, apparently, defend “the insured but which is issued to satisfy financial responsibility laws of various” *50 jurisdictions “by guaranteeing to third persons who are injured that their claims against” the insured “will be paid.” (Columbia Casualty Co. v. Northwestern Nat. Ins. Co. (1991) 231 Cal.App.3d 457, 471 [ 282 Cal.Rptr. 389 ].)

“[I]f there is a conflict in meaning between an endorsement and the body of the policy, the endorsement controls.” (Continental Cas. Co. v. Phoenix Constr. Co. (1956) 46 Cal.2d 423, 431 [ 296 P.2d 801 , 57 A.L.R.2d 914 ].)

See footnote 4, ante.

A proposed consent decree, which incorporated an earlier agreement, had been lodged with the federal and state CERCLA complaints in 1986, and had subsequently been withdrawn.

In the words of section 300.430(a)(2), which was subsequently added to title 40 of the Code of Federal Regulations: “The purpose of the remedial investigation/feasibility study (RI/FS) is to assess site conditions and evaluate alternatives to the extent necessary to select a remedy. Developing and conducting an RI/FS generally includes the following activities: project scoping, data collection, risk assessment, treatability studies, and analysis of alternatives.” (55 Fed.Reg. 8666, 8846 (Mar. 8, 1990); accord, 40 C.F.R. § 300.430 (a)(2) (1996).)

The County of San Bernardino et al., which have been granted leave to appear as amici curiae supporting Aerojet’s position, have submitted a request for judicial notice of the following commentaries: (1) Elliott, The New Comprehensive General Liability Policy (American Management Association Reprint 1966); and (2) Obrist, The New Comprehensive General Liability Insurance Policy—A Coverage Analysis (Defense Research Inst. Monograph 1966). We deny the request. We may take judicial notice only of matter that is “authorized or required by law.” (Evid. Code, § 450.) The indicated commentaries are not such. (See id.., §§451, 452.) They may nevertheless be consulted for whatever assistance they may furnish. So they were in the past, in decisions including Montrose Chemical Corp. v. Admiral Ins. Co. (1995) 10 Cal.4th 645, 671-672 [ 42 Cal.Rptr.2d 324 , 913 P.2d 878 ]. So they will be now as well.

As stated, prior to 1966, in its insuring clause the standard comprehensive general liability insurance policy covered specified harm, such as bodily injury or property damage, caused by “accident” rather than by an “occurrence.” (Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ¶ 7:26 , p. 7A-8.) As pertinent here, the difference in words does not reflect any difference in substance. (See id., ‘JH 7:25 to 7:32, pp. 7A-8 to 7A-10.)

In Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th 645 , we made the point plain. Hence, the contrary premise on which Justice Chin rests his concurring and dissenting opinion collapses as without support. In Montrose, we noted, and reaffirmed, the “settled rule” of the case law that “an insurer on the risk when continuous or progressively deteriorating [property] damage or [bodily] injury first manifests itself remains obligated to indemnify the insured for the entirety of the ensuing damage or injury.” (Id. at p. 686, italics added.) In Armstrong World Industries, Inc. v. Aetna Casualty & Surety Co. (1996) 45 Cal.App.4th 1 [ 52 Cal.Rptr.2d 690 ], the Court of Appeal observed that, in Montrose, we “relied upon existing case law holding that coverage for a manifested loss is not terminated by the expiration of the policy; coverage continues until the damage is complete.” (Id. at p. 50.) Citing such case law itself, it explained: “[T]he event which triggers an insurance policy’s coverage does not define the extent of the coverage. Although a policy is triggered only if [bodily injury or] property damage takes place ‘during the policy period,’ once a policy is triggered, the policy obligates the insurer to pay ‘all sums’ which the insured shall become liable to pay as damages for bodily injury or property damage. The insurer is responsible for the full extent of the insured’s liability . . . , not just for. the part of the [injury or] damage that occurred during the policy period.” (Id. at p. 105.) In light of the foregoing, commentators have soundly stated: “Courts reject the argument that [an] insurer should only be responsible for [injury or] damage that took place during its policy period . . . .” (Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ^ 8:73.10, p. 8-19, italics in original.)

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.” (Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at pp. 686-687.) The successive insurers are not “jointly and severally liable.” (Id. at p. 681, fn. 19, italics omitted.) Rather, “[a]llocation of the cost of indemnification” among such insurers “requires application of principles of contract law to the express terms and limitations of the various policies” (ibid.) and, in their absence, “equitable considerations” (id. at p. 687).

See footnote 10, ante.

Pace Justice Kennard, who in her concurring and dissenting opinion essentially adheres to views that we previously considered and found wanting. (Compare Buss v. Superior Court, supra, 16 Cal.4th at pp. 44-49 with id. at pp. 62-66 (dis. opn. of Kennard, J.).)

At least as a general matter, under the analysis presented in the text, the costs that the insurer must incur in fulfilling its duty to indemnify and the costsjhat it must incur in fulfilling its duty to defend are mutually exclusive. Indemnification costs, i.e., expenses to resolve liability, are expressly limited in the policy. They arise after the insured’s liability is established and as a result thereof. That is because indemnification presupposes that such liability has actually been established in the past. (See Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 659, fn. 9 .) By contrast, defense costs, i.e., expenses to avoid or at least minimize liability, are not limited by the policy, expressly or otherwise, but impliedly extend to all such expenses as are reasonable and necessary. They arise before the insured’s liability is established and apart therefrom. That is because defense presupposes that such liability may possibly be established in the future. (See ibid.) Thus, at least generally, the same costs cannot be both indemnification costs and defense costs.

In General Acc. Ins. Co. v. State Dept. of Environ. (1996) 143 N.J. 462, 473-479 [ 672 A.2d 1154, 1160-1163 ], in which it applied New Jersey law, but spoke generally, the New Jersey Supreme Court implied to the contrary. It evidently did so because it did not follow a contractual/quasi-contractual analysis such as that set out in the text, but rather strayed— erroneously, in our view—in the direction of vague “fairness” and rough “justice.” (See also pp. 73-76, post.)

Insofar as the insurers’ assumption that site investigation expenses cannot be defense costs to the extent that the underlying site investigation is ordered or requested by the federal or state government is based on the premise that such expenses are indemnification costs within the meaning of AIU Ins. Co. v. Superior Court, supra, 51 Cal.3d at page 824 , i.e., “sums which [the insured] becomes ‘legally obligated’ to pay as ‘damages’ . . . because of ‘[bodily injury or] property damage[,]’ ” it is unsound. There is a suggestion that a “legal obligation” arises from a governmental order and perhaps even from a governmental request. But any such “legal obligation” need not be “to pay . . . ‘damages’ . . . because of ‘[bodily injury or] property damage.’ ” For example, as indicated in the text, a legal “obligation” may be imposed under CERCLA even before specified harm is proved—and even if such harm is subsequently disproved.

In General Acc. Ins. Co. v. State Dept. of Environ., supra, 143 N.J. at page 477 [ 672 A.2d at page 1162 ], and Domtar, Inc. v. Niagara Fire Ins. Co. (Minn.Ct.App. 1996) 552 N.W.2d 738, 751-752 , affirmed in pertinent part, Domtar; Inc. v. Niagara Fire Ins. Co., supra, 563 N.W.2d 724 , the New Jersey Supreme Court and the Minnesota Court of Appeals, respectively, also adopted the approach of the federal district court in Ex-Cell-01 and Ex-Cell-0 II. Although each cited the federal district court, neither presented any reasoning.

In Endicott Johnson Corp. v. Liberty Mut. Ins. Co. (N.D.N.Y. 1996) 928 F.Supp. 176, 184 , appeal and cross-appeal dismissed (2d Cir. 1997) 116 F.3d 53 , the United States District Court of the Northern District of New York held that it would “allocate” Remedial Investigation/Feasibility Study or RI/FS costs as follows: “To the extent that an expense is primarily attributable to remedial investigations ... the expense will be treated as a defense cost. To the extent an expense is primarily attributable to feasibility studies ... the expenses will be treated as damages to be indemnified. Finally, to the extent the Court cannot determine . . . whether an expense is attributable to either RI or FS, the Court will have broad discretion to allocate the expense in an equitable manner.” (Fn. omitted.) The court thought its method was “simple.” (Ibid.) In light of the analysis presented in the text, we think it is simplistic.

In stating that Aerojet agreed to pay its own defense costs and indeed to defend itself, we merely follow the language of the “fronting” policies. Of course, Aerojet did not “contract” with itself to impose an “obligation” on itself.

Or, similarly, by “accident,” etc. (See fn. 9, ante.)

Each may also have had a corresponding right of some sort to require the others to share in discharging the duty or at least to contribute to its costs. (See generally, Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ¶¶ 8:73 .10 to 8:73.19, pp. 8-17 to 8-22 [surveying the topic]; see also Haskel, Inc. v. Superior Court (1995) 33 Cal.App.4th 963, 976, fn. 9 [ 39 Cal.Rptr.2d 520 ] [appearing to assume such a right]; cf. Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at p. 687 [stating that, “leaving aside the availability of excess (multiple) policies or ‘other insurance’ clauses, and absent express policy language decreeing the manner of apportionment of contribution among successive liability insurers, the courts will generally apply equitable considerations to spread the cost [of indemnification] among the several policies and insurers”].)

In a strict sense, “self-insurance” is a “misnomer.” (Nabisco, Inc. v. Transport Indemnity Co. (1983) 143 Cal.App.3d 831, 836 [ 192 Cal.Rptr. 207 ]; accord, Orange County Water Dist. v. Association of Cal. Water etc. Authority (1997) 54 Cal.App.4th 772, 777 [ 63 Cal.Rptr.2d 182 ].) “Insurance is a contract whereby one undertakes to indemnify another against loss, damage, or liability arising from a contingent or unknown event.” (Ins. Code, § 22.) “[S]elf-insurance ... is equivalent to no insurance . . . .” (Richardson v. GAB Business Services, Inc. (1984) 161 Cal.App.3d 519, 523 [ 207 Cal.Rptr. 519 ].) As such, it is “rfepugnant to the [very] concept of insurance . . . .” (Ibid.) If insurance requires an undertaking by one to indemnify another, it cannot be satisfied by a self-contradictory undertaking by one to indemnify oneself.

We find nothing contrary in City of Oxnard v. Twin City Fire Ins. Co. (1995) 37 Cal.App.4th 1072 [ 44 Cal.Rptr.2d 177 ], or Nabisco, Inc. v. Transport Indemnity Co., supra, 143 Cal.App.3d 831 . Each decision holds only that an “excess insurer” does not have a duty *73 to defend an insured until “primary insurance” in the form of a so-called “self-insured retention” is exhausted. (City of Oxnard v. Twin City Fire Ins. Co., supra, 37 Cal.App.4th at pp. 1076-1078; Nabisco, Inc. v. Transport Indemnity Co., supra, 143 Cal.App.3d at pp. 834-837.) The insurers here are not “excess insurers.” By their own admission, they are “primary insurers.”

Following decisions such as Ins. Co. North America v. Forty-Eight Insulations (6th Cir. 1980) 633 F.2d 1212 , 1224-1225, rehearing granted in part and denied in part sub nomine Ins. Co. of N.A. v. Forty-Eight Insulations (6th Cir. 1981) 657 F.2d 814 , and its progeny including Sharon Steel v. Aetna Cas. and Sur. (Utah 1997) 931 P.2d 127, 140-142 , Owens-Illinois, Inc. v. United Ins. Co. (1994) 138 N.J. 437, 479 [ 650 A.2d 974, 995-996 ], and Gulf Chemical & Metallurgical v. Associated Metals (5th Cir. 1993) 1 F.3d 365, 372 . It is perhaps in Owens-Illinois that the concern with “fairness” and “justice” instead of contract and quasi-contract is most evident. There, the New Jersey Supreme Court stated: “When periods of no insurance” along with periods of insurance “reflect a decision by an actor to assume or retain a risk, as opposed to periods when coverage for a risk is not available, to expect the risk-bearer to share in the allocation is reasonable.” (Owens-Illinois, Inc. v. United Ins. Co., supra, 138 N.J. at p. 479 [ 650 A.2d at p. 995 ].) Only if one’s expectation ignores contract in favor of “fairness” and “justice.” For if an actor shifts a risk in some periods and not in others, so far as contract is concerned its omission should generally have the same effect whether it finds its genesis in choice or in compulsion.

To the extent that they state or imply otherwise, such decisions as Ins. Co. North America v. Forty-Eight Insulations, supra, 633 F.2d at pages 1224 to 1225, and its progeny including Sharon Steel v. Aetna Cas. and Sur., supra, 931 P.2d at pages 140 to 142, Owens-Illinois, Inc. v. United Ins. Co., supra, 138 N.J. at page 479 [ 650 A.2d at page 995 ], and Gulf Chemical & Metallurgical v. Associated Metals, supra, 1 F.3d at page 372 , are unsound under the analysis presented in the text.

The result might be different if the standard comprehensive or commercial general liability insurance policy were a kind of “claims made” policy. (See Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at pp. 688-689.) Under a policy of this sort, “the insurer generally is responsible for loss resulting from claims made during the policy period”—and only for loss resulting from claims made during the policy period—“no matter when the liability-generating event took place.” (Helfand v. National Union Fire Ins. Co. (1992) 10 Cal.App.4th 869, 885, fn. 8 [ 13 Cal.Rptr.2d 295 ].) The standard policy, however, is not such. (See Montrose Chemical Corp. v. Admiral Ins. Co., supra, 10 Cal.4th at pp. 688-689.) Neither is any of Aerojet’s policies pertinent here.

Aerojet was effectively uninsured, not “self-insured.” (See fn. 20, ante.)

We are consequently in accord with decisions that have resisted temptation in this regard, such as Armstrong World Industries, Inc. v. Aetna Casually & Surety Co., supra, 45 Cal.App.4th at pages 55 to 57, and Keene Corp. v. Ins. Co. of North America (D.C. Cir. 1981) 667 F.2d 1034 , 1050 [ 215 App.D.C. 156 ], and are contra to those that have not, such as Ins. Co. North America v. Forty-Eight Insulations, supra, 633 F.2d at pages 1224 to 1225, and its progeny including Sharon Steel v. Aetna Cas. and Sur., supra, 931 P.2d at pages 140 to 142, Owens-Illinois, Inc. v. United Ins. Co., supra, 138 N.J. at page 479 [ 650 A.2d at page 995 ]. *76 and Gulf Chemical & Metallurgical v. Associated Metals, supra, 1 F.3d at page 372 . At least to some extent, the latter seem predicated on the assumption that equitable contribution is applicable between an insurer and an uninsured or “self-insuréd” party. As explained in the text, that assumption is unsound.

We therefore cannot, and will not, follow courts that do, such as the New Jersey Supreme Court in Owens-Illinois, Inc. v. United Ins. Co., supra, 138 N.J. at pages 474 to 480 [650 A.2d at pages 993-996].

In Owens-Illinois, Inc. v. United Ins. Co., supra, 138 N.J. at pages 474 to 480 [650 A.2d at pages 993-996], a case involving asbestos, the New Jersey Supreme Court all but conceded as much. It declined to amass and analyze the pertinent empirical data itself in favor of directing the trial court, at page 477 [ 650 A.2d at page 994 ], to “appoint a master, one skilled in the economics of insurance, to” attempt to do so. *77 costs, the insurer incurs indemnification costs. It is unsound. It would threaten to transform all defense costs into indemnification costs. That is certainly true of all defense costs that prove to be well spent, that is to say, all that in fact avoid or at least minimize liability. For defense costs of this sort necessarily obviate indemnification costs.

We observe that Aerojet may indeed have gotten from the insurers more in defense costs than it could have gotten in indemnification costs. But it got no more than it had a right to: Although indemnification costs were limited by the pertinent policies, defense costs were not. The insurers might perhaps have avoided such a pass, as through the issuance of “self-consuming” or “burning limits” policies, under which the indemnification limit is reduced dollar for dollar by defense costs until zero is reached and the duty to indemnify and the duty to defend are then terminated (see Croskey et al., Cal. Practice Guide: Insurance Litigation 2, supra, ¶ 7:656 , p. 7B-34). They apparently did not attempt to do so.

The insurers may be understood to argue that, if the same costs may be characterized as both indemnification costs and defense costs, they should be characterized as the former rather than the latter.

At least as a general matter, as we have explained, indemnification costs and defense costs are mutually exclusive: The latter are expenses to avoid or at least minimize liability that arise before the insured’s liability is established and apart therefrom; the former are expenses to resolve liability that arise after the insured’s liability is established and as a result thereof. Hence, at least as a general matter, the same costs cannot be characterized as both. In any event, they should not be characterized as the former rather than the latter, as it were, by default. No basis exists therefor in the pertinent policies. None shall be created here. (See Domtar, Inc. v. Niagara Fire Ins. Co., supra, 563 N.W.2d at pp. 738-739 [applying Minnesota law, but speaking generally, to the effect that, even if the same costs could be characterized as both indemnification costs and defense costs, they should not be characterized as the former rather than the latter: that the same costs may do double duty as both indemnification costs and defense costs means that they do, in fact, do duty as the latter as well as the former].)

In her concurring and dissenting opinion, in which she invokes the public policy underlying Insurance Code section 533 not to encourage “wilful acts” on the part of the insured by prohibiting indemnification by the insurer for any liability based thereon (e.g., Horace Mann Ins. Co. v. Barbara B. (1993) 4 Cal.4th 1076, 1087 [ 17 Cal.Rptr.2d 210 , 846 P.2d 792 ]; Gray v. Zurich Insurance Co., supra, 65 Cal.2d at pp. 277-278), Justice Kennard creates a novel, and concededly “complicated” (conc, and dis. opn. of Kennard, J., post, at p. 87), scheme to govern when, if ever, the same costs could be characterized as both indemnification costs and defense costs. The premise is that, by incurring defense costs that obviate indemnification