Bien & Summers and Elliot L. Bien, San Francisco, for Amicus Curiae Insurance Environmental Litigation Association.

SILLS, P.J.

This case forces us to confront what happens when the one final judgment rule in appellate procedure collides with the piecemeal adjudication of a liability insurer's two basic duties to its insuredthe duty to defend a lawsuit brought against the insured, and the duty to indemnify if the insured loses that lawsuit. As we explain below, the one final judgment rule is a statutorily based system of rationing access to the appellate courts. Here, we have a purported "appeal" by an insurance company from a minute order adjudicating only one of four of the plaintiff insured's four causes of action against the company. We are no doubt going to disappoint both sides and amicus curiae by dismissing this appeal: They might not be able to agree on much of anything else, but they all want us to reach the merits of the insurer's contentions.

However, fidelity to the rationing scheme enacted by the Legislature requires us to dismiss. The rules of appellate procedure should not be bent just because two well-funded litigants can generate a tremendous amount of paper at the trial level. Of course, the winners of this encounter are not before us, at least directly. They are the many other litigants who have appeals in this court who will not have their matters bumped back in the queue because we bent the rules so that a fraction of an insurance coverage matter could go ahead of them.

The "one final judgment rule" describes the operation of section 904.1 of the Code of Civil Procedure, which specifies the sorts of trial court actions (either "orders" or "judgments") from which an appeal may be taken. The leading case on the one final judgment rule is our Supreme Court's decision in Morehart v. County of Santa Barbara (1994) 7 Cal.4th 725 , 29 Cal.Rptr.2d 804 , 872 P.2d 143 , where the high court took to task a number of appellate decisions originating with Schonfeld v. City of Vallejo (1975) 50 Cal.App.3d 401 , 123 Cal.Rptr. 669 , which had adopted a rule that if a cause of action was "separate and independent" (or some permutation of those words) from the remaining causes of action and could be deemed "severed" from the remaining causes of action, there was a final judgment on that cause of action from which an appeal might be taken. (See Morehart, supra, 7 Cal.4th at pp. 739-740, 29 Cal.Rptr.2d 804 , 872 P.2d 143 .) Indeed, the Morehart court noted that some appellate decisions had not even bothered to ask whether the cause of action deemed severed was separate and independent from the remaining undecided issues. These appellate decisions had cited Schonfeld to justify appellate jurisdiction on the simple theory that the cause of action had "in fact" been severed. (See Morehart, supra, 7 Cal.4th at p. 740 , 29 Cal.Rptr.2d 804 , 872 P.2d 143 .)

Morehart put a stop to the practice of circumventing the one final judgment rule by deeming causes of action "severed" from the balance of the action. After noting that a party has the right to have an interlocutory judgment reviewed on appeal from a final judgment, the court declared: "Accordingly, we hold that an appeal cannot be taken from a judgment that fails to complete the disposition of all the causes of action between the parties even if the causes of action disposed of by the judgment have been ordered to be tried separately, or may be characterized as `separate and independent' from those remaining. Statements to the contrary in Schonfeld, supra, 50 Cal.App.3d 401 , 123 Cal.Rptr. 669 , and its progeny are disapproved." ( Morehart, supra, 7 Cal.4th 725, 743 , 29 Cal.Rptr.2d 804 , 872 P.2d 143 , fn. omitted, emphasis added.)

A moment's reflection will show that the one final judgment rule is not necessary to a functioning system of justice which provides for appellate review of trial court decisions. Unlike, say, the rules which govern the time in which to file appeals (see Cal. Rules of Court, rules 2 and 3), there is no structural or systemic reason for the rule. (Without rules providing for time limits in which to appeal, there would be no final judgments, and without final judgments, you might as well not even have a judicial systemnothing would be certain.) Theoretically at least, one could design a perfectly functional system of justice in which the litigants would be entitled to appellate review from every trial court decision, and efforts are sometimes made in the Legislature to provide for interlocutory appeals. For example, in the juvenile dependency area (where time is extraordinarily critical because the litigation directly affects children's lives) the Legislature has actually provided for a form of interlocutory appeal (or quasi-appeal) by requiring appellate courts to decide certain writ petitions clearly preliminary to the final judgment "on the merits by written opinion" absent extraordinary circumstances. (Cal. Rules of Court, rule 39.1B(o).) [1]

Bernard Witkin, the great maven of California law and particularly appellate procedure, [2] called the one final judgment rule a "fundamental principle of appellate practice in the United States." (See 9 Witkin, Cal. Procedure (4th ed. 1997) Appeal, § 58, p. 113.) However, the theory behind the rule is not that it is necessary to a functioning system of justice, but the oppressiveness and cost of "piecemeal" disposition and review. (Ibid.; see e.g., In re Matthew C, supra, 6 Cal.4th at p. 393, 24 Cal.Rptr.2d 765 , 862 P.2d 765 ; Rao v. Campo (1991) 233 Cal.App.3d 1557, 1565 , 285 Cal.Rptr. 691 ; Kibrej v. Fisher (1983) 148 Cal.App.3d 1113, 1115 , 196 Cal.Rptr. 454 .)

The "cost" rationale for the rule needs some elaboration, because the cost of piecemeal review is not confined to the litigants themselves, and sometimes may be willingly and eagerly borne by them. In the case before us, for example, each side is well-funded and eager to pay the additional costs of piecemeal disposition; we doubt, however, that the united front presented by appellant and respondent here would be the case if one or both of the litigants were much poorer. It takes little imagination to think of a situation where the allowance of piecemeal interlocutory appeals could readily serve to "oppress" the poorer side by driving its costs up.

But quite apart from the costs to the litigants is the cost to the taxpayers and other litigants. The Court of Appeal is what is sometimes referred to as a "court of error." (Unlike the Supreme Court, which is a "court of review.") The entitlement to an appeal to a panel of three judges who must state the reasons for their decision in writing (see Cal. Const, art. VI, § 14) affords litigants a vital protection against prejudicial error. That entitlement, however, is a right which requires tax dollars, and tax dollars are finite. Allowing for interlocutory appeals as a matter of right in addition to appeals from final judgments obviously increases the required work of the appellate courts, and the commitment of resources necessary to do that work. The one final judgment rule, then, represents the outcome of what is, quintessentially, a legislative policy decision to limit access to the Courts of Appeal in the context of finite limits on the tax dollars which the state can devote to its appellate court system. If the Legislature wanted to amend section 904.1 of the Code of Civil Procedure to provide for interlocutory appeals, it could certainly do so and gain the thanks of many litigants; but there would be costs, significant delays in other cases if no extra money were committed to the appellate court, and money might have to be taken from other projects. As it is, the Legislature has chosen to ration the right to appeal by limiting it to the one final judgment in a case, orders after that final judgment, and several miscellaneous categories of orders.

Second, we must review a few basics in the law of liability insurance coverage; when we do so the problem posed by the one final judgment rule in appellate law becomes clear.

Liability insurance is often referred to as "third party" insurance because, unlike other insurance which is bought to protect the insured's life, health or property, liability insurance is purchased to protect the insured (or "policyholder") against certain kinds of claims made by a third party. [3] Typically, and certainly where comprehensive general liability (CGL) insurance policies are involved, liability insurance entails two distinct obligations of the insurer, each of which, however, is triggered under different circumstances. (For a much more comprehensive statement of the rules governing these two duties, see Buss v. Superior Court (1997) 16 Cal.4th 35, 45-9 , 65 Cal.Rptr.2d 366 , 939 P.2d 766 .)

First there is the duty to defend. When this duty properly arises, it comes into existence upon a request for a defense, and lasts until the underlying lawsuit is concluded. ( Montrose Chemical Corp. v. Superior Court (1993) 6 Cal.4th 287, 295 , 24 Cal.Rptr.2d 467 , 861 P.2d 1153 ["The defense duty is a continuing one, arising on tender of defense and lasting until the underlying lawsuit is concluded ... or until it has been shown that there is no potential for coverage...."].) The duty to defend requires the insurance company to mount and fund a defense ( Buss, supra, 16 Cal.4th at p. 46 , 65 Cal.Rptr.2d 366 , 939 P.2d 766 ), which in practical terms means either providing the policyholder with legal representation, or paying for a lawyer or lawyers selected by the policyholder. (Cf. San Diego Federal Credit Union v. Cumis Ins. Society, Inc. (1984) 162 Cal.App.3d 358 , 208 Cal.Rptr. 494 [discussing problem of counsel selected by insurance company in context of reservation of rights over certain coverage issues and holding that in certain circumstances the insurance company must pay for additional counsel selected by policyholder].)

For our purposes, the most important aspect of the duty to defend is that it, virtually by definition, must arise before the second of the insurer's two duties, which is to indemnify the policyholder for claims actually covered by the insurance policy "in light of the facts proved." (See Buss, supra, 16 Cal.4th at p. 46 , 65 Cal. Rptr.2d 366, 939 P.2d 766 .) The two duties, of course, parallel the natural sequence of a lawsuit. First there is litigation. The policyholder needs counsel. The insurer provides itor doesn't, if it doesn't believe that there are any potentially covered claims. If there is a judgment in the underlying suit, the insurer pays the part of it which is actually covered by the substantive terms of the insurance policy.

Because of their sequential order, the two duties present an interesting relationship to each other when both are the subject of coverage litigation. In an action where all the claims are potentially covered by the duty to indemnify, the insurer has a duty to defend. (See Buss, supra, 16 Cal.4th at p. 47 , 65 Cal.Rptr.2d 366 , 939 P.2d 766 .) If none of the claims are potentially covered by the duty to indemnify, then the insurer has no duty to defend. (Ibid.) And if some of the claims are potentially covered and some are not, there is a duty to defend the claims which are potentially covered. (Id. at p. 48, 65 Cal.Rptr.2d 366 , 939 P.2d 766 .) These rules are inherent in the fact that the policy promises a defense, but no one can really know at the outset of the claim against the policyholder whether a particular claim that is "potentially covered" will result in an actual duty to indemnify based on facts proved in the underlying action. (See id. at p. 46, 65 Cal.Rptr.2d 366 , 939 P.2d 766 .)

The sequential relationship between the duty to defend and the duty to indemnify, however, creates certain conundrums in the context of appellate procedure. As we have just seen, the duty to defend can exist when no duty to indemnify ultimately arises, and for that reason it is often said that the duty to defend is "broader" than the duty to indemnify. (E.g., Buss, supra, 16 Cal.4th at pp. 46-47, 65 Cal.Rptr.2d 366 , 939 P.2d 766 .)

But another aspect of the same thought is that where there is no duty to defend, there is, a fortiori, no duty to indemnify: No claims were ever even potentially covered, and therefore no judgment on those never-even-potentially-covered claims need be paid. By this token, from the point of view of the judicial process (as distinct from substantive insurance law), a one-way dependency exists between the two duties. If there is no duty to defend at the inception of or arising during the litigation of the underlying case, then there is absolutely no duty to indemnify any judgment in the underlying case, and therefore certainly no bad faith in denying the request for a defense of the underlying case. But it does not work the other way around: If there is a duty to defend, one cannot say whether there will be a duty to indemnifythere might not be.

Thus both the duty to indemnify and the duty to defend are implicated when a policyholder tenders a case to the insurance company. And therefore, typically, when an insurance company denies the request, any ensuing litigation will usually follow one of two scenarios: One, the insurance company will first file a complaint for declaratory relief, in which it will seek a judgment declaring that it has no duty to defend the policyholderand therefore has no duty to indemnify the policyholder in case the policyholder holder loses. Two, the policyholder will file first, in which case the complaint is likely to seek not only a declaration that the insurance company owes the policyholder a defense of the underlying action, but indemnification for any judgment against the policyholder in the underlying litigation, plus damages for breach of contract for failing to defend, damages for the anticipated breach of contract in failing to indemnify, and, many times, damages for breach of the implied covenant of good faith and fair dealing for unreasonably denying the request for a defense.

When the policyholder files first, it is the rare case where there are no claims beyond the request for a defense. The policyholder usually also seeks indemnification and, many times, tort damages for bad faith. By contrast, the insurance company often seeks a declaration that it has no duty to defend, and therefore no duty to indemnify. The upshot is that when the insurance company wins on the duty to defend question, there will necessarily be a final judgment from which the policyholder can appeal: The determination in the insurance company's favor disposes of all issues in the case. The policyholder is free to appeal; the one final judgment rule doesn't stand in the way. By contrast, if the policyholder wins on the duty to defend issue, there will still be "miles to go" before there will be a final judgment from which the insurer can appeal.

Now to the case before us, which we will only describe in the broadest of outlines. We are, after all, not deciding the case on the merits now. [4] Over the years the EPA [5] sent notices regarding administrative actions (some of which would later turn into lawsuits) against St. Joe Minerals Corporation for a number of its sites around the country where it had performed mining and smelting operations. St. Joe is a New York entity which, since its purchase by the Fluor Corporation, has its principal place of business in Irvine, California. In March 1992, St. Joe requested a "defense" of these claims (i.e., that its attorney fees in responding to them be paid for) from a group of primary and excess insurers, including defendant Zurich Insurance Company, a primary insurer. Zurich turned down the request and in July 1992 sued St. Joe for declaratory relief in a federal district court in Missouri. In October 1992 St. Joe responded with this action in Orange County. [6]

St. Joe's complaint listed at least 39 insurers who, it was alleged, were each responsible to defend and indemnify St. Joe against claims at no less than 14 sites around the country, one of which was in California. The complaint listed four causes of action: (1) declaratory relief on the duty to defend, (2) declaratory relief on the duty to indemnify, (3) breach of the duty to defend, and (4) breach of the duty to indemnify "[t]o the extent" that any of the claims "had progressed sufficiently." There was no request for an injunction requiring any of the insurers to pay for defense costs in any of the underlying actions pendente lite.

Almost three years later St. Joe filed a motion for summary adjudication of Zurich's duty to defend the underlying claims involving 11 sites. (Of these 11 sites, 6 involved only administrative action, 2 involved full-fledged lawsuits instituted by the EPA, and 3 involved a contribution action by an entity that had purchased property from St. Joe and was now itself the target of administrative action.) After taking the matter under submission, the trial court granted St. Joe's motion by filing a minute order on May 11, 1995, to that effect. Later Zurich made a motion for reconsideration, and that was denied as set forth in a minute order filed July 21, 1995. This "appeal" began with a notice of appeal filed August 9, 1995, declaring that Zurich appealed to this court "from the minute order entered and served in this action on May 11, 1995 ... and the minute order entered on July 21, 1995." [7] As the case approached oral argument, we requested supplemental briefing on the appealability question, and also whether circumstances existed justifying "saving" the appeal by treating it as a writ petition (to use Witkin's words, see 9 Witkin, Cal. Procedure (4th ed. 1997) Appeal, § 88, p. 147). Both parties now argue that the case is appealable under the "collateral matter" exception to the one final judgment rule (see id. at § 60, p. 116); St. Joe argues that it is also appealable as a permanent injunction (see Code Civ. Proa, § 904.1, subd. (a)(6)), and both parties also ask that, if it is not appealable, it be treated as a writ petition.

The duty to defend is not a "collateral matter" in liability insurance coverage litigation. There are three reasons.

First, there is no doubt that an adjudication of the duty to defend is "important and essential" to the main issue in any liability insurance coverage case in which the complaint seeks both indemnification and defense. The test of the collateral matter exception, says Witkin, is whether the order "is important and essential to the correct determination of the main issue"if so, it's not collateral. (9 Witkin, Cal. Procedure (4th ed. 1997) Appeal, § 60, p. 116.) That is what we have here. As we have seen, when the insurance company prevails on the duty to defend issue, an order adjudicating the duty to defend is dispositive of the indemnification issue as well. No possibility of coverage (defense) means no actual coverage (indemnification).

But even where the policyholder prevails on the duty to defend issue, the basic legal considerationsin particular the scope of language in the policywill bear on the outcome of the indemnity claim as well. For example, where the underlying case is one for environmental cleanup, undisputed facts may establish that the policyholder (say, a large chemical company) should as an objectively reasonable personhave "expected or intended" certain damage from pollution it put into a waste system at its plant, but still might not have subjectively intended or expected some or all of that damage. (Cf. Shell Oil Co. v. Winterthur Swiss Ins. Co. (1993) 12 Cal. App.4th 715, 743-748 , 15 Cal.Rptr.2d 815 .) Adjudication on appeal of the legal question of whether the words "intended or expected" as used in a CGL policy should have an objective or subjective meaning could be dispositive of the duty to defend issue if the insurer prevailed (and the court ruled that it had an objective meaning), but in any event would constitute a vital determination in the scope of the indemnity which the insurer owed, because the court's determination on the objective-subjective problem in the context of the defense issue would also furnish the test for later ascertaining whether indemnity was owed in specific instances.

We may also illustrate the interconnection between the duty to defend and the duty to indemnify with issues present in the very case before us. If, for example (and as we are urged by Zurich in the briefing on the merits) this court, were to apply what is now the definitive California rule that administrative proceedings are not "suits" within the meaning of the CGL (see Foster-Gardner, Inc. v. National Union Fire Ins. Co., supra, 18 Cal.4th 857 , 77 Cal.Rptr.2d 107 , 959 P.2d 265 ), it would not only affect the duty to defend, but it also might affect whether the insurer would have to pay site investigation costs. (See id. at p. 886, 77 Cal.Rptr.2d 107 , 959 P.2d 265 [holding that site investigation costs incurred before instigation of a lawsuit are not "defense" costs].) Are site investigation costs "defense" or "indemnity"? Indeed, are remediation costs undertaken in the face of EPA action "defense" or "indemnity"? We need not answer these questions now; merely to ask them shows that the concepts are so intertwined that it is impossible to say that the duty to defend is merely "collateral" to the duty to indemnify.

At oral argument the parties developed the theme that environmental insurance coverage litigation somehow merits special treatment because the duty to defend underlying environmental claims may entail expensive items such as investigation and remediation.

That argument, ironically, only underscores the fundamental interrelatedness between the duty to defend and the duty to indemnify in the environmental insurance coverage context. No doubt St. Joe would ultimately want the courts to adopt the theory that site investigation and site remediation costs incurred in advance of any actual lawsuit are "defense" costs, because it is easier for a policyholder to win on the defense issue than the indemnity issue. If such costs were held to be, in substance, indemnity, the policyholder would not receive the benefit of the potentiality rule. (E.g., Horace Mann Ins. Co. v. Barbara B. (1993) 4 Cal.4th 1076, 1081 , 17 Cal.Rptr.2d 210 , 846 P.2d 792 .) By the very same token an insurance company will want to characterize these costs as "indemnity," because it would give it a fighting chance to avoid liability for them. Fortunately, we need not resolve the problem now, except maybe to note that, just as light has characteristics of both particles and waves, site investigation and remediation costs might be seen (under Missouri or New York law) as having characteristics of both defense and indemnity, something which only confirms their interrelatedness.

Second, mere adjudication of whether there is a duty to defend does not fit the traditional basis for application of the collateral matter exception, which is that the interlocutory order requires the payment of money or some act. (See, e.g., Sjoberg v. Hastorf (1948) 33 Cal.2d 116, 119 , 199 P.2d 668 ["It is not sufficient that the order determine finally for the purposes of further proceedings in the trial court some distinct issue in the case; it must direct the payment of money by appellant or the performance of an act by or against him."].) In the present case (and despite St. Joe's strenuous arguments to the contrary), the two 1995 minute orders (i.e., the one on summary adjudication and the one denying reconsideration) did not, by their terms, require either the payment of money or the performance of an act. Strictly speaking, those minute orders merely resolved one dispute at the trial level, albeit one involving a duty which, if it indeed did exist, would run concurrent with the trial itself. In fact, in this appeal, there is no order to pay any money, and even what constitutes "defense" costs is itself a major dispute between the parties. [8]

Disputes over the words in insurance contracts are classic questions of law because they focus on the interpretation of language. The fundamental nature of the relief (even declaratory relief) is legal, not equitable. While there is no adequate remedy at law for such genuine interlocutory orders to pay money such as pendente lite support orders, the same cannot be said for declarations concerning an insurer's duty to defend. It is true that substantive insurance coverage law may place insurers in a bind in regard to the assertion of their rights (because the insurer risks more if the insurer guesses in its favor the first time, but is later judged to have been wrong). But there is nothing inadequate about the traditional legal solution: Insurers can either refuse to defend at all and take their chances that their decision will, in hindsight, be approved by the courts, or defend under a reservation of rights, and take their chances that they may not be able to recoup defense money which they ultimately will be shown not to have owed. (See Eigner v. Worthington (1997) 57 Cal.App.4th 188, 197-198 , 66 Cal. Rptr.2d 808 .)

A variant of the idea that the minute orders here require the interlocutory payment of money is that they are, in effect, appealable as orders granting an injunction. (See Code Civ. Proc, § 904.1, subd. (a)(6) [orders granting injunctions are appealable].) There are three reasons this argument fails: The adjudication is not an injunction in form, or substance, and it would be grossly unfair in the context of insurance coverage litigation to retroactively deem such orders to be injunctions.

First, as to form, the plain fact is that this appeal is, as the notice of appeal says, from minute orders granting a motion for summary adjudication on a point of declaratory relief. This appeal involves no equitable order, enforcible by contempt, requiring Zurich to do anything.

Second, as to substance. St. Joe did not even seek an injunction in its original complaint. It asked, as is appropriate for this sort of litigation, for declaratory relief and damages. That is because there is no doubt that St. Joe has an adequate remedy at law for Zurich's failure to pay defense costs (assuming for sake of argument that Zurich is ultimately judged responsible for them). Characterizing the minute order as an injunction is just a creative way of trying to shoehorn a minute order into something otherwise appealable.

As if to underscore that this case does not entail a final judgment between St. Joe and Zurich, the parties have filed no less than two additional appeals (G023064 and G025002) since they filed this one (G018280). The first appeal (G023064) now in the pipeline has been brought by St. Joe, which apparently lost a summary adjudication as to the duty to defend on a particular site not at issue in this appeal. That appeal is otherwise irrelevant to the present matter except insofar as it shows that more issues remain to be adjudicated. The second appeal (G025002) appears to be a post hoc attempt to cure the nonappealability of the present appeal (G018280), and its nature requires some explication.

In 1995, just after the present appeal was filed, this court, on its own motion, filed an order stating that "[t]he court is considering dismissing the appeal on its own motion as its appears to be taken from nonappealable orders." Thereafter we issued an order saying that the court would decide the appealability issue in conjunction with the merits (which is what we are doing now).

On March 12, 1999, as oral argument loomed, Zurich filed an appeal from an order filed January 28, 1999 whichafter some considerable explicationappears to indeed require the immediate payment of money. [9] Obviously we need not decide now whether this January 1999 order is appealable, because that order is the subject of a new appeal and will come to us in due course. [10] However, in the present appeal the parties argue that the new 1999 order somehow makes the old 1995 orders appealable, for which proposition they rely on American Motorists Ins. Co. v. Superior Court, supra, 68 Cal.App.4th 864 , 80 Cal.Rptr.2d 621 .

American Motorists was a 1998 opinion that required the appellate court to allude to a prior 1993 opinion on which the Supreme Court had granted review. The earlier opinion had held that a 1989 order from a summary adjudication motion on the duty to defend, which was later "enhanced" by a 1991 "enforcement" order directing compliance with the first order, were "together" appealable because nothing further remained to be litigated on the duty to defend issue. (See id. at p. 873, 80 Cal.Rptr.2d 621 ["We held that the trial court's resolution of the duty to defend issues by means of a summary adjudication order later enhanced by a second order directing compliance with the first order were together appealable because `nothing further remained to be litigated in this action with respect to the duty to defend.'"].)

There are two reasons American Motorists is not good authority for the idea that an old (nonappealable) minute order adjudicating the duty to defend can somehow be made appealable by having a later "enforcement" order entered.

First, anything American Motorists had to say about appealability was dicta. Its discussion of the appealability question was simply a reiteration of the previous opinion, rendered in 1993, which did not have the benefit of the Supreme Court's 1994 Morehart decision. American Motorists started out as a 1993 published Court of Appeal decision entitled Montrose Chemical Corp. v. American Motorists Ins. Co. The Supreme Court then granted review on a "grant and hold" basis. Eventually the Supreme Court dismissed review and remanded the case back to the Court of Appeal, which then issued a new opinion in 1998, in which its old opinion, now "depublished" by virtue of the grant and hold (see rule 976(d) Cal. Rules of Court) was still the "law of the case." (See American Motorists Ins. Co. v. Superior Court, supra, 68 Cal.App.4th at p. 869 & 869, fn. 2, 80 Cal.Rptr.2d 621 .) Appealability, however, was not before the court the second time around; the issue was which side must bear the burden on the question of the reasonableness of attorney fees. (See American Motorists Ins. Co. v. Superior Court, supra, 68 Cal.App.4th at p. 867 , 80 Cal.Rptr.2d 621 .) The opinion's comments on appealability were only "local color" used to describe legal terrain which, because it was "law of the case," had become unalterable.

Second, the dicta in American Motorists is suspect if applied beyond its actual facts, which entail a pre-Morehart decision operating as law of the case. The American Motorists opinion described its earlier opinion as holding that the second enforcement order "`in effect severed the duty to defend from the remaining issues.'" (Id. at p. 873, 80 Cal.Rptr.2d 621 .) The whole "severance" rationale as articulated in Schonfeld v. City of Vallejo, supra, 50 Cal.App.3d 401 , 123 Cal.Rptr. 669 was condemned in the later Morehart decision. (See Morehart, supra, 7 Cal.4th at pp. 743-744, 29 Cal.Rptr.2d 804 , 872 P.2d 143 .) There is no reason to believe that the rationale of the first opinion on appealability would not now meet the same fate that met the other Court, of Appeal opinions which were disapproved in Morehart . (See id. at pp. 743 & 743 fn. 11, 29 Cal. Rptr.2d 804 , 872 P.2d 143 .)

The last refuge of a defective appeal from a nonappealable order is the power of an appellate court to treat an appeal from a nonappealable order as a petition for a writ of mandate. (E.g., Olson v. Cory (1983) 35 Cal.3d 390, 400-401 , 197 Cal. Rptr. 843 , 673 P.2d 720 .) No less an authority than Morehart points out that treating appeals from interlocutory judgments as writ petitions is an "efficient avenue" for obtaining appellate review. ( Morehart, supra, 7 Cal.4th at p. 743 , 29 Cal.Rptr.2d 804 , 872 P.2d 143 .)

The salient aspects of the treat-as-writ petition procedure are that it is discretionary, and "should not" be exercised "except under unusual circumstances." ( Olson v. Cory, supra, 35 Cal.3d at p. 401 , 197 Cal. Rptr. 843 , 673 P.2d 720 .) After all, if an appellate court has the sua sponte duty to dismiss an appeal from a nonappealable order (see 9 Witkin, Cal. Procedure (4th ed. 1997) Appeal § 14, pp. 73-74 and the many authorities cited there), the strong presumption must be against treating a defective appeal as a writ, lest the jurisdictional nature of appealability be subverted by causing the exception to swallow the rule. [11]

At least seven factors may be extracted from the cases bearing on the exercise of the court's discretion. We do not mean to offer these factors as an exhaustive list, to suggest that they are all of equal value, or even to suggest that one or more could not be dispositive. They do, however, allow us in this case to come to a reasoned and principled basis as to whether to exercise our discretion. The need for some sort of structure is particularly important here, because of the voluminous size of the record. It is a testament to the byzantine nature of litigation in the late 20th Century that the clerk's transcript in this appeal from a duty-to-defend summary adjudication order is no less than 27count `em, 27volumes. There is a temptation to treat this case on the merits just so we can have the record hauled out of our courthouse.

But that temptation must be resisted. Wealthy litigants (like both sides here) who can afford to hire law firms which become virtual billable hour machines chewing up whole forests of paper should not receive precedence over poorer litigants whose attorneys conduct their cases with enforced, spartan economy. To go back to our introductory remarks: The one final judgment rule is a way of rationing entitlement to the appellate courts. It would be a perverse circumlocution of the Legislature's allocation of resources to routinely allow an indirect entitlement just because the lawyers for the parties can generate a massive quantity of paper.

The factors which give structure to our decision are:

(1) Whether the defectiveness of the appeal was clear in advance. (See Olson v. Cory, supra, 35 Cal.3d 390, 401 , 197 Cal. Rptr. 843 , 673 P.2d 720 ["the issue of appealability was far from clear in advance"].)

(2) The degree to which the parties might have been misled by case law into believing that the order was appealable. (See People v. Garrett (1998) 67 Cal. App.4th 1419, 1423 , 79 Cal.Rptr.2d 803 ["because appellant filed his appeal in reliance on" a certain arguably incorrect case saying the particular kind of order involved was appealable].)

(3) The degree to which treatment as a writ petition facilitates judicial economy. (See Lechuza Villas West v. California Coastal Com. (1997) 60 Cal.App.4th 218, 231 , 70 Cal.Rptr.2d 399 ["Judicial economy would not be served in this case...."]; U.S. Financial v. Sullivan (1974) 37 Cal.App.3d 5, 12 , 112 Cal.Rptr. 18 ["it is unthinkable to permit this complex case to go to trial on only some of the counts because of an erroneous ruling by the trial court on defendant's demurrers when to do so would almost certainly result in an eventual appeal and reversal of the judgment because of this error"]; see also People v. Garrett, supra, 67 Cal. App.4th at p. 1423 , 79 Cal.Rptr.2d 803 ["In the interests of judicial economy...."].)

(4) The need to treat the proceeding as a writ petition because of the in adequacy of a remedy by appeal. (See Genger v. Delsol (1997) 56 Cal.App.4th 1410 , 1431 fn. 23, 66 Cal.Rptr.2d 527 , ["We deny Sachiko's request to treat her appeal as a writ petition. Sachiko has an adequate legal remedy in an appeal ...."]; cf. Vasquez v. Superior Court (1971) 4 Cal.3d 800, 806-807 , 94 Cal.Rptr. 796 , 484 P.2d 964 [hearing writ petition on merits because plaintiffs could not appeal from the disputed order].)

(5) The degree to which the case presents a public interest issue of statewide importance. (See California Casualty Ins. Co. v. Municipal Court (1998) 66 Cal.App.4th 1410, 1413 , 78 Cal.Rptr.2d 657 ["there is no public interest issue of statewide importance"].)

(6) The degree to which the briefing, already completed, covers the dispositive issues in the case. (See Connell v. Superior Court (1997) 59 Cal.App.4th 382, 394 , 69 Cal.Rptr.2d 231 ["because the merits of the dispositive issues have been fully briefed"]; Ponce-Bran v. Trustees of Cal. State University (1996) 48 Cal.App.4th 1656, 1662 , 56 Cal.Rptr.2d 358 ["More importantly, the determinative issue in this case ... is neither sharply in focus nor thoroughly briefed...."].)

(7) Whether the interests of justice and the need to "prevent unnecessary delay" require treatment as a writ petition. (See Clovis Ready Mix Co. v. Aetna Freight Lines (1972) 25 Cal.App.3d 276, 281 , 101 Cal.Rptr. 820 ["We are justified in disposing of the procedural problems in a businesslike fashion in the interests of justice and to prevent unnecessary delay"].)

One factor which we do not consider important is the fact that the parties are willing to stipulate to having the appeal treated as a writ. [12] While the objection of one party might be important in declining to so treat a defective appeal, the fact that the parties both want to have it treated as a writ is of no consequence because of the fundamental rule that parties cannot confer appellate jurisdiction on an appellate court which is otherwise not present. ( Don Jose's Restaurant, Inc. v. Truck Ins. Exchange (1997) 53 Cal. App.4th 115 , 118-119 & p. 119 fn. 5, 61 Cal.Rptr.2d 370.) Stipulation to treat a defective appeal as a writ is, in substance, an attempt to confer jurisdiction on the appellate court where none otherwise exists.

In the present case, the factors point in the direction of declining to deem the defective appeal as a writ petition.

(1) The defectiveness of the appeal was clear in advance. It has long been established that minute orders from summary adjudications are not appealable.

(2) Likewise, there is no case authority squarely holding that summary adjudications on the duty to defend issue are appealable; the dicta in the American Motorists decision was not available in 1995 when this appeal was filed, and American Motorists did not become a final decision until late March 1999, when the Supreme Court denied a petition for review.

(3) Treating this appeal as a writ petition would facilitate judicial economy in this case only if one side prevails on the merits of the duty to defend issue, and to know that, we would first have to determine the case on the merits, which means that any judicial economy is illusory. [13] Yes, it is true that a decision now in the insurance company's favor would indeed administer a vorpal blow to the galumphing, forest-eating jabberwock into which this litigation has turned. But the whole point of reaching the merits should be to do so impartially, not to have a pre-set disposition to go snicker-snack to a case and fell it by ruling in one party's favor. There is no judicial economy if the other party should prevail.

(4) There is no question that Zurich will have an adequate remedy by way of appeal from the one final judgment.

(5) The case does not present any public interest issue of statewide importance. If we apply California law, it now appears that a case that is already on the books, Foster-Gardner, will be dispositive, so this case will have nothing to add to it. If we apply either Missouri or New York law, our decision will have almost no effect on the law of California. While the interests of these litigants are no doubt vital to them, the harsh fact is that this case now appears that it will affect only these litigants.

(6) The briefing, while intensive, does not cover all the issues necessary to dispose of the indemnity claims. As we have said above, assuming that the policyholder wins, there will still be miles to go before the case can sleep. In this regard, there is no need to treat the case as a writ petition lest the briefing and record already filed go to waste. If and when this court sees an appeal from an appealable order or judgment the record here can be used again. [14]

(7) The prevention of delay, like judicial economy, is also only marginally facilitated by treating the defective appeal as a writ petition. Again we are up against the conundrum in coverage litigation that resolution of the duty to defend issue only works one way, so prevention of delay does not have much relevance in this context. Moreover, this case is still going on at the trial level, and has already generated at least one other appeal in which the winner this time round was the loser. And, ironically, this may be a case where a little delay could actually facilitate judicial economy. We like to think that it is not totally out of the realm of possibility that the high courts of New York and Missouri might read our own Supreme Court's tour de force decision in Foster-Gardner and, persuaded by its analysis, announce that it sets forth a rule of law that is coterminous with the law of those states. [15] If that happens, much of this case will go away. But even if those courts are not persuaded and articulate or adopt a diametrically opposite rule, the fact that they will have confronted Foster-Gardner will make our decision easier, because the law of Missouri and New York will be the clearer for it. Allowing the time for the courts of Missouri and New York to read Foster-Gardner while this case still percolates at the trial level without final judgment is only going to make the ultimate decision on the merits easier.

While we must dismiss this appeal even though both sides want us to hear the merits, it bears noting that the parties are not without a remedy. Particularly in environmental coverage caseswhere the litigants are both likely to be well-funded the duty to defend issue would appear to be tailor-made for alternative dispute resolution. There are now many distinguished retired trial, appellate and supreme court judges working in alternate dispute resolution who are just as capable of divining what the supreme courts of Missouri or New York will say on the question of site investigation and remediation costs as we are. If the parties here are that desperate for an adjudication of the (nonappealable) duty to defend, the logical thing is to commit the issue to a panel of retired judges or justices, not to try to circumvent the well-established rules which ration access to the publicly-funded appellate courts.

The appeal is dismissed. We will not treat it as a writ petition. In the interests of justice both sides will bear their own costs.

CROSBY, J., and SCOVILLE, J. [*] , concur.

[*] The Supreme Court ordered that the opinion be not officially published. (See California Rules of CourtRules 976 and 977.)

[1] Indeed, the special provision for interlocutory appeals in the juvenile law is all the more remarkable for being an exception to the basic structure of appellate law because the provision was the result of the Legislature's reaction to In re Matthew C. (1993) 6 Cal.4th 386 , 24 Cal.Rptr.2d 765 , 862 P.2d 765 , where the court simply applied the logic of the one final judgment to hold that the propriety of an order setting a hearing for the final termination of parental rightsan issue which might have been raised by writ petition from the order at a 12- or 18-month review which would necessarily occur several months prior to the final judgment terminating parental rightscould be raised in an appeal from the actual final judgment of termination. Previously, a number of appellate courts had adopted a "writ it or lose it" approach, holding that any attack on the 12 or 18 month review order had to be made by writ petition, and the court would not consider the attack in the appeal from the final judgment. The Legislature compromised the tension between children's need for speedy resolution and basic appellate law by rewriting subdivision ( l ) of section 366.26 of the Welfare and Institutions Code to provide that the correctness of the order setting the final termination hearing could be appealed, but only if certain requirements were met, namely that a writ be taken and denied without a written opinion on the merits. The Judicial Council then responded with amendments to the rules of court to provide for what are, in substance, interlocutory appeals from orders setting final termination hearings. In practical effect, a "39.1B writ" is an appeal from an interlocutory judgment.

[2] See generally In Memoriam (1996) 13 Cal.4th 1267 and particularly page 1268 (remarks of Justice Chin): "For me, Bernie Witkin and California law have always been synonymous." Of course, many of us honor his strictures against footnotes in legal writing (see id. at p. 1269) in the breach.

[3] One of the issues in this case is whether activities of the Environmental Protection Agency prior to filing a complaint (i.e., notifying a company that it is a potentially responsible party for pollution and requiring remediation) constitute a "suit" under standard comprehensive liability insurance policies, triggering an insurer's duty to defend. Foster-Gardner, Inc. v. National Union Fire Ins. Co. (1998) 18 Cal.4th 857 , 77 Cal.Rptr.2d 107 , 959 P.2d 265 provides the dispositive answer to that question under California law; whether California law should be applied, however, is another question, going to the merits of the case, and one which we do not now address. Our use of the word "claim" should not be construed as having any substantive bearing on the issue one way or the other.

[4] Thus no implications should be drawn from our descriptions if and when this case revisits us after a final judgment.

[5] Most acronyms make for bad copy, but some don't. More readers may recognize "EPA" than Environmental Protection Agency.

[6] Why the case is now proceeding here, and not Missouri, when it was first filed in Missouri is a separate story which does not need to be told now.

[7] Because we resolve the appealability question on the lack of an appealable final judgment, we need not address the question of whether the appeal from the May 11, 1995, minute order would be timely even assuming it was from a final judgment. For the moment let us say that motions for reconsideration on summary adjudication would not appear to qualify as new trial motions under rule 3 of the California Rules of Court. Of course, the very fact that Zurich seemed to believe that it could appeal from both the May 11 order and the July 21 order only confirms our observations about the importance of adhering to the one final judgment rule. If orders denying reconsideration after motions were appealable, there would be no end to piecemeal appeals.

[8] We address the problem of an order that on its face compels an insurer to pay defense costs below, where we also discuss American Motorists Ins. Co. v. Superior Court (1998) 68 Cal.App.4th 864 , 80 Cal.Rptr.2d 621 .

[9] The order filed January 1999 adopts something called "Finding B-I" from a referee's October 1998 report. Finding B-I from the October report says that if the trial court does not apply Foster-Gardner, Inc. v. National Union Fire Ins. Co., supra, 18 Cal.4th 857 , 77 Cal.Rptr.2d 107 , 959 P.2d 265 , then the court should "adopt the June 17, 1998, Report except as hereinafter modified as to the fees of the in-house counsel." The June 17, 1998 report in turn recommends that a certain motion to compel the payment of defense costs as a result of the May 11, 1995 order (one of the two minute orders which is the subject of this case) "be granted and that defendant be ordered to reimburse plaintiff for the expenses incurred...." The January 28, 1999 order also recites that the trial court "does not have jurisdiction to apply the Foster decision to the May, 1995 order," so presumably the condition set forth in B-I of the referee's October 1998 report has been fulfilled. Finally, the March 12, 1999, notice of appeal states on its face that Zurich "believes its original appeal" (i.e., G018280) will "resolve" the order it is now appealing from (G025002), but it is appealing "out of an abundance of caution."

Whew! Leaving aside the ludicrously convoluted and cross-referenced nature of the January 28, 1999 orderone must look at no less than three different documents to figure out what it meansit nevertheless appears that it provides for the payment of fees. Whether it is sufficiently clear to be enforcible by contempt is another matter.

[10] We have already denied Zurich's belated motion (made after our supplemental briefing request) to consolidate G025002 with G018280 so as to address the merits in GO 18280 now. The appealability of the order in G025002 is not before us and the filing of an appeal from a nonappealable summary adjudication should not be rewarded with a reserved place in line at the appellate court.

[11] Though it bears reiterating here that the rule is one of the allocation of appellate resources rather than the structure of appellate law. The world did not end when U.S. Financial v. Sullivan (1974) 37 Cal.App.3d 5, 11-12 , 112 Cal.Rptr. 18 , pioneered the idea of saving a defective appeal by treating it as a writ petition.

[12] A few courts have mentioned the parties' own readiness to so stipulate in their discussions. (E.g., U.S. Financial, supra, 37 Cal. App.3d at p. 12 , 112 Cal.Rptr. 18 ; Clovis Ready Mix, supra, 25 Cal.App.3d at p. 281 , 101 Cal.Rptr. 820 .)

[13] If truth be told, appellate courts sometimes treat defective appeals as writ petitions simply because the merits were easily determined and written up before anyone noticed that the appeal was defective. Whatever else one might say about that practice, it certainly isn't going to happen in cases where reaching the merits necessarily requires considerable work.

[14] For example, there is no reason that the record in the present matter, GO 18280, cannot be deemed to be all or part of the record in G025002.

[15] As of mid-September 1999, no published Missouri or New York decision has mentioned Foster-Gardner.

[*] Retired Presiding Justice of the Court of Appeal, Fourth Appellate District, Division Three, assigned by the Chief Justice pursuant to article VI, section 6 of the California Constitution.