Between 1996 and 2000, the City of Burlington (the "City") held two all-risk policies that insured a power generator with Indemnity Insurance Company of North America ("Indemnity"). During the period of coverage, approximately thirty-three welds on the generator's boiler developed leaks as a result of intrinsic defects. Indemnity disclaimed coverage for the defective and failed welds and the City brought suit in the United States District Court for the District of Vermont. The district court (Murtha, C.J.) held that the defective and failed welds were not covered because the City's policies excluded losses caused by a "latent defect." We conclude that the correct disposition of this case depends on two undecided questions of Vermont law: (1) whether under Vermont law all-risk policies cover only losses that are externally caused, and (2) if not, whether under Vermont law a paragraph of the City's policies that excluded from coverage losses due to "latent defect" and various other listed internal causes applied to the failed welds. Because Vermont courts have not yet ruled on the issues presented by this case and because these issues involve important public policy considerations for Vermont, we certify our question to the Vermont Supreme Court, and ask for its guidance.

In 1982, the City contracted with Zurn Industries, Inc. to design, engineer and construct a wood-fired steam electricity generator. The boiler portion of the generator included an economizer, which consisted of metal tubes welded together. These "shop welds" were performed at the manufacturing facility prior to delivery.

In 1987 and 1988, the City discovered two isolated leaks in shop welds on the lower section of the economizer. The record before us indicates that this number of leaks was not enough to cause concern. No more leaks were found until April 1995. Between 1995 and August 1999, however, the City discovered and repaired thirty-four additional leaks on the lower economizer.

The City hired David N. French to study the economizer and determine the cause of the leaks. French performed radiographic/metallographic analysis on two weld samples that had been removed from the boilers. In March 1999, he reported that the original welds were substandard, finding "a lack of full penetration which means the weld does not comply with [American Society of Mechanical Engineers] requirements." Later that same month, the City contracted with Power Specialist Associates, Inc. (PSA) to perform ultrasonic thickness testing. Like French, PSA concluded that "it appeared that the failure was caused by a lack of root weld penetration, which would cause a relatively weak area in the tube."

In October 1999, the City decided to remove and replace all existing shop welds on the lower economizer, including those that, to date, had shown no signs of leaking. Its contractor, Bremco, concluded that "[c]areful inspection of the removed welds revealed incomplete penetration at every joint." In the spring of 2000, the City contracted with Vermont Nondestructive Testing for non-invasive radiographic inspection of the center and upper portions of the economizer; these investigations also indicated pervasive weld defects.

Between August 25, 1995 and August 25, 2001, the boiler unit was insured with Indemnity. Thirty-three of the thirty-four leaks occurred during this period. The two three-year policies covering those years (the "Policies") were in all relevant respects identical. Both were "all-risk" policies, which is to say that they insured against all risks unless explicitly excluded, and both covered "risks of direct physical loss or damage to the property insured." The Policies also contained an "Excluded Causes of Loss" section, which contained, inter alia, the following coverage limitations:

This Policy does not insure loss, damage or expense caused directly or indirectly by any of the following. Such loss, damage, or expense is excluded regardless of any other cause or event that contributes concurrently or in any sequence to the loss.

....

9. Corrosion, rust, decay, depletion, deterioration, erosion, evaporation, inherent vice, latent defect, leakage, loss of weight, marring or scratching, bulging, gradual cracking, shrinkage, wear and tear, wet or dry rot or any quality in property which causes it to damage or destroy itself.

....

14. Cost of making good;

A. Error, omission or deficiency in design, plans, specification engineering or surveying;

B. Faulty or defective workmanship, materials and supplies; This exclusion shall not apply to the mechanical breakdown of:

(1) any boiler, pressure vessel, refrigerating system or any piping and accessory equipment or;

(2) any electrical or mechanical machine or apparatus used for the generation, transmission or utilization of mechanical or electrical power; which has been installed, fully tested and contractually accepted by the Named Insured and which is being or has been operated at the insured location, in the capacity for which it was designed, as part of the Named Insured's normal production process or processes.

On October 15, 1999, the City mailed Indemnity a notice of claim letter, which specified as the loss the faulty welds on the lower section of the economizer. Indemnity replied on January 7, 2000 that its review of the matter found coverage to be "in question" due to the Policies' exclusions, including those for "inherent vice," "latent defect," and "any quality in property which causes it to damage or destroy itself." The letter expressly reserved all rights and defenses.

On April 13, 2000, the City made an additional claim for the middle and upper sections of the economizer. On April 17, Indemnity replied that its investigation was winding to a close and a decision on the City's claims would be reached within thirty days. Indemnity had not yet made a claim determination on May 24, 2000, when this lawsuit was filed in the United States District Court for the District of Vermont.

The City's complaint sought damages from five separate insurance companies, each of which had insured the generator at one time or another between its installation and the initiation of the suit. The defendants each moved and the City cross-moved for summary judgment. The district court analyzed the policies separately and concluded, for different reasons in each case, that none covered the City's claimed losses. See City of Burlington v. Hartford Steam Boiler Inspection and Ins. Co., 190 F. Supp. 2d 663 (D.Vt.2002). The City appeals only the ruling on the Indemnity Policies.

The court held that coverage was precluded by the Policies' "inherent vice" and "latent defect" exclusions.1 It focused on the term "latent defect," which the Policies did not define and which Vermont courts have yet to consider in the context of an all-risk insurance policy. The district court observed that, though some courts have deemed the term ambiguous and therefore construed it in favor of the insured, a majority of courts outside of Vermont have found the term to be unambiguous. Id. at 688-89 (citing cases). It concluded that, given the "strong trend of authority," the Vermont Supreme Court would not find the term ambiguous within the context of Indemnity's all-risk policy and would read that term to mean "a defect that is hidden, or which could not have been discovered by any known or customary test or examination." Id. at 688-89. The court held that the flaws in the welds were not discoverable upon known and customary inspection, were therefore latent defects, and consequently were excluded from coverage. Id. 689-90.2

The City subsequently moved to alter or amend the judgment and for reconsideration, on the grounds that reconsideration was appropriate since Indemnity had not argued the "latent defect" exclusion in moving for summary judgment and because that term was in fact ambiguous. In a decision dated May 10, 2002, the trial court granted the motion for reconsideration and, without further explanation, declined to alter its ruling.

The district court ruled largely on the basis of the Policies' latent-defect exclusion. On appeal, Indemnity argues that the district court got the meaning of "latent defect" right — that the term refers to flaws that are hidden and cannot be discovered through known or customary tests. The City argues for a narrower interpretation of the term as it appears in the Policies, under which "latent defect" refers only to hidden flaws that are undiscoverable by all but the most searching examination, and maintains that the flaws in the welds do not meet this definition. The City also contends that the district court failed to make all possible inferences in its favor, as it was required to do on Indemnity's motion for summary judgment, when the court concluded that the defects in the welds could not have been discovered by any known or customary tests.

While the definition of "latent defect" in an all-risk insurance policy is an important question, this case does not necessarily turn on it. The parties raise a number of other arguments for or against the decision below and we may affirm a grant of summary judgment for different reasons than those relied upon by the district court. Abdu-Brisson v. Delta Air Lines, Inc., 239 F.3d 456, 466 (2d Cir.), cert. denied, 534 U.S. 993, 122 S. Ct. 460, 151 L. Ed. 2d 378 (2001). Moreover, the term "latent defect" appears in an exclusionary clause that itself raises broader questions as to the meaning of the Policies.

In addition to the question of whether the faulty or failed welds were subject to the "latent defect" exclusion, which was the ground on which district court granted Indemnity's motion for summary judgment and which we consider to raise sufficiently important questions of Vermont law to warrant certification, Indemnity continues to press two other, independent arguments for a holding that there was no coverage. If either of Indemnity's proposed resolutions were correct under Vermont law and fully resolved the suit before us, we would, of course, decide the case on that basis, rather than troubling the Vermont Supreme Court. Our analysis must therefore begin with these arguments.

First, by their own terms, the Policies protected only "against risks of direct physical loss or damage to the property insured." Indemnity argued before the district court and continues to maintain on appeal that the defective and failed welds did not qualify as a "direct physical loss or damage" and were not, therefore, covered. If this is correct, then the defective and failed welds were not covered regardless of whether they qualified as "latent defects."

Vermont courts have not yet ruled on how the term "direct physical loss or damage" should be interpreted in an insurance contract. Other courts, however, have held that "[t]he language `physical loss or damage' strongly implies that there was an initial satisfactory state that was changed... into an unsatisfactory state." Trinity Industries, Inc. v. Ins. Co. of N. America, 916 F.2d 267, 270-71 (5th Cir.1990) (Louisiana law); see also Whitaker v. Nationwide Mut. Fire Ins. Co., 115 F. Supp. 2d 612, 616 (E.D.Va.1999) (Virginia law); Wolstein v. Yorkshire Ins. Co., 97 Wash. App. 201, 985 P.2d 400, 408 (1999); North American Shipbuilding, Inc. v. Southern Marine & Aviation Underwriting, Inc., 930 S.W.2d 829, 833 (Tex.Ct.App.1996). Under this definition, while the failure of a defective part qualifies as direct physical loss or damage, the defect itself, assuming the item has not yet failed, does not. We are aware of no cases adopting a contrary interpretation of the term "direct physical loss or damage" and consider it extremely likely that Vermont courts would interpret the term as it appears in the Policies in conformity with these other courts.3

Applying this definition to the undisputed facts of this case, we conclude that coverage under the Policies extended only to the thirty-three leaks that occurred during the Policies' coverage period and not to the costs of repairing the defective welds that had not yet failed. Indemnity argues that even the welds that leaked are not covered, since the City provided no evidence that the leaks caused additional damage. But clearly the occurrence of a leak that must be repaired is itself the sort of discrete event that qualifies as a loss. We therefore believe that the thirty-three leaks fall within the scope of "direct physical loss or damage" and, unless otherwise excluded, are covered by the Policies.

The City makes two arguments that even the defective welds that have not yet failed also qualify as a "direct physical loss or damage." We find neither convincing. First, the City contends that it is undisputed that the welds did not meet the standards of the American Society of Mechanical Engineers (ASME), which Vermont has adopted in its regulations governing boiler construction. Vt. Stat. Ann. tit. 21, § 242. The City cites various decisions holding that a sufficient injury may exist "only because of a government regulation, and not because of direct damage to the insured property," General Mills, Inc. v. Gold Medal Ins. Co., 622 N.W.2d 147, 152 (Minn.Ct.App.2001), and asks us to conclude that this regulatory noncompliance qualifies as a direct physical loss or damage. But the City does not claim, and presents no evidence suggesting, that compliance with the ASME Code is required by Vermont.4 And, even under the cases that the City cites, unless such compliance is mandatory, noncompliance doesn't suffice to constitute a loss.5

Second, the City maintains that the defects in the nonleaking welds are due to the negligent actions of third parties (namely, the manufacturer, Zurn Industries, Inc.), that such actions of third parties are classic perils covered by all-risk policies, and that the date of loss should be calculated on a "manifestation theory" — a loss occurs when it is first discovered — according to which the dates of the City's "losses" fall in the period covered by Indemnity. This analysis, however, is precluded by paragraph 14 of the Policies' "Excluded Causes of Loss" section, which states that the cost of making good "[f]aulty or defective workmanship, materials and supplies" is not covered. While that paragraph also exempts from the paragraph 14 exclusion the "mechanical breakdown ... of any boiler," this exemption cannot apply to the welds that did not fail during the coverage period.

We therefore agree with Indemnity that the limitation to "direct physical loss or damage" precludes recovery for the welds that did not fail, but conclude that it does not exclude from coverage those thirty-three welds that did fail. Accordingly, Indemnity's first independent argument for summary judgment does not suffice to resolve the suit.

Indemnity's second independent argument for granting its motion for summary judgment is that the "faulty workmanship" exclusion also applied to the welds that failed. The City responds that the failed welds qualify as "mechanical breakdown[s]" of the boiler and are therefore exempt from the exclusion, i.e., within the scope of coverage. The parties argue at length as to whether a failed weld counts as a "mechanical breakdown." We conclude that the term as it appears in the Policies is ambiguous. Given Vermont's rule that all ambiguities in an insurance policy are to be resolved in favor of the insured, City of Barre v. New Hampshire Ins. Co., 136 Vt. 484, 486, 396 A.2d 121, 122 (1978), we read the term "mechanical breakdown" to include the failure of a weld. Consequently, the failed welds are not subject to the Policies' "faulty workmanship" exclusion and are, unless otherwise excluded, covered.

Because Indemnity's other proposed grounds for summary judgment are insufficient to decide the case, we are left with the question of whether paragraph 9 of the Policies' "Excluded Causes of Loss" section ("Paragraph 9") places the failed welds outside the scope of coverage.

Paragraph 9 excludes losses caused by, inter alia, "latent defect." The district court read the "latent defect" exclusion to entail that none of the defective welds were covered. In reaching this conclusion, it reasoned that Vermont courts would be most likely to read the term to mean any "defect that is hidden, or which could not have been discovered by any known or customary test or examination," and that the flaws in the welds were not so discoverable. 190 F.Supp.2d at 688, 689-90.6

The meaning of "latent defect" as an exclusion to an all-risk policy is an open question under Vermont law and, therefore, is one that might well be certified to the Vermont Supreme Court in an appropriate case. We doubt, however, that this case turns on the definition of that term alone. When interpreting all-risk policies, we are required to view "exclusionary terms as a whole, since the terms are both abstract and far from mutually exclusive." 6 Couch on Insurance § 153:87 (3d ed.1997); see also City of Barre, 136 Vt. at 486, 396 A.2d at 122 ("Resort must be had to construing the contract of insurance in its entirety."). And Paragraph 9 itself excludes not only losses caused by "latent defect," but those caused by "corrosion," "decay," "deterioration," "inherent vice," "or any quality in property which causes it to damage and destroy itself." Any one of these terms could be read to describe the failure of the faulty welds. As a result, rather than resting the outcome of the case on the legal meaning of one or another of them, we believe it to be incumbent on us to inquire into the meaning of the paragraph as a whole and to ask whether the faults in the welds fall within the ambit of that paragraph's exclusions when these are taken all together.7

Read as a whole, Paragraph 9 may well be an attempt to specify what many jurisdictions at one time held to be implicit in all-risk policies, namely, that coverage extended only to losses due to external causes. To understand why this is so, a detour into the history of all risk policies is needed.

All-risk policies like the City's policies with Indemnity cover all risks except those that are specifically excluded. It has long been recognized, however, that "all-risk" does not mean all-loss. British & Foreign Marine Ins. Co. v. Gaunt, [1921] 2 A.C. 41, 57 (Eng.); Mellon v. Federal Ins. Co., 14 F.2d 997, 1002-03 (S.D.N.Y.1926) (A.Hand, J.). A loss that is inevitable is not a risk. This conceptual truth has been expressed in terms of a requirement that to be covered, a loss must be fortuitous: "Damage, in other words, if it is to be covered by policies such as these, must be due to some fortuitous circumstance or casualty." Gaunt, 2 A.C. at 47. "[E]ven in an `all risk' policy, there must be a fortuitous event — a casualty — to give rise to any liability for insurance." Mellon, 14 F.2d at 1004.

The term "fortuity" names a problem — which is to identify which losses count as risks and are therefore covered — but does not solve it. It remains to be determined just when a loss qualifies as "fortuitous." Two aspects of the answer are relatively easy.

First, it is beyond question that losses resulting from the intentional misconduct or fraud of the insured do not qualify as fortuitous as that term is used to define the limits of all-risk insurance coverage. Gaunt, 2 A.C. at 57 ("Nor is it [an insured] loss which the assured brings about by his own act, for then he has not merely exposed the goods to chance of injury, he has injured them himself.") (quoted in Mellon, 14 F.2d at 1002). See also Avis v. Hartford Fire Ins. Co., 283 N.C. 142, 195 S.E.2d 545, 548 (1973) (defining a "fortuitous event" as one "occurring by chance without evident causal need or relation or without deliberate intention" (quoting Webster's Third New International Dictionary 895 (1961))); Aetna Ins. Co. v. Sachs, 186 F. Supp. 105, 108 (E.D.Mo.1960) (holding that the gross negligence of a dog owner in allowing his incontinent poodle to play in the vicinity of an expensive piece of furniture precluded recovery under an all-risk policy).

Second, losses due to normal wear and tear are not fortuitous. Entropy is not a matter of chance, but is our destiny. Gaunt, 2 A.C. at 46 ("[S]uch damage as is inevitable from ordinary wear and tear and inevitable depreciation is not within the policies."); Contractors Realty Co. v. Ins. Co. of N. Am., 469 F. Supp. 1287, 1293 (S.D.N.Y.1979) ("[N]ormal wear and tear... is not an insurable risk, but a certainty.").

Early judicial interpretations of all-risk policies, however, also held to be non-fortuitous — and therefore outside the scope of coverage — those losses that resulted from an inherent quality or defect in the thing insured. Gaunt, 2 A.C. at 57 ("Accordingly the expression does not cover inherent vice or mere wear and tear or British capture. It covers a risk, not a certainty; it is something, which happens to the subject-matter from without, not the natural behaviour of that subject-matter, being what it is, in the circumstances under which it is carried.") (quoted in Mellon, 14 F.2d at 1002); Greene v. Cheetham, 293 F.2d 933, 936-37 (2d Cir.1961) ("But the `all-risk' event so covered would not include an undisclosed event that existed prior to coverage, or an event caused by the consummation during the period of coverage of an indwelling fault in the goods that had existed prior to that coverage."); Glassner v. Detroit Fire & Marine Ins. Co., 23 Wis. 2d 532, 127 N.W.2d 761, 764 (1964) ("An `all-risk' policy... is not a promise to pay for loss or damage which is almost certain to happen because of the nature and inherent qualities of the property insured."). Under this definition, a loss is fortuitous only if it has an external cause. This external-cause requirement has been held to be a conceptual truth: "The addition of the phrase `external cause' to the `all risks' clause constitutes no real limitation on the scope of the latter. If the loss did not result from inherent defect, ordinary wear and tear, or intentional misconduct, its cause was necessarily external." Goodman v. Fireman's Fund Ins. Co., 600 F.2d 1040, 1042 (4th Cir.1979).

While it is fairly obvious why fortuity excludes both intentional or fraudulent losses and normal wear and tear, the limitation to externally caused losses is not so self-evident. An historical explanation may lie in the fact that all-risk policies originated as maritime policies that typically insured against "all perils of the sea," see Stephen A. Cozen & Richard C. Bennett, Fortuity: The Unnamed Exclusion, 20 Forum 222, 228 (1985), a phrase that might have been read to suggest that only externally caused losses were insured in such policies. An alternative reason was given by Judge Augustus Hand in Mellon. He there stated that he found the requirement of an external cause not in the concept of fortuity, but in that of insurance. "It must always be borne in mind that the policies are of insurance and not of warranty of soundness, and for that reason no liability arises under the perils clauses for damage [caused by latent defects]." 14 F.2d at 1002 (emphasis in original).8

Whatever the reason for the early reading of an implicit external-cause requirement in all-risk policies, many courts have come to reject that approach. They have typically done so based on a different, and subjective, interpretation of "fortuity," one that considers the question to be not whether the loss was inevitable, but whether it was one that the parties knew of or expected. The Third Circuit's decision in Compagnie des Bauxites de Guinée v. Insurance Co. of North America, 724 F.2d 369 (3d Cir.1983), illustrates this, today's more common, reading.9

The issue in Compagnie des Bauxites was whether an all-risk policy covered losses that resulted from the structural failure of the insured's building and equipment. The district court granted summary judgment to the insurer on the grounds that "an insurer is not liable for losses resulting from inherent vice, defect or infirmity in the subject matter insured, for this type of loss is not caused by chance or an extraneous event." Compagnie Des Bauxites De Guinée v. Ins. Co. of N. Am., 554 F. Supp. 1080, 1084 (D.Pa. 1983). The Third Circuit reversed, relying on the definition of "fortuitous" in the First Restatement of Contracts:

A fortuitous event ... is an event which so far as the parties to the contract are aware, is dependent on chance. It may be beyond the power of any human being to bring the event to pass; it may be within the control of third persons; it may even be a past event, such as the loss of a vessel, provided that the fact is unknown to the parties.

724 F.2d at 372 (quoting and adding emphasis to First Restatement of Contracts § 291, cmt. a (1932)). The court of appeals held that, while in hindsight the structural defects might appear inevitable, it had to credit the insured's statements that it had no knowledge of the design defects and that the loss was therefore fortuitous. Id. at 372. Because an intrinsically caused loss may be just as unexpected as an extrinsically caused one, this interpretation of "fortuity" meant that no external-cause requirement could be read implicitly into all-risk policies.

Various explanations can be given for this changed view of fortuity. One is that courts incorrectly applied the Restatement of Contract's subjective definition of "fortuity" to insurance law, where the term had a previously fixed objective meaning. Another is that courts increasingly came to adopt interpretations of insurance policies that were friendly to the insured, i.e., interpretations that broadened the scope of coverage. Finally, the expansive reading of "fortuitous" in all-risk policies could be viewed as a "penalty default." See Ian Ayres & Robert Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 Yale L.J. 87 (1989). On this account, expanded coverage to the detriment of insurers in all-risk policies is justified since such expansions give insurers, who presumably have better knowledge of insurance laws than do insureds, a powerful incentive to insert explicit language into policies, thereby informing the insureds as to the precise scope of coverage.10

We believe that this background puts Paragraph 9 in context. The paragraph lists any number of possible internal causes and concludes with the catch-all "or any quality in property which causes it to damage and destroy itself." Read in the light of the aforementioned history of judicial interpretation of all-risk policies, we think it quite possible that this is best read as an attempt to return to a world that limited coverage to externally caused losses.

Accordingly, if we did not have the option of certifying to the Vermont Supreme Court, we would be inclined to dispose of this case as follows. First, while Vermont courts have not yet provided an interpretation of the fortuity requirement in all-risk policies, we would predict that they would be more likely to adopt the contemporary, subjective reading of that requirement than to stay with the older, objective reading. That is, we would expect that all-risk coverage under Vermont law, while extending only to fortuitous losses, is not, as such, limited to losses that have external causes. Second, we would be inclined to read Paragraph 9 as opting-out of this interpretive default and excluding intrinsically caused losses from the Policies' coverage. Because it is not disputed that the cause of the leaking welds was a lack of full penetration in the welds themselves, rather than any external cause, we would therefore tend toward the conclusion that the Policies did not cover the failed welds.

Such a conclusion, however, involves significant assumptions about how Vermont law interprets all-risk insurance policies. Because these assumptions implicate important policy considerations, the determination of which lies within the province of Vermont and its courts, we deem it prudent respectfully to certify the following questions to the Vermont Supreme Court:

Under Vermont law, does an all-risk insurance policy cover unforeseen losses that are not externally caused, where the policy does not explicitly exclude such loses from coverage?

Assuming an affirmative answer to the first question, did paragraph 9 of the "Excluded Causes of Losses" provision of the Policies suffice to limit coverage to externally caused losses and hence to exclude the failed welds from coverage?

The certified questions may be reformulated or expanded to cover other pertinent questions of Vermont law that the Vermont Supreme Court finds appropriate to answer in connection with this appeal. And we welcome any guidance the Vermont Supreme Court might wish to provide with respect to any state law issues presented by this appeal. This panel retains jurisdiction to consider all questions that remain before us once the Vermont Supreme Court has either provided us with its guidance or has declined certification.

Accordingly, it is hereby ORDERED that the Clerk of this Court transmit to the Vermont Supreme Court a Certificate, as set forth below, together with a complete set of the briefs, appendix, and record filed by the parties in this Court. The parties are further ORDERED to bear equally such costs and fees, if any, as may be required by the Vermont Supreme Court.

The foregoing is hereby certified to the Vermont Supreme Court pursuant to Second Circuit Local Rule § 0.27 and Vermont Rules of Appellate Procedure Rule 14, as ordered by the United States Court of Appeals for the Second Circuit.