This is a diversity case in which, on cross motions for summary judgment, the district court dismissed a claim asserted against an insurance company under a builder's all-risk property damage policy. The claimant, a construction contractor, sought to hold the insurance company liable for the cost of replacing ceiling insulation ruined by water that had condensed within the insulation cavity after a subcontractor failed to install a vapor barrier properly.

Two affirmative defenses were set forth in the pleadings. The first rested on a policy provision that excluded coverage for loss or damage caused by "faulty workmanship." The second rested on an exclusion for "[d]ampness or dryness of atmosphere; extremes or changes in temperature." The insurance company's summary judgment motion invoked both of these affirmative defenses.

The plaintiff conceded, for purposes of analysis, that the damage in question had been caused by faulty workmanship, but the plaintiff relied on an express exception to the faulty workmanship exclusion. The exception had the effect of reinstating coverage for loss or damage "ensuing" from an insured peril, notwithstanding that no coverage was provided for the cost of correcting the faulty workmanship itself.

Applying the decision of this court in Farmers Chem. Ass'n v. Maryland Casualty Co., 421 F.2d 319 (6th Cir.1970), the district court accepted the plaintiff's argument on the ensuing loss exception and held that the claim was not barred by the faulty workmanship exclusion. The court upheld the insurance company's second affirmative defense, however, and dismissed the lawsuit on the ground that the loss had been placed outside the coverage of the policy, unambiguously, by the dampness of atmosphere exclusion.

The plaintiff contractor has appealed, and the defendant insurance company has taken a protective cross-appeal. Upon de novo review we conclude that the insurance company was not entitled to judgment on either of its affirmative defenses. The order dismissing the lawsuit will therefore be reversed.

* The plaintiff, Blaine Construction Corporation, was named as an additional insured under a $150,000,000 policy written by the defendant, Insurance Company of North America. The INA policy covered, among other things,

"The Insured Interest in:

(a) Real and Personal Property owned by the Insured and, improvements and betterments in buildings not owned by the Insured;

(b) Real and Personal Property of others in the Insured's Care, Custody or Control, and the Insured's liability imposed by law or assumed by written contract prior to loss."

Under the caption "PERILS INSURED AGAINST" the policy provided that

"This policy insures against ALL RISKS OF DIRECT PHYSICAL LOSS OR DAMAGE to property insured including general average, salvage and all other charges and shipments covered hereunder except as excluded."

Under the caption "PERILS EXCLUDED" the policy provided that

"This policy does not insure loss or damage caused directly or indirectly by any Peril excluded. Such loss or damage is excluded whether contributed to, in whole or in part, by any excluded Peril."

The policy then enumerated 20 excluded perils, including the two in question here:

"10. Errors in design, errors in processing, faulty workmanship or faulty materials, unless loss or damage from an insured Peril ensues and then only for such ensuing loss or damage.

* * *

12. Dampness or dryness of atmosphere; extremes or changes in temperature."

Although water seeping up from the ground and precipitation falling on personal property in the open were listed as excluded perils, there was no general exclusion for water damage.

Blaine Construction Corporation had become an additional insured under the policy after entering into a contract to construct a metal warehouse building for a utility company in Huron, South Dakota. The construction contract called for installation of a total of 12 inches of insulation at the roof of the building, and Blaine subcontracted the insulation and insulation support system work to Crown Metal Buildings, Inc., d/b/a High Rider Systems. We must assume, for present purposes, that the manner in which High Rider installed the insulation and insulation support system was not in conformity with the contract.

The contract called for the use of insulation batts with vapor barriers laminated on one side. These batts were to underlie thicker, unfaced, insulation batts. Both courses of insulation were to go between structural steel framing members, called "purlins," below the roof line.

The vapor barriers protruded beyond the batts to which they were laminated, forming "edge tabs" two inches in width. Designed to overlap one another, the edge tabs were supposed to be taped or sealed to form a continuous vapor barrier capable of preventing moisture migration.

Instead of sealing the edge tabs to form a continuous vapor barrier, High Rider simply tucked the tabs up against the sides of the purlins. Moisture could thus migrate above the vapor barrier, and any condensation would be trapped in the cavities between the purlins.

That seems to be precisely what happened in the winter of 1994-95, at a time when construction activities were still going on within the building. Consulting engineers hired by INA in May of 1995, after the loss had been reported, found that "[w]hen the vapor barrier was cut a copious amount of water flowed from the opening...." It appeared, according to the engineers, that copious amounts of water had been trapped in the ceiling cavities across the entire structure.

The owner of the building told INA's engineers that the roof system was substantially completed before the winter began. The building's concrete floor was poured in January and February of 1995, and heaters fired with natural gas were used to keep the building warm during those months. Interior venting of the exhaust from the heaters increased the relative humidity within the building. Thawing of the ground and construction materials and the curing of the concrete also contributed to the humidity.

Although the humidity and the resulting condensation were greater than they would have been otherwise, INA's engineers viewed this factor as a secondary one. Their report to INA concluded with an opinion couched in these terms:

"[M]ost if not all of the condensation within the insulation cavity area could have been prevented with the proper installation of the vapor barriers edge tabs and sealing of the insulation ends over the beams and at cross-bracing positions. * * * If proper insulation procedures had been carried out, condensation, if it did occur, would have formed on the underside of the vapor barrier in the interior of the building. Thus, no damage to the insulation would have resulted.

In conclusion, if the vapor barrier on the insulation had been sealed in a manner as specified and [as] is customary to the trade, the condensation and subsequent accumulation of water in the insulation blankets would not have occurred. The construction activities provided a high relative humidity level which only became a factor due to the lack of a continuous vapor barrier."

The building owner directed Blaine to remove and replace all of the water-soaked insulation. Blaine did so--allegedly at a cost of about $315,000--and then brought the present lawsuit against INA in federal district court after a claim submitted under the policy had been denied.

Blaine is a Tennessee corporation with its principal place of business in Knoxville, Tennessee. INA is a citizen of Pennsylvania. The parties are in agreement that the district court's diversity jurisdiction was properly invoked and that the governing substantive law is Tennessee's.

Some months after INA filed its answer asserting the affirmative defenses described above, Blaine moved for partial summary judgment thereon. With respect to the faulty workmanship exclusion, Blaine's motion contended, among other things, that "[e]ven if the failure to seal the vapor barrier constituted defective work, the introduction of moisture into the insulation was an 'ensuing loss' outside the scope of the first exclusion relied upon by INA." With respect to the dampness or dryness of atmosphere exclusion, the motion contended, inter al., that "[a]ir within a building space is not 'atmosphere' as that term is reasonably understood, and therefore, the second exclusion relied upon by INA is not applicable in this case."

INA filed a response and cross-motion for summary judgment in which it argued that the policy exclusions barred coverage unambiguously and had to be applied according to their ordinary meaning. Blaine then filed a reply brief arguing, among other things, that INA was advancing "a strained and uncommon use of the term 'atmosphere;' " that "[a]t the very least," caselaw cited in the brief validated "the reasonableness of Blaine's interpretation of the 'dampness of atmosphere' exclusion;" and that "[i]f the insurance contract is subject to more than one reasonable interpretation, the policy must be construed most favorably to the insured."

After the final brief was filed, the court heard oral argument on the summary judgment motions. In addressing the "dampness of atmosphere" issue at the hearing, Blaine's counsel contended first that Blaine's was "the only reasonable interpretation of the words...." As he had done in his reply brief, however, Blaine's lawyer went on to argue that, at the very least, "we have a reasonable interpretation of the words, and where we've proven a reasonable interpretation of the words we're entitled to coverage...." He did not use the word "ambiguity," but the clear implication of his argument--as opposing counsel unquestionably understood--was that even if the dampness of atmosphere exclusion was not clearly limited to atmospheric conditions in the layman's sense of the term, the exclusion was, at the very least, ambiguous.

Counsel for INA responded that "[t]he cases indicate that they're required to point to an ambiguity." With what may strike the reader as a touch of hyperbole, INA's counsel then said this:

"What they've given the Court here is a strained interpretation. They haven't pled an ambiguity; they haven't proven an ambiguity; they haven't indicated any evidence of an ambiguity. What they want the Court to do is to read one into the policy."

The rejoinder offered by Blaine's lawyer to the suggestion that he wanted the court to read a non-existent ambiguity into the policy was one that may strike the reader as more than hyperbole:

"[W]e've not argued that there is an ambiguity, your Honor. Quite often [sic] we don't think the policy is ambiguous. We think the policy clearly says what we say it says."

Our dissenting colleague reads this as a retraction of Blaine's fallback argument that the dampness of atmosphere exclusion was, at the very least, ambiguous. In its brief on appeal, however, INA has not argued that such a retraction occurred--and our de novo review of the record leaves us unpersuaded that the lawyer's reflexive remark was intended as a retraction of the fallback position previously advanced.

Consider the peroration of Blaine's oral argument, which begins, in the transcript, a few lines below the supposed retraction:

"If we're going to start calling ... water that condenses on the underside of the metal building panel and falls into the insulation dampness of atmosphere, essentially what INA is arguing is that no water damage is covered by this policy, and that would be very easy for INA to say. If INA wanted to say, no damage by water [is covered] by the policy, they just have to say that; but that's not what the policy says; and if anybody is trying to make a strained or contrived interpretation of the language, it's INA by trying to read the word atmosphere, dampness of atmosphere to exclude any kind of damage that occurs from water.

We think if the Court looks at the cases that have been cited, the Court can agree that the weight of authority is really in favor of Blaine but the logic of those cases support[s] Blaine's position in this case, and we request the Court to grant our motion for summary judgment."

"Blaine's position" on this issue, as reflected elsewhere in the oral argument transcript and in the reply brief, clearly had two parts. The first part was that "the only" reasonable interpretation of the exclusion was the interpretation offered by Blaine. The second part was that--at the very least--Blaine's was "a" reasonable interpretation. "[T]hey've chosen the word ['atmosphere']," Blaine's counsel argued at page 8 of the transcript, "and they have to live by any reasonable interpretation...." (Emphasis supplied.) "[T]he fact that two or three courts that ... have interpreted that word all agree with our interpretation," he continued, "is prima facie evidence that we've got a reasonable interpretation of that language." (Emphasis supplied.)

"So the bottom line," counsel told the district court, "... is, without even going to the cases, if we just read the words, and the law is very clear in interpreting insurance contracts, you read the words and if it's a reasonable interpretation that's offered by the insured, then there should be coverage; and in this case I think we've got the only reasonable interpretation of the words, but, certainly, we have a reasonable interpretation of the words, and where we've proven a reasonable interpretation of the words, we're entitled to the coverage, and we request the Court enter an Order explaining how the policy will be applied with regard to these facts...." (Emphasis supplied.) This, we take it, was the position reaffirmed by Blaine at the end of its oral argument.

The district court, as we have said, did not accept Blaine's position: "The Court finds no ambiguity in the policy use of the word 'atmosphere,' and concludes that it refers to the interior of [the utility company's] building as well as its exterior." Summary judgment was entered for INA solely on this basis.

"It is elementary in insurance law that a claimant under an insurance policy has the initial burden of proving that he comes within the terms of the policy.... Conversely, the insurer carries the burden if it claims that one of the policy exclusions applies to the claimant and prevents recovery." Farmers Bank & Trust Co. of Winchester v. Transamerica Ins. Co., 674 F.2d 548, 550 (6th Cir.), cert. denied, 459 U.S. 943, 103 S. Ct. 257, 74 L. Ed. 2d 200 (1982) (Tennessee law). Under Tennessee law, "[i]t is well settled that exceptions, exclusions and limitations in insurance policies must be construed against the insurance company and in favor of the insured." Allstate Ins. Co. v. Watts, 811 S.W.2d 883, 886 (Tenn.1991). Of course, an unambiguous exclusionary clause precludes coverage, and the courts should give the words in exclusionary clauses their ordinary meaning. Beef N' Bird of America, Inc. v. Continental Casualty Co., 803 S.W.2d 234, 237 (Tenn.Ct.App.1990).

The foregoing principles inform our analysis of both of the affirmative defenses asserted by INA. We turn first to the first such defense--the defense that coverage is precluded by the policy exclusion for faulty workmanship.

* The faulty workmanship exclusion, as we have seen, contains an exception--an "exclusion[ ] from the exclusion[ ]," to borrow a phrase from Judge Friendly1--for "ensuing loss or damage" that ensues "from an insured Peril...." The question facing us here is whether the INA policy's faulty workmanship exclusion, with its ensuing loss exception, unambiguously precludes coverage for damage ensuing from the chain of events we have described.

In Farmers Chem. Ass'n v. Maryland Casualty Co., 421 F.2d 319 (6th Cir.1970)--a diversity case governed, like the case now before us, by Tennessee law--a panel of this court was presented with a closely analogous question. The insurance policy at issue there was a Maryland Casualty all-risk business interruption policy covering an ammonia plant. Operations at the plant had to be suspended after a gas pipe buckled and developed leaks. The failure of the pipe was caused by faulty workmanship. (The inside of the pipe was supposed to have been lined with insulation in order to shield the steel casing from the heat of gases passing through the pipe under pressure and at high temperatures; because a contractor misunderstood the applicable specifications, however, the insulation had been placed outside the pipe instead of inside, thereby trapping the heat of the gases in precisely the wrong place.)

The Maryland Casualty policy contained a faulty workmanship exclusion (an exclusion for "[e]rror, omission or deficiency in design, specifications, workmanship, or materials") that incorporated an ensuing loss exception. Id. at 321. The exception cancelled the cancellation of coverage where "fire or other accidents otherwise recoverable hereunder ensues and then only for such ensuing loss or damage." Id. The insured party acknowledged that if remedial work to correct the contractor's error had been undertaken before the pipe failed, and if the plant had been shut down in connection with such work, the business interruption would not have been covered by the insurance. Where the interruption resulted from the actual failure of the pipe, however, the insured contended that the shutdown was covered by virtue of the ensuing loss exception. Maryland Casualty, on the other hand, contended that the exception could not apply unless some further loss or damage ensued subsequent to the failure of the pipe.

Faced with these conflicting interpretations of the ensuing loss exception, the Farmers Chemical panel concluded that "the language here involved is at best ambiguous." Id. That being so, and "in line with well-established rules of construction" (footnote omitted), the panel held that "the ambiguity must be resolved in favor of the insured...." Id. A judgment entered by the district court in favor of the insured was affirmed.

The parallel to the case at bar, it seems to us, is a very close one. Here too a contractor failed to follow applicable specifications for insulation installation--neglecting, in this instance, to seal the vapor barrier edge tabs. Here too significant property damage ensued--in this instance, water damage to the insulating material. Here too the ensuing damage was otherwise insured--in this instance, water (except for groundwater and, under limited circumstances, rainwater) was an "insured Peril." Here too the insured party acknowledges that the cost of correcting the workman's mistake--in this instance, the cost of sealing the edge tabs and creating a continuous vapor barrier--would have come within the faulty workmanship exclusion. Here too the insurance company contends that the damage for which coverage is sought lies outside the exception to the faulty workmanship exclusion, the original damage not having been followed by any further accident--or, as INA puts it here, "an ensuing loss, to be covered, must be the result of a new, separate and independent peril from the peril that is excluded, rather than a loss that follows naturally and ordinarily from an excluded peril." And here too the district court--which considered itself bound by Farmers Chemical--resolved the issue in favor of the insured.

INA argues on appeal that the district court erred in following Farmers Chemical, the earlier case being factually distinguishable. The claimed distinctions are these: (1) the claimant in Farmers Chemical was the owner, not the prime contractor; (2) the exclusionary provision in Farmers Chemical did not refer to loss or damage caused "directly or indirectly" by an excluded peril, and it did not say--as the INA policy does, rather awkwardly--that loss or damage excluded "is excluded whether contributed to, in whole or in part, by any excluded Peril;" and (3) what was damaged in Farmers Chemical was something other than the material that had been installed incorrectly.

These are distinctions without a difference. (1) As a named insured under the INA policy, Blaine's position was in no way inferior to that of the owner of the building. (2) The differences in the wording of the exclusionary provision are immaterial; it is not the breadth of the exclusion about which we are concerned, but the breadth of the exception to the exclusion. (3) What was damaged here was not the vapor barrier edge tabs--the items that should have been sealed and weren't--but the adjacent material. We fail to see what difference it would have made, moreover, if it had been the vapor barrier itself that was damaged.

Finally, INA asks us to reconsider Farmers Chemical. This panel cannot do that, however, absent an indication by the Tennessee courts that they would have decided Farmers Chemical differently.2 Farmers Chemical clearly constituted the law of the circuit at the time the INA policy was issued, and the rule that precludes us from overruling a prior published decision of another panel thus serves to protect the presumed expectations of the contracting parties at the time the contract was made. The district court did not err in rejecting INA's first affirmative defense.

We turn now to the question whether INA has sustained its burden of showing that the damage to the insulation was caused or contributed to by the peril described in the policy as "[d]ampness or dryness of atmosphere; extremes or changes in temperature." If the quoted words necessarily encompass the high relative humidity produced within the building by the use of gas-fired heaters and other construction activities, and if such humidity caused or contributed to Blaine's loss within the meaning of the verbs and adverbs used in the exclusionary cause, the judgment rendered in favor of INA must be affirmed. If not--if the exclusionary clause contains ambiguities that render uncertain its applicability to the particular facts presented here--the judgment must be reversed.

INA contends that although the word "atmosphere" often refers to the external atmosphere surrounding the earth, the term is also used, not infrequently, to include the air inside a building or other confined space. This is certainly true--although it is also true that "[t]he ordinary, popular meaning of the phrase 'the atmosphere' connotes the external atmosphere which surrounds the earth...." United States Fidelity & Guar. Co. v. Wilkin Insulation Co., 144 Ill. 2d 64, 161 Ill. Dec. 280, 578 N.E.2d 926, 933 (1991). Cf. Continental Casualty Co. v. Rapid-American Corp., 80 N.Y.2d 640, 593 N.Y.S.2d 966, 609 N.E.2d 506, 513 (1993) ("differing [dictionary] definitions point toward ambiguity in the meaning of 'atmosphere' "), and Lumbermens Mut. Casualty Co. v. S-W Indus., Inc., 39 F.3d 1324, 1336 (6th Cir.1994) (fumes confined to an indoor work area are not "discharged into the 'atmosphere,' as that word is ordinarily understood").

The insurance policies at issue in the cases just cited were liability policies that excluded coverage for injury arising out of the discharge of pollutants into "the atmosphere," as well as discharges "into or upon land ... or any watercourse or body of water." The insurance policy at issue here, in contrast, is not a liability policy, does not speak of dampness of "the" atmosphere, and does not contain the references to land and sea found in the pollution causes. These are significant differences, and the cited cases are obviously not dispositive of the issue presented in the case at bar. That said, however, we believe it is fair to read these cases as lending at least some support to Blaine's argument that ordinary men and women would be likely to understand a reference to "[d]ampness or dryness of atmosphere; extremes or changes in temperature" as a reference to weather conditions, and not as a reference to artificially enhanced humidity inside a building.

If INA had wanted to exclude dampness or dryness generally, the layman might think, there would have been no need to use the word "atmosphere" at all. The insurance company could simply have described the excluded peril as "dampness or dryness." By adding a word that would not have been necessary if the company had not intended to refer to the "atmosphere" in the most commonly used sense, and by speaking in the next breath of "extremes or changes in temperature," INA ran the risk, it seems to us, of being taken to mean that it was simply talking about weather conditions--the extreme heat and humidity of Florida, e.g., or the sub-zero temperatures of Alaska.

We do not have the guidance of much directly relevant caselaw on this point. The Appellate Division of the Supreme Court of New York, however, has issued an opinion that is quite instructive here. The case--Purpura v. Continental Casualty Co., 143 A.D.2d 741, 533 N.Y.S.2d 302 (N.Y.App.Div.1988)--involved a delicatessen and catering business that was insured under a policy containing an exclusion for loss caused by "[c]ontamination, dampness of atmosphere, change of temperature, corrosion or rust." Id. at 302. When the business premises were left without power during Hurricane Gloria, $12,000 worth of perishable foods went bad for lack of refrigeration. The insurance company denied coverage on the strength of the quoted exclusion, among other things. The insured contested the denial of coverage, claiming "that the 'change of temperature' exclusion applied purely to atmospheric conditions or, at the very least, was ambiguous...." Id.

The trial court sided with the insurance company on this issue, but the appellate court took the opposite view. Although the Appellate Division ultimately allowed the insurance company to avoid coverage on the strength of an exclusion for losses resulting from "[w]indstorm," the court went out of its way to express disagreement with the trial court's interpretation of the exclusion for "[c]ontamination, dampness of atmosphere, change of temperature, corrosion or rust:"

"We find the 'change of temperature' exclusion contained in the policy at bar to be ambiguous since, read in context, and according it the meaning ' "which would be given it by the average man" ' ... it may reasonably refer to changes in the weather ... which was not the cause of the loss suffered at bar." Id. at 303 (emphasis supplied) (citations omitted).

Mutatis mutandis, it seems to us, while the INA exclusionary clause may not necessarily refer to the weather alone, it can reasonably be read that way in light of the meaning that the average man would be likely to ascribe to it. Purpura strongly suggests that the exclusion should be found to be ambiguous in its application to the loss suffered here.

The strongest precedent to which INA points as support for its claim that no reasonable person could read the clause as referring just to the weather is Judge Friendly's opinion in Aetna Casualty and Surety Co. v. Yates, 344 F.2d 939 (5th Cir.1965). But the exclusionary clause at issue in Yates, and the facts to which the court applied that clause, differ significantly from the exclusion and the factual situation with which we are concerned here.

Yates involved a homeowners policy that contained an exclusion for the following:

"Loss caused by inherent vice, wear and tear, deterioration; rust, rot, mould or other fungi; dampness of atmosphere, extremes of temperature; contamination; vermin, termites, moths or other insects." Id. at 940-41.

The court described the nature of the plaintiffs' loss thus:

"[T]he plaintiffs discovered that the joists, sills and subflooring of their home were almost completely rotted away. The cause, as the evidence of both parties showed, was that the 'crawl space' under the house was inadequately supplied with vents. Contact between air trapped in the crawl space and the subfloors and sills, which had been chilled by air conditioning, produced condensation of moisture and consequent rotting." Id. at 940.

The exclusionary clause was "disheartening" for the plaintiffs, as Judge Friendly put it, because, in his words,

"Plaintiffs' loss could be said to be 'caused' by the defective construction of the house, arguably an 'inherent vice'; it was a 'deterioration,' although perhaps not 'caused' by deterioration; it surely was caused by 'rot'; the rot almost most [sic] certainly had been caused by 'fungi'; and 'dampness of atmosphere' had produced the condition in which the fungi could grow and do their work." Id. at 941.

The Yates exclusionary clause contained an exception for "ensuing loss caused by ... water damage ... provided such losses would otherwise be covered...." Id. But the court held that the plaintiffs' loss did not rise to the dignity of "water damage," within the meaning of the policy, so judgment was rendered in favor of the insurance company.

In the case at bar, although the exclusionary clause at issue here has no ensuing loss exception for water damage, INA insists that the damage to the insulation was not water damage. We disagree; it seems to us that water damage is precisely what it was that led the building owner to require Blaine to replace the insulation. As INA's own experts attested, after all, the insulation was so "saturated" with water that when the vapor barrier was pierced, "water flowed from the opening" in a quantity described as "copious." The immediate cause of the damage, obviously, was what the engineers referred to as "an accumulation of water," not an accumulation of humidity.

Blaine argues that because it was water by which the insulation was damaged, and not moist air, the damage bears a closer resemblance to water damage caused by a leaking roof (damage that presumably would be covered by the INA policy) than to damage--mildew or rot, for example--caused by "dampness of atmosphere." INA counters with a reference to the "directly or indirectly" language in the exclusionary clause and a further quotation from Yates--a passage where Judge Friendly rejected the notion that rot and water damage were in some sense separable events:

"We do not think that a single phenomenon that is clearly an excluded risk under the policy was meant to become compensable because in a philosophical sense it can also be classified as water damage; it would not be easy to find a case of rot or dampness of atmosphere not equally subject to that label and the exclusions would become practically meaningless.3 " Id.

In the case at bar, of course, there is no contention that the insulation had rotted. Here the damage consisted of the direct intrusion of water, and if we were to accept INA's reading of the exclusionary clause as the only reasonable one, an interesting question would be presented as to whether the damage was caused "indirectly," within the meaning of the exclusionary clause, by dampness of atmosphere. Blaine's answer, we surmise, would invoke the leaking roof analogy: If workmen had left open seams in the roof, and if rain (which would not have fallen but for humidity in the atmosphere) had seeped through and damaged the insulation, the loss would clearly have been covered--in this circuit, at least--notwithstanding the faulty workmanship exclusion. That being so, it would seem passing strange to exclude coverage where exactly the same type of damage results from seams left open in a vapor barrier as opposed to a roof.

This is a question that has not been thoroughly ventilated in the briefs, and we need not decide it here. Assuming, arguendo, that the causation question would be decided in favor of INA, we must ask whether there is anything in the logic of Judge Friendly's opinion in Yates that compels us to rethink our conclusion that more than one meaning can reasonably be attributed to the phrase "dampness of atmosphere." We conclude that there is not.

INA tells us that "while the word 'atmosphere' was not directly at issue in Yates, the court clearly recognized that it referred to the space inside the building...." The Yates court certainly assumed this, at least, but the assumption was not determinative of the outcome--the end result would obviously have been the same if the court had found the phrase "dampness of atmosphere" ambiguous--and we have no reason to suppose that the validity of the assumption was even argued in Yates. It was undisputed, after all, that the subflooring, joists and sills "were almost completely rotted away," and the policy contained an express exclusion for loss caused by "rot." The only real question in Yates was whether the ensuing loss exception applied; it could not have made the slightest difference to anybody whether "dampness of atmosphere" was ambiguous or not. In the case at bar, on the other hand, it makes a huge difference.

The question is a close one. We reject out of hand Blaine's contention that its reading of the exclusion is the only reasonable one, and we readily acknowledge that INA's reading is reasonable. We cannot say that Blaine's is unreasonable, however--and because "the insurer must establish that the exclusion applies in the particular case and that it is subject to no other reasonable interpretation," Purpura, 533 N.Y.S.2d at 303, we conclude that Blaine is entitled to coverage.

The judgment entered by the district court is REVERSED, and the case is REMANDED with instructions to grant Blaine's motion for partial summary judgment.

BOGGS, Circuit Judge, dissenting.

I disagree that we should consider this policy language to be ambiguous. I also believe that even if we treat it as ambiguous, Blaine's interpretation is not a reasonable one. Therefore, I dissent.

* For some reason, Blaine conceded at oral argument in the district court that the policy was unambiguous and did not attempt to argue otherwise. INA started the discussion with the following statement: "The cases indicate that they're required to point to an ambiguity. What they've given the Court here is a strained interpretation. They haven't pled an ambiguity; they haven't proven an ambiguity; they haven't indicated any evidence of an ambiguity." (Transcript of oral argument at 11). Blaine's response to this was as follows: "[W]e've not argued that there is an ambiguity, your Honor. Quite often we don't think the policy is ambiguous. We think the policy clearly says what we say it says." Id. at 15.

The district court relied on this concession in its opinion. The majority concludes that this reliance was improper. If Blaine wanted to abandon its advantage, however, why should the district court have stopped it? This is not an issue like subject-matter jurisdiction, which we either have or do not have, regardless of concessions or agreement by the parties. A plaintiff must come into court and specify the relief it wishes to receive. When a plaintiff, through counsel, does not ask for all that it is entitled to, it is not the duty of the court to second guess counsel. To be sure, the district court could have refused to accept the parties' agreement on this matter, but I do not think that we should hold, as we do today, that the district court must do so. To put this another way, Blaine is raising on appeal an issue that it did not raise before the district court, and we should not "reverse" the district court for failing to make a decision that it had no reason to make. Although we often state that we may affirm on grounds different than those relied upon by the district court (an extension of the concept of harmless error), I do not think we can reverse on grounds not presented below.

This point is ultimately irrelevant to me, however, because I do not believe that the policy is ambiguous. A policy is ambiguous only when it supports more than one reasonable interpretation, see Smith v. Shelby Ins. Co. of Shelby Ins. Group, 936 S.W.2d 261, 263 (Tenn.App.1996), and I do not believe that Blaine's interpretation is reasonable.

* This case illustrates the peril of lawyers citing dictionaries. Dictionaries are designed to be comprehensive, but the meaning of words is dependent on context, and individual definitions in a dictionary do not always offer context. To take an extreme analogy to Blaine's technique here, let us assume that I had an insurance policy that excluded coverage for any injury caused by a dog. One day, after being attacked by an irascible pooch, I decide that it would be nice to have the insurance company pay my medical bills. But what of the troublesome exclusion in the policy? No worry. I turn to my dictionary and note that "dog" is defined as an andiron. AMERICAN HERITAGE DICTIONARY 414 (2d ed.1982). Eureka! Being a good lawyer, I search for more support, and find it in the OED. See 4 OXFORD ENGLISH DICTIONARY 922 (2d ed.1989) (referred to below as OED). I can now confidently proclaim that there is (to use Blaine's term) a "consistent and uniform treatment" of the term "dog," which clearly means "andiron." I can even cite several cases in which injuries are caused by andirons. See, e.g., People v. DiMarcantonio, 117 A.D.2d 612, 498 N.Y.S.2d 160, 161 (N.Y.App.Div.1986); Vaughn v. State, 52 Ala.App. 377, 292 So. 2d 671, 672 (1974), rev'd, 293 Ala. 365, 304 So. 2d 6 (Ala.1974); State v. Silva, 153 Me. 89, 134 A.2d 628, 632 (Me.1957), overruled by State v. Brewer, 505 A.2d 774 (Me.1985).

Admittedly, Blaine's argument is not quite as unreasonable as the hypothetical one I offer here. Indeed, my two brethren think that Blaine's interpretation is reasonable, and I would not question their sagacity. However, Blaine's selective quotation of reference sources, purportedly to offer a "consistent and uniform treatment," is no less illogical than in the andiron example. This attempt to ignore context (indeed, to deny it) is unreasonable. The fact that at least one favorable definition of a word exists should not end an inquiry as to the meaning of that term in a contract. This is true whether we are defining a dog or a damp atmosphere.

So what does "dampness of atmosphere" mean?

First, we should consider what Blaine must demonstrate in order to win this appeal. The majority has overlooked something fundamental in this regard. To prevail, Blaine must do more than show us that "dampness of atmosphere" can mean outdoor humidity. It must show us that "dampness of atmosphere" cannot mean indoor humidity. The damage done in this case is covered by the insurance policy only if the latter is a reasonable interpretation. Put another way, if Blaine can reasonably contend that "dampness of atmosphere" means only outdoor humidity, it wins. If all it can do is to reasonably contend that "dampness of atmosphere" also means outdoor humidity, it cannot win.

Consider the dog/andiron example. Anyone making that argument would deserve to be laughed out of court, but not because "dog" cannot mean "andiron." Even though that interpretation is probably not contextually "reasonable," this tells us nothing about what the policy does mean. Being bitten by a neighborhood mutt is excluded from the hypothetical policy because any reasonable interpretation of the word "dog" in the policy must include members of the species Canis familiaris. Furthermore, if I am bitten by a mutt, I cannot prevail by saying that among the things that are called "dog" are Dandie Dinmont Terriers, a breed of which the mutt was not a member. Nor can I reasonably say that since "dog" is also used to mean a male dog, in distinction to a bitch, see 4 OED 921, def. 2, that I am still covered by the policy so long as the dog that damages me is a female (unless of course the policy gives us a contextual basis to so interpret the word "dog"). Rather, I must show that it is reasonable to interpret the insurance exclusion so as to exclude mutts or bitches. A mutt may not be a Dandie Dinmont, and a bitch may be female, but it is unreasonable to say that either of these is not a dog. Similarly, indoor humidity may not be outdoor humidity, but it is unreasonable to say that it is not "dampness of atmosphere."

My last statement requires proof. Luckily, there is ample evidence not just that no reasonable definition of "dampness of atmosphere" may exclude indoor humidity, but furthermore that indoor humidity is the most common meaning of the term. More exactly, the most common use of "atmosphere," when used in conjunction with a term like "dampness," is to refer to the air in an particular place, whether outdoors or indoors. The distinction Blaine draws is thus not supported by typical usage.

My earlier attack regarding the use of dictionaries in cases such as this should not be taken to mean that I find dictionaries useless in interpreting insurance contracts. Rather, I am frustrated with their mis-use, through selective quotation. What follows is my attempt to undo this insult, and thus to do right by the dictionary.

The most definitive dictionary of the English language is the Oxford English Dictionary. Not only does the OED attempt to list every word used in the English language in the last thousand years, it attempts to provide every definition that each word has had. More importantly, for our purposes, it also offers quotations, using each definition in context.

"Atmosphere" has two relevant definitions, among the many listed in order of their entry into the language:

1b. The mass of aeriform fluid surrounding the earth; the whole body of terrestrial air [first listed use 1677].

5. The air in any particular place, esp. as affected in its condition by heat, cold, purifying or contaminating influences, etc.; = AIR sb. 4 [first listed use 1767].

1 OED 750 (2d ed.). Definition 5, which is clearly the better of the two, given the context of the insurance policy and its reference to "dampness," draws no distinction between outdoor and indoor air. It is thus unreasonable to define "dampness of atmosphere" so as to exclude indoor air.

In determining the meaning of "atmosphere" in the context of dampness, I turn again to the OED for several helpful quotations. What follows are all of the places in the OED where some form of the word "damp" is used with some form of the word "atmosphere" in a quotation, reproduced as they appear in the OED [year of the citation in bold]:

1887 H.E.F. Garnsey tr. A. de Bary's Compar. Morphol. & Biol. Fungi iii. 89 As long as the Fungus remains shut up in the damp atmosphere no amount of shaking will cause it to puff. [12 OED 796; puff, v., 1d.]

1882 Garden 11 A.K. Marsh. 168/2 A damp, cool atmosphere, with little artificial heat, causes the flowers to spot. [16 OED 324; spot, v., I.3.]

c1900 Buck's Handbk. Med. Sci. III. 265 (Cent.Dict.Suppl.) Damp soil serves to keep the super-ambient atmosphere damp. [17 OED 206; super-ambient, prefix, I.2.]

1934 E. Little Mod. Rhythmic Drumming (rev. ed.) 26 No outfit is complete without at least one tomtom. The 'tuneable' models are the best, because any dampness in the atmosphere can be counteracted by the use of the tensioning handles. [18 OED 668; tunable, a., 2.]1855 Orr's Circ. Sci., Inorg. Nat. 215 The resulting lime..sets rapidly in a damp atmosphere, and even under water. [18 OED 948; under, prep., I.4e.]

1766 Smollett Trav. xi. I. 174, I have always found a cold and damp atmosphere the most unfavorable of any to my constitution. [19 OED 15; unfavorable, a., 1b.]

All of these quotations use "atmosphere" in the sense of Definition 5 rather than Definition 1b. None of them imply that there is any per se distinction between indoor and outdoor air in the use of the term "atmosphere," as though the word could mean the latter but not the former. Four of the six quotations can be read as discussing indoor air, and at least two definitely are.

Its definitiveness notwithstanding, the OED is not the last place to look in understanding how a term is used. Indeed, modern databases offer a good opportunity to do exhaustive and comprehensive research in this regard. What follows is the result of a search of all of the American cases contained in Westlaw (fifty-four in total) in which some form of the word "damp" appears within five words of some form of the word "atmosphere," and there is some hint as to the meaning of the phrase.

In fifty-one of the fifty-four cases, "atmosphere" is used in the sense of OED Definition 5, rather than Definition 1b. (In the other three, this is less clear, but there are important distinguishing factors.) That is, in almost every case, the "atmosphere" referred to is the air in a particular place, sometimes outdoors and sometimes indoors. In twenty-six of the cases, the damp atmosphere is definitely indoors.1 In another eight cases, it is unclear where the damp atmosphere is, but it is clearly the air surrounding a particular place.2 In seventeen cases, the damp atmosphere is outside, but it is localized and is being discussed because of its effect on activities that only occur outdoors.3 Two of the cases cited above involve insurance language (and the same defendant), and both use Definition 5.4

In the three cases that remain, the reference to damp atmosphere relates to the weather in general. In two of them, however, the reference is to "the atmosphere," a distinction that will be discussed more below. Reames v. Jones Dry Goods Co., 99 Mo.App. 396, 73 S.W. 935, 938 (Mo.Ct.App.1903) (Ellison, J., concurring); Simmons v. Prudential Ins. Co., 269 A.D. 1048, 58 N.Y.S.2d 578, 579 (N.Y.App.Div.1945). In the third, a ceiling was damaged when it was exposed to the "damp outdoor atmosphere" because windows were kept open. Cackowski v. Jack A. Halprin, Inc., 133 Conn. 631, 53 A.2d 649, 651 (Conn.1947) (emphasis added).

These cases thus show that it is unreasonable to define "dampness of atmosphere" in a way that excludes Definition 5.5 The court's opinion has given me no reason to conclude otherwise. With respect to the court's argument at page 20, I reiterate that the conflict is not between two mutually exclusive definitions. It is not that Blaine says "atmosphere" means only the outside air and INA says that it means only the inside air around a particular place. Instead, the conflict is that INA says that atmosphere can mean air that is either outside or inside, whereas Blaine can only prevail if a reasonable interpretation is to exclude indoor air altogether. On that dichotomy, the authorities cited support the view that there is no ambiguity.

Contrary to the argument made by Blaine below, cited at page 9, INA's interpretation of its own language does not exclude all water damage, nor does INA wish to do so. If someone played a firehose on the walls of the building, or if water spurting from a broken hydrant caused the damage, even by rendering the walls "damp," INA's language would not exclude coverage. However, had INA chosen the language suggested by the court, at pages 15--16, such damage would be excluded. Thus, if INA wanted to express its carefully considered wishes, the language it chose does the job nicely.

Finally, I would like to explain further why the remainder of Blaine's argument is unconvincing to me, and why INA's arguments are more availing. Blaine cites pollution cases in which discharges into the atmosphere have been limited to those into outside air. See, e.g., Lumbermen's Mut. Cas. Co. v. S-W Indus. Inc., 39 F.3d 1324, 1336 (6th Cir.1994) (holding that policy was meant to protect against liability for damages to neighbors and governmental environmental agencies, not indoor injuries from toxic materials). Indeed, pollution cases involving general-liability policies with exclusions for discharges "into the atmosphere" use Definition 1b fairly uniformly. See, e.g., Gamble Farm Inn, Inc. v. Selective Ins. Co., 440 Pa.Super. 501, 656 A.2d 142, 145 (Pa.Super.1995); Continental Cas. Co. v. Rapid-American Corp., 80 N.Y.2d 640, 593 N.Y.S.2d 966, 609 N.E.2d 506, 512 (N.Y.1993); Owens-Corning Fiberglas Corp. v. Allstate Ins. Co., 74 Ohio Misc.2d 144, 660 N.E.2d 746, 754 (Ohio Com.Pl.1993); United States Fidelity & Guar. Co. v. Wilkin Insulation Co., 144 Ill. 2d 64, 161 Ill. Dec. 280, 578 N.E.2d 926, 933 (Ill.1991); see also Essex Ins. Co. v. Avondale Mills, Inc., 639 So. 2d 1339, 1341-42 (Ala.1994); Bd. of Regents of University of Minnesota v. Royal Ins. Co. of America, 517 N.W.2d 888, 892-93 (Minn.1994).

But, as INA notes, invoking context, this case does not involve a general-liability policy, in which coverage is given for damage inflicted upon others. Rather, it is an all-risk policy covering damage suffered by the insured. This makes an "internal" interpretation of "atmosphere" more appropriate. Furthermore, the exclusion here is not for discharges into the atmosphere, which implies outdoors, but rather for dampness of atmosphere, which is neutral at best. In a case closer to this, Aetna Cas. & Sur. Co. v. Yates, 344 F.2d 939 (5th Cir.1965), the Fifth Circuit considered the exclusions in an all-risk policy. The policyholder in the case suffered damage when a combination of inadequate ventilation and air conditioning caused condensation and rotting in a crawl space. See id. at 940. An exclusion for "[l]oss caused by ... rot, mould or other fungi; dampness of atmosphere, extremes of temperature; contamination; vermin, termites, moths or other insects" was held to apply, even when read in conjunction with an un-exclusion for water damage, and even with ambiguities construed against the insurer. Id. at 940-41. The court relied on both the "rot" and "dampness of atmosphere" provisions, and took as a given (without saying so) that "atmosphere" included indoor air. Id. at 941; see also Clark v. Barnwell, 53 U.S. (12 How.) 272, 283, 13 L. Ed. 985 (1851) (referring to "dampness of the atmosphere" in a ship's hold).

To the extent that Blaine's argument from the pollution cases proves anything, it proves that context is essential. It is no accident that all of Blaine's cases involve third-party liability for discharges of pollution. It is also no accident that all of Blaine's cases involve insurance policies that refer to "the atmosphere" (implying the singularity of the OED's definition 1b rather than the more multiplicitous definition 5) and none of which refer to "dampness." If the policy in this case had excluded damage from "pollution released into the atmosphere," I would agree with the cases that Blaine cites, and vote to hold that "the atmosphere" could reasonably be interpreted as meaning only the air outside the building in question. But the policy does not say that. It speaks of a different context, one of dampness, and Blaine and the majority are extending an interpretation from the pollution context, where it makes sense and has been neatly contained until today, to the dampness context, where it makes no sense.

The majority is propounding an unreasonable interpretation of the "dampness of atmosphere" exclusion, and I must dissent from its reversal of the judgment in favor of INA.

I concur with the majority's rejection of INA's cross-appeal.