This case presents a novel and important question of insurance law. A manufacturer of asbestos products faces huge potential liability because of numerous lawsuits filed around the country by persons who inhaled asbestos fibers allegedly manufactured by the company. The claimed basis for liability is that the asbestos manufacturer failed to warn asbestos workers and other ultimate users of its products that asbestos was a dangerous product which, if inhaled, could cause an early death from cancer or other disease. The company had various products liability insurance policies over a twenty-year period of time. These policies were issued by five different insurance companies. Which of the insurance companies is obligated to provide a defense to the nu*1214merous lawsuits? And, assuming the manufacturer is found liable, which of the insurance companies must cover the judgment? These are the issues presented by this case. They are not easy to resolve and they have split the American insurance industry, many of whose member companies have filed amicus curiae briefs.
FACTS
Background
From 1923 until 1970, Forty-Eight Insu-lations, Inc. manufactured products which contained asbestos. Asbestos is a mineral compound which has high tensile strength and flexibility. Asbestos also withstands high temperatures. For these reasons, asbestos has many commercial uses, especially in the construction industry. Asbestos is used in home insulation, cements, paints and tile.
The problem is that tiny asbestos particles can become airborne when asbestos is mined and processed, when asbestos materials are used at a construction or other site, and when old buildings containing asbestos are demolished. When these asbestos particles become airborne, a number of them are inhaled by persons in the area. The asbestos particles are deposited in the lungs. If, over the years, enough asbestos particles are inhaled, they can cause a variety of pulmonary diseases. Medical science is not certain exactly how these diseases develop, but there is universal agreement that excessive inhalation of asbestos can and does result in disease. These asbestos-caused diseases include mesothelioma, broncheo-genic carcinoma (lung cancer), and asbestosis.1
The most common disease is asbestosis. Asbestosis occurs when fibrous lung tissue surrounds small asbestos particles in the lungs to prevent the particles from moving around or causing irritation to neighboring cells. Ordinarily, this encapsulation of the asbestos particles is a good thing. However, if too many asbestos particles are inhaled, then the encapsulation process diminishes pulmonary function and makes breathing difficult. When this occurs, the disease of asbestosis is said to be present.
Asbestosis is a progressive disease. It ordinarily takes years of breathing asbestos fibers for asbestosis to occur. And asbestosis varies greatly from person to person. Obviously, the concentration of the asbestos in the work environment is a critical factor. The more asbestos fibers a worker inhales, the more quickly a worker will contract asbestosis. Even so, there are many vagaries. The average human lung has much excess capacity and can absorb a fair amount of asbestos particles. How much varies. Many construction workers exposed to asbestos for forty years or more do not become diseased. Others, exposed to asbestos for shorter periods of time at lower concentrations, do contract asbestosis.2
The typical person who has contracted asbestosis is a worker who was employed in an industry in which many asbestos-laden *1215products were used, e. g. construction. Usually, the worker has been exposed to asbestos particles for many years-at least ten, and generally for twenty or more.
By the late 1960s so many workers had contracted asbestosis that manufacturers' began to cut back on the production of asbestos-containing products. Forty-Eight Insulations stopped using asbestos in its products in 1970. However, there remained the question of liability for the past sale of these products containing asbestos.
Suits on behalf of the injured workers were filed against the asbestos manufacturers. The workers’ theory of liability was that under Restatement Torts 2d § 402(A), asbestos was an inherently dangerous product which the manufacturers had to warn them about. The workers claimed that the manufacturers had not so warned them and that therefore the workers should recover. To the extent that it was impossible to tell which particular manufacturer was responsible, they contended that all relevant asbestos manufacturers should be jointly and severally liable.3
The above-outlined theory of liability was endorsed in the leading case of Bore1 v. Fibreboard Paper Prods., 493 F.2d 1076 (5th Cir. 1973), cert. denied, 419 U.S. 869, 95 S.Ct. 127, 42 L.Ed.2d 107 (1974). See Prosser, Law of Torts, § 52 at 315-20 (4th Ed. 1971). Borei has triggered an avalanche of law suits against the entire asbestos manufacturing industry. The typical complaint names a multitude of companies which manufactured asbestos-containing products over the years. One of these companies is Forty-Eight Insulations. As of the date of the district court opinion, Forty-Eight had been named as a defendant in 251 “asbestos suits”. By the end of 1978, Forty-Eight had over 800 such suits filed against it. By the summer of 1979, over 1,370 cases were filed.
The Instant Case
In recent years, most manufacturers have taken out insurance policies to protect them against products liability suits. Before 1955, Forty-Eight was apparently self-insured.4 Starting in that year, Forty-Eight was covered by various insurance policies issued by different companies. The Insurance Company of North America (INA) insured Forty-Eight from October 31,1955 to October 31, 1972 with six consecutive insurance policies whose coverage limits varied. Affiliated FM Insurance Company (Affiliated FM) insured Forty-Eight from October 31, 1972 to January 10, 1975. Illinois National Insurance Company (Illinois National) insured Forty-Eight from January 10, 1975 to January 12,1976. Travelers Indemnity Company of Rhode Island (Travelers) insured Forty-Eight from January 12, 1976 to November 8, 1976. Liberty Mutual Insurance Company (Liberty Mutual) has insured Forty-Eight since that date.
The various policies issued by the above companies have varied in their coverage over the years.5 However, each of the policies uniformly defined what it covered, when coverage applied and the definitions of the various terms used.6 The reason is *1216that the insurance industry uses standardized language in its general liability policies. Thus each of the policies taken out by Forty — Eight contained the same relevant language. This language was as follows:
[The insurer] will pay on behalf of the insured all sums which the insured shall be legally obligated to pay as damages because of ... bodily injury or ... property damage to which this policy applies caused by an occurrence.7 “Bodily injury” means bodily injury, sickness or disease sustained by any person which occurs during the policy period, including death at any time resulting therefrom.
“Occurrence” means an accident, including injurious exposure to conditions which results, during the policy period, in bodily injury. . . .
The above-cited policy language causes little difficulty in the ordinary case. Asbestosis, however, presents thorny problems since it does not clearly fit within the above language. A worker who contracts asbestosis and sues Forty-Eight and/or other asbestos manufacturers must establish liability and injury in order to win a judgment. If the worker succeeds in his suit and Forty-Eight is directed to pay, then it is clear that the above-cited insurance coverage is triggered.
But which insurance company has to pay the judgment? At first glance, it would appear that the insurance company which provides coverage when asbestosis is diagnosed should pay. However, asbestosis is a slowly progressive, insidious disease. As more and more asbestos particles settle in the lungs over years of exposure, the disease worsens. At some point, asbestosis interferes with gas exchange in the lungs and clearly manifests itself. It is thus arguable that a worker’s asbestosis “occurred” during the many years that the worker was breathing in tiny asbestos particles and accumulating them in his lungs.
This case principally concerns a dispute among Forty-Eight’s different insurance carriers over the years as to which carrier or carriers is liable under the above-cited policy provisions. IN A filed this diversity action in the district court, seeking a declaratory judgment on this issue.
I.
We are called upon in this case to construe uniform provisions in widely-used Comprehensive General Liability (CGL) insurance policies. We are presented with two different theories of construction.8
INA, Affiliated FM, Illinois National and Liberty Mutual, supported by several amici curiae, advocate the manifestation theory of liability. Under this theory, bodily injury in a case of asbestosis could not be deemed to have occurred until the asbestosis manifests itself. The date of manifestation is the date when the worker knew or should have known he has asbestosis, or the date that asbestosis is medically diagnosed, whichever came first. Accordingly, under the manifestation theory, those insurance *1217companies on the risk when asbestosis manifests itself must pay any resultant judgment of liability. Simply put, the manifestation theory is that no “bodily injury” took place until asbestosis became apparent.
Forty-Eight and Travelers, supported by numerous amici curiae, advocate the exposure theory. They argue that when asbestosis manifests itself has nothing to do with when “bodily injury” took place. They emphasize that the medical testimony establishes that tissue damage starts to occur shortly after the initial inhalation of asbestos fibers and that the tissue damage worsens as the victim breathes in more and more asbestos fibers. The advocates of the exposure theory characterize asbestosis as a series of continuing injuries to the body which accumulate to cause death or disability. Under this theory, asbestosis is a “continuing tort” and all insurance companies which provided coverage from the time of the worker’s initial exposure to time of the manifestation of the disease are jointly and severally liable to defend and to indemnify Forty-Eight if liability is found.
The district court adopted the exposure theory.9 The court first noted that many of the insurance companies that now advocate the manifestation theory had initially embraced the exposure theory. The court found this initial conduct instructive as to the proper meaning of the policy provisions.
The principal basis for the district court’s position, however, was the medical evidence. The medical testimony established that “each tiny deposit of scar-like tissue causes injury to a lung”. From this, the court reasoned that “each such insult-causing injury is an ‘occurrence’ for the purpose of determining which coverage applies.” 451 F.Supp. 1230, 1239 (E.D.Mich. 1978).
Finally, the district court thought that the exposure theory best served the expectations of the contracting parties. The court emphasized that an injured worker who sues Forty-Eight makes out a prima-facie case by showing injury and exposure to Forty-Eight’s products. The plaintiff in such an underlying lawsuit can recover even if he cannot pinpoint the exact time when each injury occurred. Under Borel, supra, all asbestos manufacturers to whose products a worker was exposed are jointly and severally liable for the resultant, cumulative injury. By analogy, the district court reasoned that the manufacturers’ insurance coverage should parallel their liability, and that each insurance company should be jointly and severally liable to defend and indemnify the manufacturer. Otherwise, concluded the district court, the manufacturer’s insurance coverage would be illusory.
All parties have appealed.
II.
In each case where a plaintiff sues an asbestos manufacturer, a hearing could be held to determine at what point the buildup of asbestos in the plaintiff’s lungs resulted in the body’s defenses being overwhelmed. At that point, asbestosis could truly be said to “occur”. From then on, all companies which insured the manufacturer would be treated as being “on the risk”.10
*1218The only problem with this Solomonian interpretation is that no one wants it. The principle reason is cost. If medical testimony as to asbestosis’ origin would have to be taken in each of the thousands of asbestosis cases, the cost of litigation would be prohibitive. This appears to be especially true since many of the asbestosis cases are settled before trial. In addition, it is almost impossible for a doctor to look back and testify with any precision as to when the development of asbestosis “crossed the line” and became a disease.
The only thing on which all parties agree is that there is a need for us to arrive at an administratively manageable interpretation of the insurance policies-one that can be applied with minimal need for litigation. Reaching such a beneficial result is certainly desirable, but it greatly complicates our task. In the real world, there are few Solomonian possibilities. And, as we have just seen, those that do exist are often impractical. Thus, we must turn to the arguments advanced on appeal.
A.
As noted above, the medical testimony is not in dispute. Asbestosis is a slowly progressive disease. Injury, in the sense that there is tissue damage, occurs shortly after the initial inhalation of asbestos fibers. It was because of this early tissue damage that the district court concluded that the insurance companies would be liable from the date of a worker’s initial exposure to asbestos. The district court thus equated the “bodily injury” which is covered by the insurance policies with the development of the tiny, scar-like tissue in the victim’s lungs. Because each additional inhalation of asbestos fibers results in the build-up of additional scar tissue in the lungs, the district court deemed the “bodily injury” to occur whenever asbestos fibers were inhaled.
The medical testimony strongly supports the exposure theory. The appellants-who are proponents of the manifestation theory-seek to blunt the significance of this testimony. They agree that tissue damage takes place shortly after initial inhalation of asbestos fibers. However, they contend that this is irrelevant. Appellants argue that what should matter under the contract is compensable “bodily injury”. They rely heavily on analogous insurance law cases. These cases will be discussed below.
We concede that appellants have a good argument. How one views “bodily injury” or “disease” depends on how one defines those terms. The doctors who testified openly stated that this was so. Dr. George Wright, for example, testified as follows:
Well, you get, first of all, you get into something of a matter of semantics. What is a disease? ... At what stage of quantitative abnormality does something become a disease? Its a judgment call.
Dr. Wright explained that a physician or biologist might not regard asbestosis as a disease so long as it was not diagnosable by x-ray or if the patient had no symptoms. In contrast, a histologist, who studies tissue systems, would regard asbestosis as a disease from the very beginning. Dr. Wright explained:
Now, is tissue injury a disease? You go to the histologist and he will say, “You bet your boots it is a disease”. He will find asbestos fibers, asbestos bodies, and he will say that is asbestoses], period. That is a disease. And he can do that long before the clinician can do it.
This “judgment call” is what is before us for review. As a matter of contract law, in this case, when did the disease of asbestosis occur?
Appellants urge that date of manifestation is the preferable interpretation of when bodily injury occurs. Appellants claim that this interpretation best serves the intent of the parties, which is to treat cumulative disease cases the same as ordinary accident cases. Otherwise, argue appellants, the limits and deductibles contained in the various insurance policies make no sense.11 Appellants emphasize that manifestation is the only theory that *1219reasonably allows the operation of the insurance industry’s claims machinery and administration of suits and coverage. In a nutshell, the proponents of the manifestation theory urge that we treat asbestosis the same as any other disease and that we not make any “special rules” for the cumulative disease situation which asbestosis presents.
We cannot agree. Cumulative disease cases are different from the ordinary accident or disease situation. First, the underlying theory of tort liability is that the asbestos manufacturers continually failed to warn the asbestos workers and that, as a result of this, continuous breathing of asbestos particles allowed asbestosis to progress to the point where it caused death or injury. The insurance policies before us are comprehensive general liability policies which are designed to insure the manufacturer against products liability suits. The contracting parties would expect coverage to parallel the theory of liability. Otherwise, as the district court noted, the manufacturer’s coverage becomes illusory since the manufacturer will likely be unable to secure any insurance coverage in later years when the disease manifests itself.12
Second, a question might exist as to whether the microscopic tissue injury which occurs upon initial inhalation of asbestos should be defined as the occurrence of a “disease” or “bodily injury”. However, there is universal medical agreement that the time when asbestosis manifests itself is not the time when the disease occurred. No doctor would say that asbestosis occurred when it was discovered. As Dr. Wright testified:
Suppose a man never goes to see anybody or never has a chest [x-ray] film made or anything else done until he is far advanced and the diagnosis is then made on that date. Well, nobody, I think, would ask you to believe that is when the disease began. That is when the diagnosis was made, but it was a temporal phenomenon based on things having nothing to do with tissue injury; namely, when did the man go to see a doctor so the diagnosis could be made?
Third, although there are solid arguments to support the manifestation theory, we are bound to broadly construe the insurance policies to promote coverage. This is a diversity case and we must apply Illinois and New Jersey law. We adopt the district court’s cogent discussion of this issue:
In a contract diversity action such as this, Michigan conflict of law rules require the application of the law of the place where the insurance policies were issued and countersigned, Chrysler Corp. v. Insurance Company of North America, 328 F.Supp. 445 (E.D.Mich.1971). The parties agree that Illinois law applies to the interpretation of all the policies except the Liberty Mutual policies, which are governed by New Jersey law. The law of both states holds that where the terms of an insurance contract are clear and unambiguous, they must be given their common and ordinary meaning. See Canadian Radium & Uranium Corp. v. Indemnity Insurance Company of North America, 411 Ill. 325, 104 N.E.2d 250 (1952); Boswell v. Travelers Indemnity Co., 38 N.J. Super. 599, 120 A.2d 250 (1956); Wilkinson & Son, Inc. v. Providence Washington Ins. Co., 124 N.J.Super. 466, 307 A.2d 639 (1973). In both states insurance contracts are to be liberally construed in favor of the insured and against the insurer. J. L. Simmons Co. v. Fidelity and Casualty Co., 511 F.2d 87 (6th Cir. 1975) (applying Ill. law); Last v. West American Co., 139 N.J.Super. 456, 354 A.2d 364 (N.J.1976). If there are ambiguities in the policy, or uncertainty over its inter*1220pretation, the policy is to be construed against the insurer, and in favor of the insured. Tiffany Decorating Co. v. General Accounts Fire and Life, 12 Ill.App.3d 597, 299 N.E.2d 378 (1973); Bryan Construction Co. v. Employer’s Surplus Lines Ins. Co., 60 N.J. 375, 290 A.2d 138 (N.J. 1972). See also Corbett Co. v. Ins. Co. of North America, 43 Ill.App.3d 624, 357 N.E.2d 125, 2 Ill.Dec. 148, 357 N.E.2d 125 (1976).
451 F.Supp. at 1237-38.
B.
Because of policy considerations and because of the medical testimony, we do not find controlling the analogous insurance cases urged upon us by the appellants.
First, there are the statute of limitations cases. In the past, defendants have argued that since a disease “occurs” at or shortly after exposure to a foreign substance, that is the time when the statute of limitations should start running. The problem is that such a ruling would bar relief to many plaintiffs who were unaware that they were being injured until years later when the disease manifests itself. To avoid such an unfair result, most courts 13 have adopted a manifestation rule:
“inasmuch as the injurious consequences of the exposure are the product of a period of time rather than a point of time; consequently, the afflicted employee can be held to be “injured” only when the accumulated efforts of the deleterious substance manifest themselves”. Urie v. Thompson, 337 U.S. 163, 170, 69 S.Ct. 1018, 1024, 93 L.Ed. 1282 (1949) (citation omitted)
Here, we find that the policy considerations are the opposite of those present in the statute of limitations cases. A manifestation rule would deny coverage to the insured manufacturer. Moreover, it is the injury and not its discovery that makes the manufacturer liable in the underlying tort suit. As noted above, such underlying liability should also trigger insurance coverage. We disagree with appellant’s argument that identical language used in statutes of limitations and insurance policies must necessarily be interpreted identically. Linguistic uniformity should not dictate how contracts or statutes are interpreted. Statutes of limitation are meant to protect defendants against stale claims, nor bar injured plaintiffs who have acted in good faith. Insurance contracts are meant to cover the insured.14
*1221Second, there are the worker’s compensation cases. Assume that a worker is debilitated by a progressive disease like asbestosis. Where that worker has been exposed to asbestos while working for various employers over the years, which employer is liable for the worker’s compensation coverage? Courts have generally adopted a “last employer pays” rule.15 This rule, however, is the general rule applicable in worker’s compensation cases.16 The underlying rationale for the rule is “the overriding importance of efficient administration in this area.” Travelers Ins. Co. v. Cardillo, 225 F.2d 137, 145 (2d Cir.), cert. denied, 350 U.S. 913, 76 S.Ct. 196, 100 L.Ed. 800 (1955). While such considerations are present here, they do not override the rules of contract interpretation we have already noted. Moreover, because all that is required to establish workers compensation liability is a work-related injury, administrative considerations are of greater significance in workers compensation cases than in liability insurance cases.
Third, and of most relevance, are health insurance cases. Under these cases, there is no coverage until a disease is diagnosable as such.17 Thus, in Metropolitan Life Ins. Co. v. Reynolds, 48 Ariz. 205, 60 P.2d 1070 (1936), a court found coverage for insanity which manifested itself during the policy period even though it was caused by syphilis contracted before there was insurance coverage. Similarly, Cohen v. North American Life & Casualty Co., 150 Minn. 507, 185 N.W. 939 (1921) also found coverage where the insured underwent surgery for a condition which had a cause which predated the policy period.
These cases certainly appear to provide support for the manifestation theory. We do not, however, view them as controlling here. On the contrary, we think that they support the exposure theory. The reason is that the health insurance cases rely on the same rules of construction that we think are applicable here: insurance policies must be strictly construed in favor of the injured and to promote coverage; similarly, a policy must be construed to favor the legitimate expectations of the parties.
In Metropolitan Life Ins. Co., supra, the court construed an insurance policy which provided coverage where the insured had become totally and permanently disabled, as the result of “bodily injury or disease occurring and originating after the issuance of said Policy ...” In Cohen, the court construed a similar insurance policy which provided coverage for a disability resulting from “a disease which shall originate and begin after this policy shall have been in continuous force for 30 days.”
Both cases reached the same result for the same reasons. Health insurance policies protect the insured in case of disabling sickness. This is why people buy such policies in the first place. It would be unfair to the insured-and contrary to his expectations when he bought the insurance-to allow a *1222hidden condition to defeat the coverage which he bought.18
In this case, we are faced with two possible constructions, one of which is likely to leave the manufacturer insured, the other of which leaves the manufacturer uninsured for all practical purposes. We think that Illinois and New Jersey courts would try to construe the contract language to embrace the exposure theory.
C.
We do not rely on judicial fiat. Initially, we note that Forty-Eight strenuously argues that the policy language is ambiguous. It points out that the insurance industry is divided between the manifestation and exposure theories and that some insurance companies which initially accepted coverage under the exposure theory only belatedly changed their position. It further emphasizes that words such as “bodily injury” and “occurs” are inherently ambiguous.
We think that a better view is that the contractual terms in issue here, “bodily injury” and “occurrence” are inherently ambiguous as applied to the progressive disease context before us. A cumulative, progressive disease does not fit the disease or accident situation which the policies typically cover. There is usually little dispute as to when an injury occurs when dealing with a common disease or accident. As can be seen, in this case there is considerable dispute as to when an injury from asbestosis should be deemed to occur. For this reason, we can apply the above-stated canons of construction and resolve doubts in favor of maximizing coverage.
Aside from this, however, we believe that the policy language itself is best construed along the lines of the exposure theory. We need only look at the definition of “bodily injury” in the policy. Bodily injury is defined as “bodily injury, sickness or disease ...” It is tautological that bodily injury can be “bodily injury” and is not necessarily just a “disease”. The medical evidence is uncontroverted that “bodily injury” in the form of tissue damage takes place at or shortly after the initial inhalation of asbestos fibers. Thus, it requires only a straightforward interpretation of the policy language for us to adopt the exposure theory. Indeed, for insurance purposes, courts have long defined the term “bodily injury” to mean “any localized abnormal condition of the living body.” See Appleman, Insurance Law and Practices § 355 (1965).
We do not find persuasive appellants’ contention that “bodily injury” means *1223“compensable bodily injury.” The manufacturer here paid for protection from bodily injury resulting in liability. It should make no difference when the bodily injury happens to become compensable. Put another way, we see nothing in the policy which requires that the underlying plaintiffs’ cause of action accrue within the policy period. There exists a clear distinction between when bodily injury occurs and when the bodily injury which has occurred becomes compensable.
Although the problem presented by this case is one of first impression for an appellate court, the parties were aware of the potential problems which cumulative tort cases might present throughout the contract period. The etiology of asbestosis has been known since at least the 1930’s. Although Borel was not decided until 1973, it was certainly foreseeable that asbestos manufacturers might be held liable for failing to warn workers of the danger of asbestos. “But though the application is novel, the underlying principle is ancient. Under the law of torts, a person has long been liable for the foreseeable harm caused by his own negligence. This principle applies to the manufacture of products as it does to almost every other area of human endeav- or.” Borel, supra at 1103.
The policies themselves show that the insurance industry contemplated the problem presented by cumulative trauma cases.19 Since 1962, each of the policies in question has included in its definition of “occurrence” a provision referring to continuous or repeated exposure to conditions which result in injury.20 There is strong evidence that coverage of diseases such as asbestosis was contemplated. Indeed, appellants openly admit that this is true. There is no similar provision in the policies restricting coverage to when asbestosis manifests itself. The insurance industry doubtless did not foresee the extent of the liability problem asbestosis cases would present. However, that in itself is no basis for construing the contract the way the appellants want.
In sum, we think that the exposure theory adopted by Chief District Judge John Feikens is the superior interpretation of the contract provisions.21 “Bodily injury” should be construed to include the tissue damage which takes place upon initial inhalation of asbestos. That is both a literal construction of the policy language and the construction which maximizes coverage. It is also the construction which, we think, best represents what the contracting parties intended.
The thrust of the appellants’ arguments is that we should apply a uniform manifestation rule in all cases, including this one, and not be concerned with medical intricacies. We see no evidence that this is what the parties meant.
III.
Although victorious below, Forty-Eight urges that we modify the district court. *1224opinion in three respects. We shall discuss these questions seriatim.
A.
The district court, as discussed above, adopted the exposure theory for purposes of liability and prorated liability among all of the insurance companies which were on the risk while the injured victim was breathing in asbestos. For those years that Forty-Eight did not have insurance, the district court treated the manufacturer as self-insured and responsible for a pro rata share of the cost of indemnification. Thus, if insurer A provided 3 years of coverage, insurer B an additional 3 years, and the manufacturer was uninsured for the remaining 3 years, liability would be allocated at l/z for each of the three concerns.
The district court applied an identical rule apportioning the costs of defending the underlying suits. To the extent that the manufacturer was uninsured, it has to bear its pro-rata share of the costs of defense.
Forty-Eight agrees that it must bear its share of the liability risk for those years in which it had no insurance-that is, before October of 1955. However, Forty-Eight vehemently argues that the district court was wrong in prorating the cost of defending the law suits against it.
Forty-Eight argues that the insurer’s obligation to defend is very broad-broader than its obligation to indemnify. Under the very wording of the insurance policies, the insurance companies had a duty to defend based on the allegations in the complaint, even if the allegations are groundless, false, or fraudulent ...” See Maryland Casualty Co. v. Peppers, 64 Ill.2d 187, 355 N.E.2d 24 (1976).
Forty — Eight argues further that the duty to defend is sufficiently broad so that it arises when one count of the complaint is within policy coverage and other counts are not. In many such cases, courts have not permitted apportionment of defense costs between the insurer and the insured.22 From these cases, Forty-Eight argues that so long as any insurance company had a duty to defend, it (Forty-Eight) should not be liable for any costs of defense, even if part or most of the underlying lawsuit concerned periods of time when Forty-Eight was uninsured.
We cannot agree. We fully accept the legal principles and cases on which Forty-Eight relies. What Forty-Eight ignores is the rationale on which its authorities are based. An insurer must bear the entire cost of defense when “there is no reasonable means of prorating the costs of defense between the covered and the not — covered items.” Nat’l. Steel Construction Co. v. Nat’l. Union Fire Ins. Co., 14 Wash.App. 573, 543 P.2d 642, 644 (1975). Thus, in the typical situation, suit will be brought as the result of a single accident, but only some of the damages sought will be covered under the insurance policy. In such cases, apportioning defense costs between the insured claim and the uninsured claim is very difficult. As a result, courts impose the full cost of defense on the insurer.23
These considerations do not apply where defense costs can be readily apportioned. The duty to defend arises solely under contract. An insurer contracts to pay the entire cost of defending a claim which has arisen within the policy period. The insurer has not contracted to pay defense costs for occurrences which took place outside the *1225policy period. Where the distinction can be readily made, the insured must pay its fair share for the defense of the non-covered risk.24
In this case Forty-Eight’s own exposure theory, substantially adopted by the district court, establishes that a reasonable means of proration is available. Forty-Eight has urged that indemnity costs can be allocated by the number of years that a worker inhaled asbestos fibers. By embracing the exposure theory, we have agreed. There is no reason why this same theory should not apply to defense costs. The different insurance companies will pro-rate defense costs among themselves. It is reasonable to treat Forty-Eight as an insurer for those periods of time that it had no insurance coverage.25 Were we to adopt Forty — Eight’s position on defense costs a manufacturer which had insurance coverage for only one year out of 20 would be entitled to a complete defense of all asbestos actions the same as a manufacturer which had coverage for 20 years out of 20. Neither logic nor precedent support such a result.
B.
Forty-Eight also argues that liability for asbestos-caused injuries should not be permitted for any time period unless it can be shown that its products were involved. Although the district court used the term “joint and several liability”, we read its decision as implicitly adopting this position.
In an underlying asbestosis suit, the plaintiff must show that Forty-Eight’s products injured him in order to be able a maintain a cause of action against Forty-Eight.26 Under Borel, Forty-Eight would be jointly and severally liable along with the other asbestos manufacturers. 493 F.2d at 1094-96. However, in allocating the cost of indemnification under the exposure theory, only contract law is involved. Each insurer is liable for its pro rata share. The insurer’s liability is not “joint and several”, it is individual and proportionate. Accordingly, where an insurer can show that no exposure to asbestos manufactured by its insured took place during certain years, then that insurer cannot be liable for those years. The reason is simple: no bodily injury resulting from Forty-Eight’s products, took place during the years in question. The same thing would be true if an insurer could show that a worker used an effective respirator during certain years. Again, no “bodily injury” would have taken place.
We think that this is implicit in the district court’s opinion and only state it here for clarity. However, the burden of disclaiming coverage will be on the insurer for the year or years in question.27
*1226C.
Finally, Forty-Eight urges that we modify the district court’s proration formula. Forty-Eight argues that the formula should reflect not only the injury caused by asbestos during the policy years when the worker breathed in asbestos fibers, but also the injury which developed afterward. Thus, if three insurance companies were on the risk for a 9 year period of exposure, under Forty-Eight’s formula the first insurer would be on the risk for the first 3 years plus the remaining 6 while the disease progressed. The second insurer would be on the risk for 3 years plus the following 3 years. The final insurer would be on the risk only for the final 3 years. Thus liability would be apportioned 9/is for the first insurer, 6/is for the second, and 3/is for the third. Under the district court’s formula, of course, each insurer would be liable for % of the costs.
We cannot accept this interesting formula because of the explicit language of the policies which define “occurrence” as “an
Because of these unique circumstances, we think it appropriate to presumptively view each manufacturer as being on the risk for each of the years in which a worker was exposed to asbestosis. If an insurance company can show that a certain manufacturer’s products were not or could not have been involved for certain years, it will be absolved from paying its pro rata share for those years. Given the impossibility in most cases of ascertaining which company provided asbestos products in different years, we think that this is the fairest way to apportion liability. Thus, we simply reverse the ordinary burden of proof and place it on the insurer. We are keenly aware of the need to apply a straightforward formula and not one which will lead to additional litigation.
accident, including continuous or repeated exposure to conditions”, (emphasis added) Moreover, this proposed formula would place different obligations upon insurers who are in the same position. The extent of the insurer’s liability would turn on the accident of when an insurer provided coverage. We reject this anomalous result.
Conclusion
Because the insurance industry did not fully appreciate the extent of potential liability for asbestos-caused injuries, we have this impossible problem before us. There is no truly satisfactory solution we can make. Each theory has its flaws and anomalies.28 Judge Feikens struggled with this case, as we have, and thought that the exposure theory was preferable. Although his opinion has been challenged by virtually every party and amicus on one ground or another, we think that he reached the right result for the right reasons.29
The judgment of the district court is affirmed.
*1227APPENDIX A
APPENDIX B
*1228
. Throughout this opinion, we shall be discussing the liability situation as it relates to asbestosis. The reason, is that mesothelioma and lung cancer present very different considerations. Mesothelioma is cancer of the mesothe-lial cells which line the chest wall and surround the organs of the chest cavity. This cancer is associated with the inhalation of asbestos fibers. Mesothelioma generally does not occur until 20 years or more after there has been excessive inhalation of asbestos. Mesothelio-ma is easily discovered and diagnosed shortly after it develops. Unfortunately, there is no satisfactory treatment for the disease, and the victim almost always dies within several years of the tumor’s initial development.
Lung cancer’s development is similar to that of mesothelioma. However, the correlation between the inhalation of asbestos and the development of lung cancer is not established. Rather, it appears that inhalation of asbestos accelerates the development of lung cancer in persons who smoke.
. Two leading articles in the medical literature on asbestosis are Selikoff, Bader, Bader, Churg, and Hammond, “Asbestosis and Neoplasis”, 42 Am.J.Med. 487 (1967); Selikoff, Churg, and Hammond, “The Occurrence of Asbestosis among Insulation Workers”, 132 Ann. New York Acad. Sci. 139 (1965). A good discussion of asbestosis is contained in Borel v. Fibreboard Paper Products, 493 F.2d 1076, 1083-85 (5th Cir. 1973), cert. denied, 419 U.S. 869, 95 S.Ct. 127, 42 L.Ed.2d 107 (1974).
. Joint and several industry-wide liability is itself an interesting question. In asbestosis cases, plaintiff-workers in underlying lawsuits have been able to show that various asbestos manufacturers actually applied the asbestos products to which the workers were exposed. However, in some jurisdictions, joint and several industry-wide liability can be established even if a plaintiff cannot show that a particular company’s or companies’ products were actually involved. The mere likelihood that a company provided a defective product to the plaintiff may be sufficient to trigger liability. See Sindell v. Abbott Labs, Inc., 26 Cal.3d 588, 163 Cal.Rptr. 132, 607 P.2d 924 (1980); Abel v. Eli Lilly & Co., 94 Mich.App. 59, 289 N.W.2d 20 (1979); Hall v. DuPont, 345 F.Supp. 353, 370-80 (E.D.N.Y.1972); See also Restatement 2d, Torts § 433(B) (1972); Comment, DES and a Proposed Theory of Enterprise Liability, 46 Fordham L.Rev. 963 (1978).
. Forty-Eight thought that it did have insurance coverage before 1955. However, any such policies were lost and/or destroyed and Forty-Eight was unable to prove it had any coverage before that date.
. See the breakdown which is reproduced as ' appendix “A” to this opinion.
. Coverage limits ranged from $300,000 (per accident)/$100,000 (per person) from 1955 to 1962; $500,000 (per accident) from 1962 to *12161973; $400,000 (per accident) from 1973 to 1975; $300,000 (per accident) in 1976; $1,000,-000 (per accident) from 1976 to the present. In addition, starting in 1976, Liberty Mutual’s policy contained a $100,000 per person deductible for asbestosis cases. Since most asbestosis cases have been settling for less than $100,000, this means that, as a practical matter, Forty-Eight is uninsured for asbestosis occurring after 1976. A chart containing each of the policies and their coverage is reproduced as appendix “B” to this opinion.
. In 1966, the provision providing coverage for liability “caused by an occurrence” was changed from the previous provision which provided coverage for liability “caused by accident”. This case, however, centers on interpretation of the term “bodily injury”. No party to this suit ascribes any significance to this 1966 change as far as this suit is concerned.
. The issues presented by this case have been the subject of varying judicial interpretation and commentary. See Porter v. American Optical, (E.D.La. No. 75-2202, Dec. 23, 1977), on appeal to the Fifth Circuit (manifestation theory adopted); American Motorists Ins. Co. v. E. R. Squibb, 95 Misc.2d 222, 406 N.Y.S.2d 658 (Sup.Ct.1978); (same); Tenneco Chemicals v. Employers Mut. Liab. Ins. Co. (S.D.N.Y. No. 76 Cir. 809, April 8, 1977) (exposure theory adopted); Comment, Liability Insurance for Insidious Diseases: Who Picks up the Tabs? — Ford.L. Rev. — (1980); Note, 26 Wayne L.Rev. 1127 (1980).
. More accurately put, the court adopted a version of the exposure theory. In its cross-appeal, Forty-Eight advocates a series of changes to the district court’s exposure theory. These issues are discussed below in part III.
. Judge Merritt’s ingenious opinion tries to find this optimal point by looking to X-ray diagnosis and establishing a 10-year point. The problem with this innovative theory is that it disregards the medical testimony that “bodily injury” takes place at or shortly after inhalation. Moreover, this theory places too much weight on the state of the art in medical diagnosis. As Judge Merritt acknowledges, X-ray diagnosis is simply one tool used to find whether a worker has asbestosis. The medical experts termed it a “crude” tool. In contrast, a lung biopsy would reveal damage shortly after initial exposure. Of course, this is not currently feasible. But what if medical advances make biopsy or a similar procedure feasible? We prefer to say that bodily injury is bodily injury from the beginning, rather than to say that bodily injury is bodily injury ten years from initial exposure because that’s when an x-ray shows it.
. See fn. 28, infra.
. In its brief, INA states that the district court’s reliance upon the Borel case “as controlling precedent for the interpretation of insurance contracts is almost embarrassing. Contractual language controls contractual interpretation. Rules of Tort do not.” INA oversimplifies the issue in its effort to discredit the district court. The insurance policies which we must interpret are meant to protect a manufacturer from products liability judgments. The legal theory used in the underlying tort suits is certainly relevant in helping a court define what was meant by the policy language.
. See United States v. Kubrick, 444 U.S. 111, 123 24, 100 S.Ct. 352, 360-61, 62 L.Ed.2d 259 (1979); Urie v. Thompson, 337 U.S. 163, 170-71, 69 S.Ct. 1018, 1024-25, 93 L.Ed. 1282 (1949); Goodman v. Mead Johnson & Co., 534 F.2d 566, 570, 574-5 (3rd Cir. 1976); cert. denied, 429 U.S. 1038, 97 S.Ct. 732, 50 L.Ed.2d 748 (1977); Roman v. A. H. Robins Co., 518 F.2d 970, 972 (5th Cir. 1975); Schenebeck v. Sterling Drug Inc., 423 F.2d 919, 924 (8th Cir. 1970); R. J. Reynolds Tobacco Co. v. Hudson, 314 F.2d 776, 780 (5th Cir. 1963); Brush Beryllium Co. v. Meckley, 284 F.2d 797, 800 (6th Cir. 1960); Sylvania Elec. Prods., Inc. v. Barker, 228 F.2d 847-8 (1st Cir. 1955), cert. denied, 350 U.S. 988, 76 S.Ct. 475, 100 L.Ed. 854 (1956); Thrift v. Tenneco Chems., Inc., 381 F.Supp. 543, 546 (N.D.Tex.1974); Withers v. Sterling Drug, Inc., 319 F.Supp. 878, 881 (S.D.Ind.1970); Steiner v. Ciba -Geigy Corp., 364 So.2d 47, 48-49 (FIa.Dist.Ct.App.1978); Chrischilles v. Griswold, 260 Iowa 453, 463, 150 N.W.2d 94, 199 (1967); Raymond v. Eli Lilly & Co., 117 N.H. 164, 175-76, 371 A.2d 170, 177 (1977); Frohs v. Greene, 253 Or. 1, 2-4, 452 P.2d 564, 565 (1969); Gilbert v. Jones, 523 S.W.2d 211, 213 (Tenn.Ct. App. 1974).
Indeed, asbestosis cases themselves are a prime example of the application of the “discovery rule” applied in statute of limitations cases. See Karjaia v. Johns-Manville Prods. Corp., 523 F.2d 155, 160-61 (8th Cir. 1975); Borel, supra, 493 F.2d at 1102; Velasquez v. Fibreboard Paper Prods. Corp., 97 Cal.App.3d, 881, 889, 159 Cal.Rptr. 113, 118 (1979); Louisville Trust Co. v. Johns-Manville Prods. Corp., 580 S.W.2d 497 (1979); Harig v. Johns-Manville Prods. Corp., 284 Md. 70, 71, 394 A.2d 299, 300 (1978).
We note, however, that a minority of cases have rejected this liberal view and strictly hold that the statute of limitations begins to run from the time of injury, even if undiscovered. See Thornton v. Roosevelt Hospital, 47 N.Y.2d 780, 417 N.Y.S. 920, 391 N.E.2d 1002 (1979).
. See e. g. Tijsseling v. General Accident Fire & Life Assurance Corp., 55 Cal.App.3d 623, 127 Cal.Rptr. 681 (1976) (that the cause of action accrued at manifestation did not mean that damages for insurance coverage purposes had to be measured from that point as well).
. See, e. g. General Dynamics Corp. v. Benefits Review Bd, 565 F.2d 208 (2d Cir. 1977); Travelers Ins. Co. v. Cardillo, 225 F.2d 137, cert. denied, 350 U.S. 913, 76 S.Ct. 196, 100 L.Ed. 800 (1955); Grain Handling Co. v. Sweeney, 102 F.2d 464 (2d Cir.), cert. denied, 308 U.S. 570, 60 S.Ct. 83, 84 L.Ed. 478 (1939); Sylvia's Case, 313 Mass. 313, 47 N.E.2d 293 (1943); Yurow v. Jersey Hat Corp., 131 N.J.L, 265, 36 A.2d 296 (1944), aff’d 132 N.J.L. 180, 39 A.2d 371 (1944).
. We note that Illinois has passed specific worker’s compensation legislation to deal with asbestosis and silicosis cases. Under this legislation, the last employer where the worker was exposed to asbestosis is liable. Ch. 48 Ill.Rev. Stat., § 172.36(d) (1975). New Jersey Law is similar. See Bucuk v. Zusi Brass Foundry, 49 N.J.Super. 187, 139 A.2d 436 (1958).
. See e. g. Royal Family Ins. Co. v. Grimes, 42 Ala.App. 481, 168 So.2d 262 (1964); Metropolitan Life Ins. Co. v. Reynolds, 48 Ariz. 205, 60 P.2d 1070 (1936); Broccolo v. Horace Mann Mut. Cas. Co., 37 Ill.App.2d 493, 186 N.E.2d 89 (1962); Cohen v. North American Life & Casualty Co., 150 Minn. 507, 185 N.W. 939 (1921); Kissil v. Beneficial Nat. Life Ins. Co., 64 N.J. 555, 319 A.2d 67 (1974); Reiser v. Metropolitan Life Ins. Co., 262 App.Div. 171, 28 N.Y.S.2d 283 (1st Dept. 1941), aff’d 289 N.Y. 561, 43 N.E.2d 534 (1942). See generally Annot., 53 A.L.R.2d 686 (1957).
. There is yet another line of cases relating to property insurance. Under these cases, latent defects in property which later cause damage do not trigger insurance coverage. See Maples v. Aetna Casualty & Surety Co., 83 Cal.App.3d 641, 148 Cal.Rptr. 80 (1978); Tijsseling v. Gen'l Accident Fire & Life Assur. Corp., 55 Cal. App.3d 623, 127 Cal.Rptr. 681 (1976); Remmer v. Glens Falls Indemnity Co., 140 Cal.App.2d 84, 295 P.2d 19 (1956); Oceanonics Inc. v. Petroleum Distrib. Co., 280 So.2d 874 (La.App. 1973), aff’d; 292 So.2d 190 (La.1974); Deodato v. Hartford Ins. Co., 143 N.J.Super. 396, 363 A.2d 361 (1976), aff'd, 154 N.J.Super. 263, 381 A.2d 354 (1977); Annot. 57 A.L.R.2d 1385 (1958), contra Sylla v. United States Fidelity & Guaranty Co., 54 Cal.App.3d 895, 127 Cal.Rptr. 38 (1976). In these cases, insurance coverage did not begin until the defect showed itself. However, in each of these cases, no damage or injury of any kind took place until manifestation.
However, when courts are dealing with property damage situations where damages slowly accumulate, courts have generally applied the exposure theory. So long as there is tangible damage, even if minute, courts have allowed coverage from that time. Champion Int’I Corp. v. Continental Casualty Co., 546 F.2d 502 (2d Cir. 1976), cert. denied, 434 U.S. 819, 98 S.Ct. 59, 54 L.Ed.2d 75 (1971); Union Carbide Corp. v. Travelers Indemnity Co., 399 F.Supp. 12 (W.D.Pa.1975); Gruol Constr. Co. v. Ins. Co. of N. America, 11 Wash.App. 632, 524 P.2d 427 (1974). But see United States Fidelity & Guar. Co. v. American Ins. Co., 169 Ind.App. 1, 345 N.E.2d 267.
It is true that the above cases stand for the general rule that general liability insurance policies protect against injury and not just against the occurrence of a negligent act where that act does not result in injury. However, as the district court noted below, “[t]he negligent acts alleged in the underlying lawsuits occurred when the products were sold without sufficient warning, not when the plaintiffs were exposed. The medical evidence in this case indicates clearly that injury to lung tissue occurs soon after exposure, not when the disease manifests itself.” 451 F.Supp. at 1241.
. We additionally note that published interpretations of the standard clauses in the industry-wide comprehensive general liability policy recognized the possibility that a court would adopt a proration scheme. See R. Elliott, The New Comprehensive General Liability Policy 12-3 to 12-5 in Practicing Law Institute, Liability Insurance Disputes (Sol Schreiber, ed. 1968) (“In some exposure types of cases involving cumulative injuries it is possible that more than one policy will afford coverage. Under these circumstances, each policy will afford coverage to the bodily injury or property damage occurring during its policy period.”); W. Abrist, The New Comprehensive General Liability Insurance Policy: A Coverage Analysis (Defense Res. Inst., Inc. 1966) (“In some exposure types of cases involving cumulative injuries it is possible that more than one policy will afford coverage each to apply to bodily injury or property damage occurring during its policy period.”) This shows that there was at least a recognized possibility that a court would adopt some form of proration scheme.
. See appendix A.
. Appellants’ argument in terrorem is that a ruling in favor of the exposure theory will create vast liability for a host of latent diseases. The appellants point out that the insurance policies in question expired years ago. We here rule only concerning asbestosis, a disease which has been long-known. If a disease is unknown, we see no underlying basis for liability. The problems of liability under old insurance policies are no different here than in cases where a minor was injured and we did not sue while he was under a disability.
. Tampa Elec. Co. v. Stone & Webster Corp., 367 F.Supp. 27 (M.D.Fla.1973); Hogan v. Midland Nat’l. Ins. Co., 3 Cal.3d 553, 91 Cal.Rptr. 153, 476 P.2d 825 (1970); St. Paul Fire & Marine Ins. Co. v. Hodor, 200 So.2d 205 (Fla.App. 1967); Prince v. Universal Underwriters Ins. Co., 143 N.W.2d 708 (N.D.1966); Mannheimer Bros. v. Kansas Cas. & Sur. Co., 149 Minn. 482, 184 N.W. 189 (1921); Home Ins. Co. v. Pinski Bros. Inc., 160 Mont. 219, 500 P.2d 945 (1972).
. See e.g. All Star Ins. Corp. v. Steel Bar, Inc., 324 F.Supp. 160, 163 (N.D.Ind. 1971); Apex Mutual Ins. Co. v. Christner, 99 Ill.App.2d 153, 240 N.E.2d 742 (1968); Marston v. Merchants Mutual Ins. Co., 319 A.2d 111, 114 (Me.1974); Satterwhite v. Stolz, 79 N.M. 320, 442 P.2d 810, 812 (1968); Steel Erection Co. v. Travelers Indemnity Co., 392 S.W.2d 713, 715-6 (Tex.Civ. App.1965); Waite v. Aetna Casualty & Surety Co., 77 Wash.2d 850, 467 P.2d 847, 852-3 (1970).
. See cases cited at n.22.
. Forty-Eight raises the spectre of a conflict of interest between the insurers and Forty-Eight, since the insurance company would have an interest in minimizing the amount of exposure during its policy period and maximizing that of the manufacturer for those periods that the latter was self-insured. We cannot agree that this is a significant problem in light of an attorney’s duty to protect the interest of the insured. See e.g. Parsons v. Continental Nat’l. Am. Group, 113 Ariz. 223, 550 P.2d 94 (1976).
. As we note in n.3, supra, in some cases, courts have adopted a theory of industry-wide enterprise liability even absent proof that a given company’s product or products caused injury. So far as we know, however, this broad theory has not been extended to asbestosis cases.
. We recognize that the general rule is that the burden of proof to show coverage is on the insured. See e.g. Hayes v. Country Mutual Ins. Co., 28 Ill.2d 601, 192 N.E.2d 855, 858 (1963); Warren v. Employers' Fire Ins. Co., 100 N.J.Super. 464, 242 A.2d 635, 639 (1968), rev’d on other grounds, 53 N.J. 308, 250 A.2d 578 (1969); 46 C.J.S., Insurance § 1316; 44 Am. Jur.2d Insurance § 1964.
In an asbestosis case, however, under Borel, all that an injured worker need do is prove that a given manufacturer’s products contributed to the worker’s asbestosis at any time. That is enough to trigger joint and several liability. As the briefs filed in this case forcefully remind us, few companies keep records more than 15 or 20 years, if that long. Thus, in many if not most cases, it will be impossible to tell which company’s products were used at what time. At best, there will be vague testimony that manufacturer X’s products were used at a certain job site during certain years. Considering that much exposure took place during the 1930’s and 1940’s, vague testimony is the best evidence that can be unearthed in many cases.
. Appellants are correct that the exposure theory we adopt has problems with “stacking”. From 1955 through 1977, Forty-Eight held twelve different insurance policies issued by five different companies. Eleven of these policies had aggregate limits of from $300,000 to $500,000 per occurrence. The twelfth policy had an aggregate limit of $1,000,000. The combined aggregate limits of the twelve policies is $5.6 million.
The problem is that if the inhalation of each asbestos fiber is deemed to be a separate “bodily injury”, this results in the “stacking” of liability coverage to produce coverage that is many times $5.6 million. This amounts to giving Forty -Eight much more insurance than it paid for. The district court recognized the problem which stacking presented. The court stated:
In any event, no insurer should be held liable in any one case to indemnify Forty -Eight for judgment liability for more than the highest single yearly limit in a policy that existed during the period of the claimant’s exposure for which judgment was obtained. 451 F.Supp. at 1243.
We agree with the district court. The initial exposure to asbestos fibers in any given year triggers coverage. However, under the terms of the policies, additional exposure to asbestos fibers is treated as arising out of the same occurrence. Thus, on its face, the liability of each insurer is limited to maximum amount “per occurrence” provided by each policy. We have no problem with the district court’s extending the policy language so that each insurer would face no more liability per claim than the maximum limit it wrote during any applicable year of coverage.
. Appellant Liberty Mutual warns us in its brief that the district court’s opinion, “if not corrected, will plague the nation’s trial courts and appellate courts, together with insurers and insureds alike, for many years to come.” We think that the “plague” will come regardless of how we decide this case. The cause of the plague, however, is an insurance industry which adopted a standard policy which is inadequate to deal with the problems of asbestosis.