J. SKELLY WRIGHT, Circuit Judge:

We review a grant of summary judgment in favor of Eli Lilly and Company in its declaratory judgment suit against various insurance companies. Eli Lilly's suit concerns the scope of coverage for claims arising from the company's manufacture and sale of the drug DES (diethylstilbestrol). The District Court determined that Indiana law controlled this diversity action and granted summary judgment for Eli Lilly on the basis of its interpretation of Indiana law. The insurance companies appeal.

We agree with the District Court that Indiana law controls. However, we find ourselves uncertain about the application of that law to the issues at hand. Importantly, Indiana law provides a procedure by which uncertain questions of state law may be certified from federal appellate courts to the Indiana Supreme Court. Ind.R.App.P. 15(O) (1985). The use of such certification procedures lies within the discretion of the federal courts. Lehman Brothers v. Schein, 416 U.S. 386, 391, 94 S.Ct. 1741, 1744, 40 L.Ed.2d 215 (1974). We now exercise that discretion and certify the issues to the Indiana Supreme Court for resolution.

I A.

The sad tale of DES continues to unfold. Hailed as a boon for women with a risk of miscarriages, DES was prescribed to pregnant women from the late 1940's until 1971. In 1970, however, researchers reported a significant connection between (1) ingestion of DES by pregnant women and (2) later development of vaginal clear-cell adenocarcinoma and other diseases in those daughters of the DES users who were in utero at the time of the DES ingestion. As DES daughters increasingly discovered what they thought were DES-related illnesses, they filed lawsuits against DES manufacturers. The complexities of liability for DES-related illnesses have presented courts with difficult, challenging questions, see, e.g., Sindell v. Abbott Laboratories, 26 Cal.3d 588, 163 Cal.Rptr. 132, 607 P.2d 924 (Cal.), cert. denied, 449 U.S. 912, 101 S.Ct. 285, 66 L.Ed.2d 140 (1980), and massive DES litigation continues to loom as people who believe they are DES victims seek a measure of compensation and redress in courts of law.

B.

From 1947 to 1967, Eli Lilly was one of the major manufacturers and sellers of DES. As a result, more than six hundred lawsuits have already been filed against Eli Lilly for DES-related illnesses.

On March 9, 1982, Eli Lilly brought this declaratory judgment action against the various insurance companies that insured it from the time it first manufactured DES (1947) to the time insurance companies finally refused to insure DES risks (1976).

The declaratory judgment turns on the interpretation of Eli Lilly's insurance policies. During the 29 years at issue, Eli Lilly was covered by some 242 insurance policies. The relevant clause of the policies appears in four slightly different versions. Three of the four versions define coverage in terms of an "injury" that "occurs" during the policy period. 2 Joint Appendix (JA) 96, 98, 99-100. The fourth version — reflected in 21 of the policies — refers to injuries "sustained" during the policy period. 2 JA 97. No party contends that these slight differences require different interpretations. See Joint Brief of Appellants-Defendants at 8-9; Brief of Appellee at 8-9. Furthermore, the "injury"/"occurrence" formulation is, in all material respects, identical to the insurance industry's Comprehensive General Liability Policy (CGL) provision that has been in effect since the 1960's.

The policies do not define the relevant terms with precision. For instance, in 199 of the 242 policies the insurance companies agreed to insure Eli Lilly for "personal injuries * * * caused by or arising out of each occurrence anywhere in the world." 2 JA 96. "Personal injury" was defined in part, and not altogether helpfully, as "bodily injury, mental injury, mental anguish." Id. "Occurrence" was defined, in turn, as "an accident or a happening or event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally results in personal injury ... during the policy period." Id. at 96-97. Our task is to interpret these terms with respect to claims arising from DES-related illnesses.

C.

The determinative question is when the DES-related "injury" "occurs" for purposes of the policy. At least two possible time pegs are relevant: (1) the time of exposure (when the DES mother ingested the DES), and (2) the time of manifestation (when the DES daughter discovered her illness). The insurance companies vigorously disagree as to which of these times should be relevant, but they have put that dispute aside for the present. Their argument in the current proceeding is that only one of those time pegs — exposure or manifestation — can be relevant; they will fight out later which one is relevant. Eli Lilly, in contrast, argues that the policies should be governed by what some courts have called the multiple trigger theory: the injury should be understood to have "occurred" at exposure, manifestation, and the period between the two while the disease was latent.

Courts have split in deciding how to interpret this insurance policy provision for illnesses in which there is a delay between exposure and manifestation. In the context of asbestosis, which results from exposure to asbestos, courts have split three ways. The Fifth and Sixth Circuits have held that state courts would adopt an "exposure" theory. See Porter v. American Optical Corp., 641 F.2d 1128 (5th Cir.), cert. denied, 454 U.S. 1109, 102 S.Ct. 686, 70 L.Ed.2d 650 (1981) (citing Louisiana law); Insurance Co. of North America v. Forty-Eight Insulations, Inc., 633 F.2d 1212 (6th Cir. 1980), cert. denied, 454 U.S. 1109, 102 S.Ct. 686, 70 L.Ed.2d 650 (1981) (citing Illinois and New Jersey law). The First Circuit has held that state courts would adopt a "manifestation" theory. See Eagle-Picher Industries, Inc. v. Liberty Mutual Insurance Co., 682 F.2d 12 (1st Cir. 1982), cert. denied, 460 U.S. 1028, 103 S.Ct. 1279, 75 L.Ed.2d 500 (1983) (citing Illinois and Ohio law). Finally, the District of Columbia Circuit has held that several states would adopt a "multiple trigger" theory, see Keene Corp. v. Insurance Co. of North America, 667 F.2d 1034 (D.C. Cir. 1981), cert. denied, 455 U.S. 1007, 102 S.Ct. 1644, 71 L.Ed.2d 875 (1982) (citing "basic principles" shared by Delaware, New York, the District of Columbia, Pennsylvania, Connecticut, and Massachusetts), and the Pennsylvania courts have themselves adopted a multiple trigger theory, see Crown Cork Seal Co. v. Aetna Casualty Surety Co., No. 1292 September Term (Philadelphia Court of Common Pleas, Aug. 2, 1983).

Other delayed manifestation illnesses have also produced differences in judicial interpretation of the CGL-type "trigger" provision. The Fifth Circuit has tentatively extended its exposure holding to silicosis. See Ducre v. Executive Officers of Halter Marine, Inc., 752 F.2d 976 (5th Cir. 1985). Similarly, the Pennsylvania courts have extended their multiple trigger holding to DES-related illnesses. See Vale Chemical Co. v. Hartford Accident Indemnity Co., ___ Pa.Super. ___, 490 A.2d 896 (1985), petition docketed, No. 491 E.D. Allocatur Docket 1985) (Pa. April 22, 1985) ( reproduced in Joint Supplemental Filing of Certain Defendants-Appellants as Appendix 1).

The status of the law in some other jurisdictions is less clear. In New York one state court has held that manifestation triggers coverage for DES-related illnesses, but that court did not explicitly reject the possibility that other triggers might exist as well. American Motorists Insurance Co. v. E.R. Squibb Sons, Inc., 95 Misc.2d 222, 406 N.Y.S.2d 658 (Sup.Ct. 1978). Applying New York law, however, the Second Circuit has rejected the multiple trigger theory and adopted an "injury in fact" interpretation. See American Home Products Corp. v. Liberty Mutual Insurance Co., 565 F. Supp. 1485 (S.D.N.Y. 1983), aff'd as modified, 748 F.2d 760 (2d Cir. 1984). In other cases the Second Circuit has suggested that, under New York law, no determination can be made until an extensive consideration of extrinsic evidence is completed. See, e.g., Schering Corp. v. Home Insurance Co., 712 F.2d 4 (2d Cir. 1983); cf. Emons Industries v. Liberty Mutual Fire Insurance Co., 567 F. Supp. 335 (S.D.N.Y. 1983). The Wisconsin Supreme Court, moreover, finding differences in the insurance contracts before it, held that ingestion (exposure) and market participation triggered coverage for policies until 1968, and an "injury" triggered coverage for policies after 1968. See Kremers-Urban Co. v. American Employers Insurance Co., 119 Wis.2d 722, 351 N.W.2d 156 (1984).

Thus, in the context of delayed manifestation illnesses, three principal interpretations have emerged: exposure, manifestation, and multiple trigger. In addition, some courts have considered "injury in fact" theories, and some have seemed to suggest that interpretation of this insurance policy provision must be extremely contextual.

D.

In the proceedings below the District Court granted summary judgment to Eli Lilly and interpreted the insurance policy provisions to provide multiple trigger coverage. The court determined, first, that a conflict of laws existed, and that Indiana law should govern. The court determined, second, that, under Indiana law, extrinsic evidence is not relevant in construing insurance contracts. As a result, concluded the court, under the Federal Rules there was no genuine issue of material fact precluding summary judgment. See Fed.R.Civ.P. 56(c). Finally, noting that the interpretation of the provision at issue has not been decided in Indiana, the court held that the Indiana courts would adopt a multiple trigger interpretation because that interpretation would best effectuate Indiana's principle of construing ambiguities in favor of the insured and giving force to the insured's reasonable expectations.

The insurance companies appeal. They vigorously argue that, in construing the insurance policies at issue, Indiana courts would consider extrinsic evidence. The companies have submitted extrinsic evidence which, they claim, is relevant to (1) the parties' intent in adopting the policies, (2) the parties' course of conduct in applying the policies, (3) Eli Lilly's sophistication and strength as a bargaining partner, and (4) the medical nature of DES-related illnesses. Since this extrinsic evidence should be considered under Indiana's contract and insurance law, the insurance companies argue, there is a genuine issue of material fact under the Federal Rules, and the summary judgment should be reversed.

II A.

We must first determine whether the District Court correctly determined that Indiana law should be applied.

In a diversity case a federal court must apply the choice of law principles of the jurisdiction in which it sits. Klaxon Co. v. Stentor Electric Mfg. Co., 313 U.S. 487, 61 S.Ct. 1020, 85 L.Ed. 1477 (1941). Under District of Columbia principles, we must first determine whether there is a conflict between the laws of the relevant jurisdictions. Fowler v. A A Co., 262 A.2d 344, 348 (D.C. 1970); Gaither v. Myers, 404 F.2d 216, 222 (D.C. Cir. 1968). As the District Court observed, at least three states would have an interest in applying their substantive law. Indiana has an interest: Eli Lilly's principal place of business and place of incorporation are in Indiana, and many of the policies were negotiated and concluded there. New York and Pennsylvania also have an interest: along with Indiana, those are the principal states where the policies were negotiated and concluded. Furthermore, as the District Court also found, there is a conflict between at least two of the relevant jurisdictions. Pennsylvania has adopted a multiple trigger theory for interpreting the CGL-like trigger provision, see Vale, supra, but New York apparently has not, see American Home Products Corp., supra.

Since there is a conflict, we must determine which jurisdiction has the "more substantial interest," DeMontmorin v. Du Pont, 484 A.2d 582, 585 (D.C. 1984); Blair v. Prudential Insurance Co., 472 F.2d 1356, 1359 (D.C. Cir. 1972). Indiana, Pennsylvania, and New York have similar interests with respect to their role as the site of contract negotiation and agreement. Indiana, however, has a strong additional interest because it is Eli Lilly's principal place of business and place of incorporation. Apparently, no appreciable difference emerges in the location of the insurance companies and of the potential Eli Lilly DES victims; no such differences have been argued, and our review of the record reveals none. In light of these circumstances, we agree with the District Court that Indiana law should govern.

The insurance companies do not dispute the holding that Indiana law should govern. In addition to its arguments that the District Court's interpretation of Indiana law was correct, however, Eli Lilly argues, in the alternative, that Indiana law should not control.

Eli Lilly offers two reasons for this alternative argument. First, it argues that there is no "real" conflict in the jurisdictions at issue — or in any jurisdiction — because the holdings of all jurisdictions have ultimately given coverage to the insured. While we agree that construing ambiguities in insurance contracts in favor of the insured is a venerable and important policy in most, if not all, jurisdictions, we do not agree with Eli Lilly's apparent proposition that the complexities of state law are therefore irrelevant. We note that federal and state courts have grappled with this issue, and we are not prepared to say that those efforts were merely gyrations toward a preordained result. Eli Lilly claims that none of the holdings is inconsistent with a multiple trigger theory. The short answer is that various courts have specifically rejected a multiple trigger theory. See, e.g., American Home Products, supra.

Second, Eli Lilly argues that this court is bound by our prior holding in Keene. Eli Lilly contends that Keene rested on "general principles of insurance contract interpretation" and that "the basic principles governing the interpretation of insurance policies under Indiana law are the same as those in most states and the same as those stated in Keene." Brief of Appellee at 26-27. However, this argument tends to ignore the fact that our mission here is to interpret and apply state law. Erie Railroad Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed.2d 1188 (1938). In Keene, it is true, this court did not particularly dwell on the nuances of state law. Rather, we observed that the multiple trigger theory reflected "basic principles" of contract law and that we saw no reason to believe that the six jurisdictions which might have an interest differed on these general principles. See Keene, supra, 667 F.2d at 1041 n. 10. Since Keene, however, it has become clear that not all states share the view expressed in Keene — or at least not all federal courts interpreting the laws of state jurisdictions do so. See American Home Products, supra; Eagle-Picher, supra. The interpretation of the "trigger" provision at issue must, then, be rooted in Indiana law; Keene is relevant only insofar as we determine that the Indiana courts would adopt it.

B.

Having determined that Indiana law governs, we must now determine the substance of that law. As noted previously, the District Court's holding rested on three determinations: (1) under Indiana law, extrinsic evidence should not be considered in interpreting this type of insurance policy provision; (2) since extrinsic evidence should not be considered, there is no genuine issue of material fact; and (3) under Indiana's principle of construing ambiguities in favor of the insured, the Indiana courts would adopt a multiple trigger analysis for DES-related illnesses, as the Pennsylvania courts and this court have done for asbestosis, and as the Pennsylvania courts have done for DES-related illnesses.

Our review of Indiana law leaves us uncertain as to the validity of these holdings. On the one hand, Indiana courts have sometimes interpreted insurance policy provisions without any reference to extrinsic evidence. See, e.g., Mutual Hospital Insurance v. Klapper, 153 Ind. App. 555, 288 N.E.2d 279, 281-282 (1972), transfer denied, 262 Ind. 144, 312 N.E.2d 482 (1974) (interpreting meaning of term in Indiana insurance policies without any reference to extrinsic evidence). On the other hand, however, some Indiana courts have emphasized that insurance contracts should be interpreted like other contracts, see, e.g., Asbury v. Indiana Union Mutual Insurance Co., 441 N.E.2d 232, 236 (Ind.App. 1982), and, at least in other contexts, Indiana courts have permitted introduction of extrinsic evidence to resolve contractual ambiguities. See, e.g., Goeke v. Merchants National Bank Trust Co., 467 N.E.2d 760, 765 (Ind.App. 1st Dist. 1984).

Additionally, some Indiana courts have emphasized that the question of extrinsic evidence turns on whether the ambiguity is facial (sometimes called "patent") or contextual (sometimes called "latent"). See, e.g., Michigan Mutual Insurance Co. v. Combs, 446 N.E.2d 1001 (Ind.App. 1983); Churchwell v. Firestone Industrial Products Co., 431 N.E.2d 853, 854 (Ind.App. 2d Dist. 1982); Huntington Mutual Insurance Co. v. Walker, 392 N.E.2d 1182, 1185 (Ind.App. 1st Dist. 1979). However, as one Indiana court has pointed out, that principle has been abandoned in many jurisdictions, Hauck v. Second National Bank of Richmond, 153 Ind. App. 245, 286 N.E.2d 852, 862 (2d Dist. 1972), and the Indiana Supreme Court has not addressed the distinction for many years. Id.

Furthermore, although the Indiana Supreme Court has emphasized that "[t]he terms of an insurance policy should be construed liberally in favor of the insured * * if there is an ambiguity in the policy's language," Miller v. Dilts, 463 N.E.2d 257, 265 (Ind. 1984); See also State Security Life Insurance v. Kintner, 243 Ind. 331, 185 N.E.2d 527 (1962), we have not found an Indiana Supreme Court decision that settles the question of how extrinsic evidence should be used in that analysis, if at all, and the parties have pointed us to none.

Thus, we find ourselves uncertain on an important point of Indiana law. The State of Indiana provides a procedure for federal appellate courts to certify uncertain questions of state law to the Indiana Supreme Court, Ind.R.App.P. 15(O), and the Seventh Circuit has used this procedure on several occasions, see, e.g., Metropolitan Life Insurance Co. v. Tallent, 445 N.E.2d 990 (Ind. 1983); Gabhart v. Gabhart, 267 Ind. 370, 370 N.E.2d 345 (1977). See also Walko Corp. v. Burger Chef Systems, Inc., 554 F.2d 1165 (D.C. Cir. 1977) (certifying question to Maryland Court of Appeals).

The United States Supreme Court has stressed the propriety of certification if the state law is uncertain. See Lehman Brothers v. Schein, supra, 416 U.S. at 391, 94 S.Ct. at 1744 (certification of uncertain questions "does, of course, in the long run save time, energy, and resources and helps build a cooperative judicial federalism"); Clay v. Sun Insurance Office Limited, 363 U.S. 207, 212, 80 S.Ct. 1222, 1225, 4 L.Ed.2d 1170 (1960) (Florida statute "which permits a federal court to certify * * * a doubtful question of state law" is an example of "rare foresight"). See also Elkins v. Moreno, 435 U.S. 647, 98 S.Ct. 1338, 55 L.Ed.2d 614 (1978) (certifying question to Maryland Court of Appeals); Bellotti v. Baird, 428 U.S. 132, 96 S.Ct. 2857, 49 L.Ed.2d 844 (1976) (certifying question to Supreme Judicial Court of Massachusetts). The Supreme Court has emphasized, moreover, that the use of certification "in a given case rests in the sound discretion of the federal court." Lehman Brothers, supra, 416 U.S. at 391, 94 S.Ct. at 1744.

We think this an appropriate case to exercise our discretion to certify. In addition to the uncertainty of state law, two other factors weigh in our analysis. First, this case is one of extreme public importance. As noted, the State of Indiana has a very substantial interest in this dispute. We point out an example that illustrates the stakes of this dispute. Since the insurance companies refused to issue insurance for DES-related illnesses after 1976, if a manifestation theory is adopted Eli Lilly will have no coverage for claims arising from diseases that "manifest" after 1976. Although firm figures are impossible at this point, Eli Lilly's counsel estimates that Eli Lilly would thus have no coverage for at least a third of the claims arising from its manufacture and sale of DES. The outcome of the insurance dispute, of course, affects not only the pharmaceutical company and insurance companies before us, but also the potential DES victims. Given the magnitude of the public interest, we feel it especially important to be certain that we do not misapply the governing law.

Second, we note that Eli Lilly has suggested only one reason for its decision to bring suit in this forum: the Keene precedent. Since Indiana law controls, however, Keene is relevant only insofar as it would be persuasive to Indiana courts. In view of the fact that different jurisdictions have reached differing results on the interpretation of this "trigger" provision, we think it appropriate to seek guidance from the Indiana Supreme Court on this question of first impression in Indiana law.

Taken together, these factors convince us that certification is an appropriate exercise of our discretion. We thus certify the following questions to the Indiana Supreme Court:

1. Under Indiana insurance contract law, should the types of extrinsic evidence proffered by the insurers be considered in the interpretation of the disputed "trigger" provisions.

2. If any aspect of the insurers' extrinsic evidence should be considered, would Indiana courts require a determination of the parties' actual intent concerning the application of the "trigger" provision to delayed manifestation injuries? Or would Indiana courts permit a determination, after consideration of the extrinsic evidence, that the parties did not hold or convey a clear understanding of the trigger provision's applicability to delayed manifestation injuries and that, thus, the provision must be interpreted by the court as a matter of Indiana insurance law?

3. If the insurers' extrinsic evidence should not be considered, or if that evidence is not determinative of the parties' intent, how should the insurance policy provision at issue be interpreted under Indiana law? In other words, would Indiana courts adopt an exposure, a manifestation, a multiple trigger, or some other interpretation of the "injury"/"occurrence" language in Eli Lilly's policies?

This court's opinion, together with copies of the briefs and of the record submitted on appeal, are transmitted herewith to the Indiana Supreme Court.

So ordered.