William O'Connell is in prison for sexual acts with a minor. When he committed his crimes, O'Connell was a priest in the Roman Catholic Diocese of Providence, Rhode Island. The Diocese and its insurers settled the victim's tort claim. The current litigation, under the alien-citizen jurisdiction of 28 U.S.C. § 1332(a)(3), requires us to determine how much of the settlement is the responsibility of the Diocese and its primary insurers, Lloyd's of London and Centennial Insurance Company (collectively "Lloyd's"), and how much must be paid by the excess carrier, Interstate Fire & Casualty Company. The answer depends on how many insured "occurrences" there were: the Diocese is self-insured for the first $100,000 per occurrence, and Lloyd's share likewise depends on the number of occurrences. O'Connell sexually abused one boy during two policy years, and in two distinct places. That's at least two occurrences from the victim's perspective, but O'Connell's crimes were intentional torts, excluded from coverage. The Diocese's liability arises from negligence of the Bishop of Providence (and his subordinates) in supervising O'Connell. The Diocese and Lloyd's contend, and the district judge held, that negligent supervision is a single "occurrence" no matter how many years or places the abuse spans. The Lloyd's policy has been sold across the country. No state court has addressed the meaning of "occurrence" under this policy, but two courts of appeals have held that there is one "occurrence" per priest, per abused child, per policy year. Roman Catholic Diocese of Lafayette v. Interstate Fire & Casualty Co., 26 F.3d 1359 (5th Cir.1994) (Louisiana law); Interstate Fire & Casualty Co. v. Archdiocese of Portland, 35 F.3d 1325 (9th Cir.1994) (Oregon law).

Because this suit was filed in Illinois, that state's choice-of-law rules govern. Klaxon Co. v. Stentor Electric Manufacturing Co., 313 U.S. 487, 61 S. Ct. 1020, 85 L. Ed. 1477 (1941). The policy does not contain a choice-of-law clause, and Illinois' rules for selecting a body of law in the absence of such a clause are obscure, as we have remarked on several occasions. Palmer v. Beverly Enterprises, 823 F.2d 1105 (7th Cir.1987); Florida Risk Planning Consultants, Inc. v. Transport Life Insurance Co., 732 F.2d 593 (7th Cir.1984). It has been 52 years since the Supreme Court of Illinois devoted sustained attention to the choice-of-law principles for contract cases. See Oakes v. Chicago Fire Brick Co., 388 Ill. 474, 58 N.E.2d 460 (1944). The law of other states has undergone a dizzying evolution in that time--from a formal last-act or "delivery" approach to a complex weigh-the-contacts approach. Applying a weighing approach, the district court held that the location of the insured risk is the dominant contact, requiring application of Rhode Island law. 826 F. Supp. 1156 (N.D.Ill.1993).

Recently the Supreme Court of Illinois demonstrated that clarity remains elusive. Ruling on the choice-of-law issues presented by an insurance contract that lacked a choice-of-law clause, that court stated that the appropriate law is

governed by the location of the subject matter, the place of delivery of the contract, the domicile of the insured or of the insurer, the place of the last act to give rise to a valid contract, and place of performance, or other place bearing a rational relationship to the general contract.

Lapham-Hickey Steel Corp. v. Protection Mutual Insurance Co., 166 Ill. 2d 520, 526-27, 211 Ill. Dec. 459, 462, 655 N.E.2d 842, 845 (1995), quoting from Hofeld v. Nationwide Life Insurance Co., 59 Ill. 2d 522, 528, 322 N.E.2d 454, 457-58 (1975). This formulation does not choose between the old, formal approach and the modern, "contacts" approach; instead it includes elements of both, without offering any guidance for what happens when a contract of insurance is delivered in one state but covers a risk in another.

Lapham-Hickey chose the place of the policy's delivery, rather than the place of the insured risk, to supply the law for an insurance contract. Nonetheless, we have no doubt that Rhode Island is the only sensible choice of law for the current dispute. Two policies with identical language covering the same risk should have identical effects. Lloyd's issued the basic policy; Interstate sold a "follow form" excess policy with identical substantive terms. It would be absurd if the primary policy were governed by United Kingdom law, and the excess policy by Illinois law, just because that is the location of the insurers (and therefore of the last acts necessary to make the insurance effective). We do not think that Illinois would apply multiple bodies of law to a single disputed term. That is in the end what Lapham-Hickey stands for: in that case a single policy covered risks in many states, so applying the law of the delivery state produced a consistent interpretation. The same objective here points to Rhode Island.

Like other states, Rhode Island starts with the language of the policy. Hodor v. United Services Automobile Ass'n, 637 A.2d 357 (R.I.1994). Lloyd's policy defines "occurrence" as

an accident or a happening or event or a continuous or repeated exposure to conditions which unexpectedly and unintentionally results in personal injury, or damage to property during the policy period. All such exposure to substantially the same general conditions existing at or emanating from one location shall be deemed one occurrence.

Lloyd's tells us that the victim suffered "a continuous or repeated exposure to conditions"--that is, to O'Connell's sexual proclivities--and that repeated exposure to "substantially the same general conditions" can be only one "occurrence." It depicts a pedophilic priest as similar to hazardous waste: living next to a church from which oil has seeped into the ground is one "occurrence" no matter how long the conditions exist. Because the insured event is the Bishop's negligent supervision, all consequences of that negligence are a single occurrence. See CPC International, Inc. v. Northbrook Excess & Surplus Insurance Co., 668 A.2d 647 (R.I.1995). Multiple injuries with a single cause do not count as multiple occurrences. Bartholomew v. Insurance Co. of North America, 502 F. Supp. 246 (D.R.I.1980), affirmed under the name Bartholomew v. Appalachian Insurance Co., 655 F.2d 27 (1st Cir.1981) (Rhode Island law); accord, Rozenfeld v. Medical Protective Co., 73 F.3d 154 (7th Cir.1996) (Illinois law). Interstate responds that O'Connell's wrongful acts occurred at more than one location, and that sexual abuse cannot be called "continuous." Each episode was discrete; O'Connell could have stopped at any time; and by continuing to commit new wrongs, with cumulative injury, O'Connell brought about multiple occurrences. Of course the priest's acts are not covered by the policy, but supervision itself is (or can be) discrete. The Archdiocese of Portland received multiple reports about its priest's misconduct, over a period of four years, and did nothing. Portland, 35 F.3d at 1327. Perhaps there were also multiple supervisory lapses in Providence.

Rhode Island law tells us that if "negligent supervision" is a unitary act, then multiple losses do not create multiple occurrences. But is that the right way to describe careless supervision? At oral argument, counsel for Lloyd's conceded that if O'Connell had abused two boys in a single policy year, that would be two "occurrences." Presumably two priests abusing four boys would be four occurrences. From the victim's perspective, this makes sense. Each loss is independent, and this understanding affords both the victim and the insured Diocese one full "occurrence" worth of coverage. What counsel could not explain is how this understanding is reconcilable with Lloyd's interpretation of the policy's definition. A slumbering or indifferent supervisory structure is one "occurrence" by Lloyd's lights--and if one priest abusing two boys in one policy year is two occurrences, why not one priest abusing the same boy twice, in two policy years?

The language defining "cause" does not speak directly to this question. It assumes a two-party perspective--that an insured tortfeasor has harmed a victim. Its language is a mismatch for a case in which the tort is negligent supervision of an intentional wrongdoer. "[C]ontinuous or repeated exposure to conditions" sounds like language designed to deal with asbestos fibers in the air, or lead-based paint on the walls, rather than with priests and choirboys. A priest is not a "condition" but a sentient being, and of course the victim was never "exposed" to the Diocese's negligent supervision. Invocation of the contra proferentum principle under Rhode Island law, see Textron, Inc. v. Aetna Casualty & Surety Co., 638 A.2d 537, 539 (R.I.1994), is no solution: Interstate chose to "follow form" to the Lloyd's policy and is therefore equally its author. Anyway, what it means to construe this definition against the author is itself ambiguous. Winners and losers will change with the circumstances. Interstate today wants to call sustained sexual abuse multiple occurrences to increase the number of deductibles the Diocese must cover and the number of contributions the primary carrier must make. But if tomorrow the victim's loss exceeds the maximum coverage for a single occurrence, the roles will be reversed. The excess carrier would want to call the sexual abuse a single occurrence to cap its own exposure, while the Diocese would favor multiple occurrences in order to maximize its insurance coverage.

Following the fifth and ninth circuits, both Lloyd's and Interstate assume that every child abuse case produces either one "occurrence" or many, according to the number of victims and policy years involved. We do not think that Rhode Island would find either end of this continuum attractive. Hodor and Bartholomew show that multiple injuries to a single victim cannot create multiple occurrences, no matter how many policy years the injuries span, if there is only one insured negligent act. But a single negligent act undoubtedly can produce multiple "occurrences" if the injuries are independent--consider the pharmaceutical company that negligently prepares a batch of drugs, injuring many users; or consider the diocese that digs a basement for a cathedral and fails to erect a fence, attracting several children who fall to their deaths. Similarly, the same kind of negligent act can occur several times with separate injuries, producing several occurrences. A physician's malpractice during an operation may injure the patient; during a second operation to repair the damage, the physician may bungle the job again, aggravating the injury. These are two occurrences, no different from two physicians negligently performing two surgeries. Cf. Goodhand v. United States, 40 F.3d 209, 214-15 (7th Cir.1994). Just so if the surgeon is an intern, and the negligence lies in a senior physician's careless supervision of the student during each operation.

Similar principles are at work here. The tort is poor supervision, not negligent hiring, which occurs only once per employee-employer pair. An employer may leave employees to their own devices, but even the decision not to supervise may be revisited now and again. If as in the Portland case a diocese receives multiple warnings about a priest's misconduct and ignores all of them, it would be appropriate to call those lapses multiple occurrences-for intervention after any one of them could have avoided some of the injury. But if instead the diocese receives no danger signals, and its negligence lies in failure to investigate on its own (if that is negligence at all), or if the diocese intervenes but takes action that in retrospect is inadequate, then the lapse looks more like a single occurrence under the policy's definition.

The victim's suit against the Diocese was settled; the record in that case does not reveal what the "negligent supervision" entailed. Did the Diocese receive warnings? How many, and when? Should the Bishop have been suspicious in the absence of warnings? How does the Diocese supervise its priests? Canon law prescribes the manner of supervision, but was it followed? The parties to this case did not offer information pertinent to these questions, and neither side has asked us to remand so that such information may be offered. Both sides were content to submit the case for decision on the theory that the policy itself chooses between one occurrence or many. As we do not agree with that position, the question arises: Should we remand for development of a factual record? We think not, because the plaintiffs, who bear the burden of persuasion (and of production), have never suggested that the details of the Diocese's conduct matter or asked for an opportunity to submit evidence. The Diocese and its three insurers chipped in resources, for purposes of settlement, on the assumption that there were two "occurrences," and they agreed to work out their differences in this declaratory judgment action. After this case began, the Diocese resolved its differences with Interstate. That left Lloyd's with the laboring oar, and it decided--perhaps because it knew what the facts about the Diocese's supervision were--to submit the case for decision as a pure question of law, without any fallback position. It is inappropriate to kibitz this litigation strategy. Because Rhode Island would not treat negligent supervision as invariably one "occurrence," Lloyd's has not shown an entitlement to relief. The parties' allocation of shares in the settlement pool should not have been disturbed.

REVERSED.