Plaintiff Scirex Corp. ("Scirex"), a firm specializing in clinical testing of new drugs for pharmaceutical companies, brought this suit against defendant Federal Insurance Company ("Federal") in the District Court for the Eastern District of Pennsylvania, seeking payment under Federal's "Blanket Employee Dishonesty" policy, which covered losses caused by Scirex employees' fraudulent and dishonest acts. Although the protocols for the four clinical trials at issue required Scirex's nurses to observe patients for eight hours and record their observations every thirty minutes, in many cases the nurses sent patients home after as little as an hour, yet they recorded and submitted observations allegedly covering the full eight hours. Their misrepresentations made it impossible for supervisors to discover the breaches of protocol until a former employee tipped them to the practice, at which advanced point the four studies were unfit for Food and Drug Administration review, and therefore were worthless.

Scirex replicated the studies, which cost a combined $1.2 million, at no charge to the sponsors. Federal, however, refused to cover Scirex's losses. It defended on the ground that the nurses' actions were not dishonest because they had acted on their belief that strict adherence to protocol was unnecessary, and that while their actions might have been negligent, "dishonesty" implies a cognizance of one's wrongdoing that they did not possess. Federal also maintained that even if the nurses' actions were dishonest, its policy covered only "direct" losses, and Scirex's losses more strongly resembled ordinary business expenses from failed ventures than losses, such as false claims of working overtime, due directly to employee dishonesty. Finally, Federal contended that even if it were liable for Scirex's losses, it would be liable only in the amount of $280,000, the policy limit for one occurrence, because the losses across the four ruined studies were related.

Following a bench trial, the District Court held that Scirex could not recover against Federal because "dishonesty" implies a culpable intent, and the nurses' "stubborn belief that the drug companies... were imposing unnecessary requirements" did not equate to dishonesty. Scirex Corp. v. Fed. Ins. Co., 2001 WL 1491133, *3-4, 2001 U.S. Dist. LEXIS 19088, *8-9 (E.D.Pa.2001). It therefore denied Scirex's claim, but noted in dicta that, were the nurses' actions to be found dishonest, the losses they caused would be direct, and therefore Federal's policy would cover them. Id. 2001 WL 1491133 at *4, at *11. It also opined that, at all events, the policy would limit Federal's liability to $280,000, the ceiling for one occurrence. Id. 2001 WL 1491133 at 5, at *14.

Because we conclude that the nurses' actions were dishonest, as well as negligent, we hold that they are covered by Federal's policy, and we reverse the District Court's holding to the contrary. We are satisfied that, whatever may be said of the decision to send patients home early in violation of protocol, the nurses' practice of submitting records containing observations they did not make is ineluctably and irrefutably dishonest. However, we agree with the District Court that the nurses' actions directly caused Scirex's losses. Pennsylvania law equates "direct cause" with "proximate cause," Jefferson Bank v. Progressive Casualty Ins. Co., 965 F.2d 1274, 1281-82 (3d Cir.1992), and the nurses' conduct rendered those studies worthless to their sponsors, and therefore worthless to Scirex. Finally, although we believe that Federal's policy covers Scirex's losses, we agree with the District Court that Federal's liability is limited to $280,000 for the four studies.

In 1997 and 1998, Scirex conducted four clinical studies for three different sponsors. Each tested a pain medication for patients who had undergone dental surgery, and each had a protocol, written by the pharmaceutical company sponsoring the study, which specified in great detail the procedures to be followed in that study. The four protocols required patients to remain under the observation of Scirex nurses, at Scirex's clinic, for at least eight hours after being given the first medication dose. Because the effects of pain medication to some extent vary by individual, the pharmaceutical companies anticipated that there would be a certain number of test subjects for whom the medication would be ineffective, and for these subjects, the protocols mandated providing supplemental pain medication. Subjects who received such supplemental medication were termed "rescued," while those who took only the drug being studied were termed "unrescued."

Although the protocols provided for rescuing certain subjects, they did not provide for treating those subjects otherwise differently; specifically, they stated that "[s]ubjects are required to remain at the study facility for the entire eight-hour postdosing observation period, even if the supplemental analgesic medication is taken." (R.W. Johnson Protocol.) During one of the four studies, however, Algos, the study's sponsor, told Scirex that rescued patients did not have to remain in the clinic for the full eight hours. That excepting instruction applied only to Algos's study, and even then, only to rescued subjects. The other three studies were to proceed strictly according to protocol.

During the course of a subject's eight-hour stay, the protocols required Scirex's nurses to observe the subject and record their observations. The records had to be "timely, accurate and complete," because they were what the FDA evaluated: "[t]imely, accurate, and complete reporting and analysis of safety information from clinical trials is crucial for the production of subjects, investigators, and the sponsor, and is mandated by regulatory agencies worldwide." (R.W. Johnson Protocol.) Nurse Mary Ellen Conforto, a Scirex supervisor and tenured employee, testified that she understood the need for the records to be accurate, and agreed that if a patient were released early, the records should have reflected the change. She explained, however, that it was common for nurses to fill out a patient's record ahead of time, in order to cut down on paperwork at the end of the day, and that the nurses in doing this would predict and record in advance the time that a patient would leave if he or she remained for the required eight hours. Still, she acknowledged that if a patient whose record had been so prepared then left earlier than anticipated, an addendum to that patient's record should have been prepared and submitted.

Despite these understandings, over the course of the four studies, at various times Scirex nurses released both rescued and unrescued patients early, that is, before eight hours had passed, but they submitted records for those patients that made it appear that the nurses had followed the protocols and observed them for the full eight hours. These discrepancies were discovered when Scirex received a tip from an ex-employee, following which it conducted an "audit" of test subjects that involved calling them and asking what time they were discharged. The audit showed that unrescued patients were often released early, sometimes after as little as one hour. One of the audits reported that "[a]ll of the unrescued subjects stated they were discharged from the unit before the completion of the 8-hour follow-up period, indicating follow-up periods of 1 to 7 hours. However, in all cases the nurses' notes indicated the subjects were discharged after 8 hours." (Compliance Audit Report, Algos Protocol.) As the FDA's clinical testing requirements are exacting and inflexible, the discrepancies between the actual release times and the recorded release times rendered the four studies worthless, and Scirex had to make financial amends to the pharmaceutical companies. Indeed, Scirex performed each study again without charge to its sponsor.

At the time of these events, Scirex had an insurance policy with Federal that had a limit of $280,000 and covered direct loss caused by any fraudulent or dishonest acts committed by its employees. It provided that:

The most we will pay for any loss under Blanket Employee Dishonesty for any loss caused by any employee whether acting alone or in collusion with others, either resulting from a single act or any number of acts, regardless of when those acts occurred during the period of this insurance or prior insurance, is the amount of loss, not to exceed the Limit of Insurance for Blanket Employee Dishonesty shown in the Declarations.

(Policy at 6, "Limits of Insurance.") A separate provision limited liability for multiple related acts, stating that "[a]ll losses resulting from an actual or attempted fraudulent or dishonest act or series of related acts at the premises ... whether committed by one or more persons will be deemed to be one occurrence or event." (Policy at 7.)

Scirex submitted a claim to Federal under this policy, contending that the actions of the nurses in sending patients home early, and submitting data sheets that falsely indicated that the patients had remained for the required eight hours, were "fraudulent or dishonest" and were therefore covered under the policy. It ultimately sought compensation for the loss of its investment in the four studies, an amount that included employee salaries, facility rentals, and other task-specific expenses. It claimed $185,000 on the R.W. Johnson study, $575,408 on the Forest Labs study, and a combined $473,679 on the two Algos studies, for a total of approximately $1.23 million.

In response to Scirex's claim, Federal's claim adjuster, Patricia Duffy, testified that "for the purposes of the discussion with the insured, I accepted the fact that what they were presenting to me seemed to be fraudulent — or seemed to be a dishonest act." (Duffy Dep. at 17.) She therefore requested that the claim be paid at policy limits. Upon receiving Duffy's request, however, an attorney at Federal's home office questioned whether Scirex's loss had been "direct." The policy does not define "direct" or "direct loss," and eventually Federal denied the claim, finding that Scirex's losses were more akin to ordinary expenses from a failed business venture than losses, such as fraudulently-claimed overtime, caused by employee dishonesty. It also asserted that even if there had been a direct loss, Scirex would have been entitled only to one policy limit of $280,000, rather than four, because the four ruined studies constituted only a single "occurrence or event." Scirex subsequently filed suit for the losses it incurred, up to the limit of $280,000, on each study.

At trial, Federal asserted for the first time that the nurses' actions had not been "fraudulent or dishonest." The District Court agreed, based on its finding that the Scirex nurses did not believe their conduct to be wrongful. The Court reasoned that "the words `fraudulent' and `dishonest' both focus on the intent of the actor, and connote intentional conduct by the actor as wrongful," Scirex, 2001 WL 1491133 at 3, 2001 U.S. Dist. LEXIS 19088 at *9, but the nurses acted on their "stubborn belief that [they] were right and that the drug companies were imposing unreasonable and unnecessary requirements." Id. 2001 WL 1491133 at *3, at *8-9. It reasoned that they "gained no personal benefit from their actions," although it acknowledged that by sending patients home early, they might have gained "a slight reduction in paperwork, yielding an earlier end to their workday." Id. 2001 WL 1491133 at 3, at *9. In summary, the Court found that while the nurses' actions might have been ill-considered, they did not believe they were being duplicitous or dishonest, and therefore their actions were not dishonest within the meaning of the policy. It accordingly found against Scirex. The District Court also stated that, although the nurses' actions ruined four studies, the losses all stemmed from a "series of related acts," and therefore only $280,000 would be recoverable if coverage were found to exist. Id. 2001 WL 1491133 at *5, at *14.

The District Court had jurisdiction under 28 U.S.C. § 1332. We have jurisdiction under 28 U.S.C. § 1291. Although the District Court made certain fact findings which would be subject to deferential review, primarily at issue on appeal are the District Court's legal conclusions that Federal's insurance policy does not cover the Scirex nurses' acts, and that if it does cover those acts, Scirex is entitled to only one payout for the four studies, over which we exercise plenary review.

A. Does Federal's Policy Cover the Nurses' Acts?

1. Were the nurses' acts dishonest?

Federal's policy covers "fraudulent or dishonest acts." We will assume that the nurses did not commit fraud in ordering patients' early discharge and failing to record those discharges accurately; aside from a very slight reduction in their workload, they did not benefit from their questionable conduct. The District Court noted this lack of fraud, but also found no dishonesty because the word "`dishonest' focus[es] upon the intent of the actor, and connote[s] intentional conduct perceived by the actor as wrongful." Scirex, 2001 WL 1491133 at *3, 2001 U.S. Dist. LEXIS 19088 at *9. "The evidence as a whole makes clear that Ms. Conforto and the other nurses honestly believed that they were substantially complying with the requirements of the protocols. In their view... there was no real need to keep the patients at the clinic when they wanted to go home early." Id. 2001 WL 1491133 at *3, at *8. Dishonesty, the Court concluded, must mean something more than an error in judgment.

In its brief, Federal enthusiastically endorses this view. If the evidence supports the notion that the nurses honestly believed they were substantially complying with the protocols' requirements, it asks, how can their acts be fairly termed "dishonest?" Federal points to many cases where courts have held that "willfulness and an intent to deceive must be present in order for an employee's actions to be dishonest." Rock Island Bank. v. Aetna Cas. & Sur. Co., 706 F.2d 219, 222 (7th Cir.1983); see also Jellico Grocery Co. v. Sun Indem. Co., 272 Ky. 276, 114 S.W.2d 83 (1938) (finding no dishonesty where an employee violated express instructions not to extend credit to a certain customer, to the employer's detriment, because there was no showing of intent); Couch on Insurance § 161:27 (stating that mere irregularities, mistakes, negligence, errors in judgment or incompetence, committed without intent to deceive, do not constitute fraud or dishonesty).1

Federal relies most heavily on Universal Credit Co. v. United States Guarantee Co., 321 Pa. 209, 183 A. 806 (1936). In that case, an insurer contracted to indemnify an employer for all direct pecuniary losses sustained by "any act or acts of fraud, dishonesty, larceny, embezzlement, forgery or wrongful extraction" committed by any of its employees. Id.

In order to ascertain whether any of the employer's vehicles had been sold without repayment of the amount loaned, an employee, Gregory, was charged with the duty of checking the cars at the dealer's place of business. He recorded the results of his inspection on a report that contained the printed statement: "I have personally seen and checked the auto numbers on the cars listed above, and certify that the information given is correct." In reality, Gregory had relied on the dealer for his information and had not personally checked to see whether particular cars were on hand. As a result, the dealer doublesold some cars and thereby suffered a loss. Id.

The Pennsylvania Supreme Court said, "[i]t seems too clear to require citation of authority to support it that negligence is not fraud, and acts resulting from mistake of judgment are not acts of fraud or dishonesty any more than acts done negligently," id. at 807, and it found no insurance coverage. Notably, however, the fidelity bond in Universal Credit Co. covered only "acts done for the purpose of harm or with a view to personal profit," id. at 806, qualifying language that does not exist in Federal's bond with Scirex.

Moreover, many cases expressly hold motive and intent irrelevant to the concept of dishonesty. In National Newark and Essex Bank v. American Ins. Co., 76 N.J. 64, 385 A.2d 1216 (1978), a bank manager made records that misstated the value of securities pledged as collateral for loans. When the borrower defaulted and the bank learned that the loans were inadequately secured, the bank filed a claim under a bond covering losses resulting from "dishonest or fraudulent acts" of employees. At the same time, however, the bank informed the insurer that there was no evidence of any "defalcation, mysterious disappearance, kiting operation or other purported [criminal] act." Id. at 1220. The insurer denied the claim based on this concession, on the grounds that the employee's acts were not dishonest or fraudulent. The New Jersey Supreme Court held that the employee's acts of "misrepresenting or failing to disclose important facts concerning the loans" were dishonest, id., and noted that, in a fidelity bond, the words "dishonest" and "fraudulent" "extend beyond criminal acts and are to be given a broad signification and taken most strongly against the surety company." Id. at 1222. It concluded that those words "encompass[] any acts which show a want of integrity or a breach of trust." Id.

Other courts are in accord. See, e.g., Mortgage Corp. of New Jersey v. Aetna Casualty & Sur. Co., 19 N.J. 30, 115 A.2d 43, 46 (1955) ("[W]here an employee ... is employed to perform a series of acts and to certify a set of facts on physical inspections, and that employee certifies to those facts, without having made the physical inspection ... he is unfaithful to his employer," and his lack of integrity "brings him squarely within the definition of `dishonest.'"); First National Bank of West Hamlin v. Maryland Casualty Co., 354 F. Supp. 189, 193 (S.D.W.Va.1973) (holding that an employee acts dishonestly when he certifies that automobiles are on the car dealer's premises when in fact he does not know if they are).

In this case, Scirex's nurses understood that they were to record their observations of test subjects every thirty minutes for eight hours, yet in many cases their records included "observations" for times when the patients were sitting at home. Even worse, these fictionalized records falsely implied that the patients had remained in the clinic for the full eight hours, and therefore that the tests had proceeded according to protocol. Thus, not only did the nurses fictionalize the records, they made it virtually impossible to discover the fictionalization until disclosure by the informant. Faced with such flagrant misrepresentation in a field characterized by strict adherence to procedure, we conclude that the nurses' conduct was clearly dishonest, as well as highly unfaithful. Our conclusion is buttressed by our practice of construing a policy's ambiguities against its drafter. See Medical Protective Company v. Watkins, 198 F.3d 100, 105 (3d Cir.1999).

Federal explains that it is common nursing practice to fill out observation reports ahead of time, then to change them if the actual observation does not fit the expected observation. (See Federal Br. at 17-18.) From that it argues that the nurses failed to change their pre-recorded observations because "they did not perceive an early discharge to be an adverse event that needed to be recorded," (id.), the implication being that they were negligent, not dishonest. We disagree, for, even if the nurses were merely negligent in deciding that an early discharge is a trivial event, they were dishonest in submitting records implying that no early discharges took place. Indeed, Federal relies upon Nurse Conforto's testimony that it is nursing practice to record observations ahead of time, yet she herself testified that she could think of no reason why the record of a patient who was released early should not note that. (Conforto Dep. at 93.6-9.) We therefore reverse the District Court's holding that the nurses' actions were not dishonest within the meaning of Federal's policy.

2. Were Scirex's losses direct?

Even though we conclude that the nurses' actions were dishonest, that does not end our inquiry because Federal's policy covers only "direct" losses. The District Court stated in dicta its belief that the losses were direct, but on appeal Federal seeks to affirm the denial of coverage on the alternate ground that the losses were indirect. See University of Maryland v. Peat Marwick Main & Co., 923 F.2d 265, 275 (3d Cir.1991) (Court of Appeals can affirm the judgment below on any ground, including a ground not relied upon in the decision below). Federal's policy covers only "direct loss caused by fraudulent or dishonest acts," and the loss must occur "to money, securities, or other property." It maintains that Scirex's losses are not covered for two reasons: because they are "ordinary and legitimate business expenses," not direct losses; and because they are not recoverable losses "to money, securities, or other property." We examine each contention in turn.

Federal's first argument is that Scirex's losses are indirect, likening them to ordinary operating expenses for a business venture that ultimately fails. It seeks conceptually to distinguish direct losses from indirect losses by means of example:

If the nurses and falsified their work to make it appear that they were entitled to overtime and Scirex had paid them $1,000 in fictitious overtime, then Scirex would have incurred a "direct loss" in that amount. Similarly, if the nurses had pilfered office supplies worth $100,000, then Scirex would have sustained a "direct loss" in that amount.

(Federal Br. at 36.) The implication is that covered losses must be ones that relate to the employees themselves, rather than ones that occur because an employee does his or her job negligently, dishonestly or improperly.

Here, according to Federal, Scirex seeks to recover what amounts to ordinary expenses. Although Scirex seeks to recover for the labor, equipment, and expertise that went into the studies, Federal contends that "Scirex was not `tricked' into incurring these expenses by the nurses' alleged misconduct. Insofar as they may be called `losses,' they are losses only because Scirex was unable to charge the sponsors for the studies on which they were incurred. That is an `indirect' loss." [Federal Br. at 36-37.] It cites several cases to support its position. See, e.g., Lynch Props., Inc. v. Potomac Ins. Co., 962 F. Supp. 956 (N.D.Tex.1996) (embezzlement of customer's funds held to be "direct" loss; embezzler's employer's obligation to replace those funds held "indirect" loss); Continental Bank, N.A. v. Aetna Cas. & Sur. Co., 164 Misc. 2d 885, 626 N.Y.S.2d 385, 387-88 (N.Y.Sup.Ct. 1995) (in policy containing "directly from" language, no coverage for insured whose employees caused "actual loss" to customers of insured).

Scirex counters that it "made a substantial investment in [the] studies, and up to the time of the employee falsifications, those studies were valuable. When the employees falsified data in the studies, however, those studies became worthless, and Scirex' investment was lost. This was a `direct loss' within the meaning of the policy." (Scirex Rep. Br. at 9.) It submits that our jurisprudence favors that interpretation.

In Jefferson Bank v. Progressive Casualty Ins. Co., 965 F.2d 1274 (3d Cir.1992), plaintiff bank sought indemnification under a loan created with the aid of an imposter notary, who affixed her invalid notarization to the mortgage, and then failed to record it. The customer subsequently granted other mortgages on the same property, at least one of which was recorded, and when the customer defaulted on all loans, the plaintiff bank was unable to take possession of the property. The defendant insurer argued that the bank's loss was not covered under the bond's coverage for losses "resulting directly from" fraudulent signatures, because the loss was caused not by the forged signature of the notary, but by the fact that the building was so heavily encumbered. We disagreed, holding that under Pennsylvania law, the "direct cause of a loss" does not have to be the "sole cause" or "immediate cause," but need only be a proximate or substantial cause:

[D]irect cause or immediate cause is a nebulous and largely indeterminate concept, and one that does not enjoy favor under Pennsylvania law. As we have suggested, Pennsylvania, consistent with general notions of proximate causation, requires that plaintiffs in negligence cases show substantiality, rather than immediacy, in order to demonstrate proximate cause.

Id. at 1281-82.

We believe that Jefferson Bank's proximate cause approach is proper here, and under that test, we conclude that the nurses' acts directly caused Scirex's losses. It is uncontested that their failure to follow protocol and their deceptive record-keeping singlehandedly rendered the studies worthless. We find Scirex's reasoning persuasive: it is in the business of producing products, pharmaceutical studies, that are tailored specifically to individual clients' needs. It produces a study only when commissioned by a particular sponsor, and for that reason, it has relatively little overhead or ongoing costs of doing business. Rather, as Scirex states, "the salaries of the employees who conducted the ruined studies, as well as the cost of office space and supplies allocated to the four studies, would not have been incurred if Scirex had not contracted to produce those studies." [Scirex Rep. Br. at 18.] Scirex's losses were directly tied to these studies, and by rendering those studies worthless, the nurses' behavior proximately, and therefore directly, caused Scirex's losses.

Federal also submits that, to recover under its policy, "the loss must occur to money, securities or other property," and that Scirex cannot recover under any of these categories. (Fed. Br. at 35.) It concludes that this is not a "loss to money" because money has not been stolen, and while it might (as Scirex argues) be a loss to property, that would lead to no recovery because the policy values property as the least of: (1) actual cash value of the property on the day the loss was discovered; (2) cost to repair; or (3) cost of replacing the property with material of like kind and quality. While Scirex would favor (3), the cost of replacing the studies, Federal submits that the policy would force choice (1), which would yield the value of the studies on the day the nurses' actions were discovered. Since the studies were at that point already ruined, Federal argues that they were valueless and therefore Scirex could recover nothing. Even if Scirex could recover under its desired option (3), Federal submits, it would recover only the cost of replacing a study of like kind, which is an imperfectly-conducted study.

We agree that this is not a "loss to money," because if it were, certainly it would be an indirect loss and therefore not covered under the policy. Scirex's theory is that it made a product, pharmaceutical studies, which it sold for money. By damaging the product, the nurses' actions directly caused a loss, but under this theory, it is a direct loss only to property, that is, the studies. Any loss "to money" would be indirect, because it would be a derivative loss caused by Scirex's inability to sell its damaged product. If Scirex is to show a direct loss, that loss must therefore be to property.

In our view, this is a loss "to property," and it is therefore subject, at least in theory, to the policy's property valuation rubric. But we disagree that this taxonomy leads to a recovery of zero, because it is possible to read the policy in two ways. The "strict literalism" method, which Federal supports, would force us to ascertain property value at the time the loss was discovered, which here would lead to a trivial recovery since the botched studies had no real value. We agree with Scirex that such a reading would essentially render the policy's property protections illusory, because it would effectively protect goods only in their damaged state — by definition, once losses are discovered, the damage has already been done. Such protection is no protection at all, and we will not reach that result when a reasonable alternative interpretation is present.

We conclude that the policy's property valuation language is not designed to limit recovery in this situation. By fixing the amount of recovery to "the actual cash value of the property on the day the loss was discovered," the policy seems to us to anticipate a loss to property whose market value fluctuates over time. In such situations, the policy language would preempt any debate over the proper time of valuation, but it is not useful where, as here, there is no real market for the product — the studies are produced at a fixed price specially for one consumer, and they are valueless to all others. Applying the policy language literally in this situation would render recovery almost impossible for any producer of custom-made products, a clearly counter-intuitive result that we are unwilling to reach.

We conclude that the nurses' actions caused a direct loss to property, and the policy's property valuation language does not limit Scirex's recovery.

B. Did the Nurses' Actions Cause a Single Loss or Four?

Scirex claims that the nurses' repeated failure to follow protocol and accurately record their observations ruined four enormous studies valued at $185,000, $575,000, $317,000, and $156,000 respectively. It seeks recovery up to the $280,000 policy limit for each study, for a total of $880,786. Federal argues that the entire recovery, if there is one, should be limited to $280,000. The District Court in dicta concluded that the various instances of dishonesty were sufficiently related to constitute one occurrence or event, and limited the recovery to $280,000. We agree.

The policy's Limits of Liability Clause reads:

The most we will pay for any loss under Blanket Employee dishonesty for any loss caused by an employee whether acting alone or in collusion with others, either resulting from a single act or any number of acts, regardless of when those acts occurred during the period of this insurance or prior insurance, is the amount of loss, not to exceed the Limit of Insurance for Blanket Employee Dishonesty shown in the Declarations.

* * *

All losses resulting from an actual or attempted fraudulent or dishonest act or series of related acts at the premises ... whether committed by one or more persons will be deemed to be one occurrence or event.

Stipulation ¶ 35(c) (emphasis added).

Scirex argues that the key word in the first passage is "any," which can be interpreted to mean "each" as well as "all." It further submits that the "each" interpretation is more reasonable than the "all" interpretation because the policy uses the singular "loss" rather than the plural "losses." "Each loss," it concludes, is a more natural reading than "all loss," and any ambiguity should be resolved against Federal. (Scirex Br. at 23.) Regarding the second passage, Scirex suggests that its purpose is merely to define what constitutes "one occurrence or event," and it points out that no language suggests that Federal will pay only one policy limit per occurrence or event.

Reading the policy language as a whole, however, we are satisfied that it limits Federal's liability to $280,000 for an event or series of related events. Although it does not say so explicitly, the accepted purpose of defining "an occurrence or event" is to limit liability, and in the insurance industry "occurrence" is commonly understood to mean all loss caused by a single act or related events. See, e.g., Couch on Insurance, § 160:61; see also Appalachian Ins. Co. v. Liberty Mut. Ins. Co., 676 F.2d 56, 61 (3d Cir.1982) (if there is but one cause for all of the losses, they are part of a single occurrence). Scirex's interpretation, which would make possible a separate recovery for each loss even if those losses are part of the same occurrence, would lead us to grant a separate recovery for each forged check passed as part of an employee's forgery scheme, a result that has been squarely rejected. See, e.g., Business Interiors, Inc. v. Aetna Cas. & Sur. Co., 751 F.2d 361 (10th Cir.1984) (concluding that the employee's fraudulent acts constituted a single loss for policy purposes).

The District Court found that, "as to all of the four studies in question, plaintiff's losses resulted from a `series of related acts': The same nurses were involved in all of the alleged wrongdoing, they acted in concert, and all of the alleged wrongful acts constituted a series of related acts." Scirex, 2001 WL 1491133 at *5, 2001 U.S. Dist. LEXIS 19088 at *14. This finding is not clearly erroneous, for the nurses themselves did not seem to distinguish among the four studies in terms of their responsibilities. We therefore conclude that their conduct caused a single loss. Federal's policy limits its liability to $280,000 per loss, so that is the amount to which Scirex is entitled for its ruined studies.

For the foregoing reasons the judgment of the District Court will be reversed and the case remanded with directions to enter judgment for Scirex for $280,000, parties to bear the costs equally.