This appeal, from a grant of summary judgment in a diversity case governed by Pennsylvania law, requires us to construe a Banker's Blanket Bond, a policy that insures against bank losses resulting from various types of criminal activity. Plaintiff-appellant Jefferson Bank ("Jefferson") brought suit on a Banker's Blanket Bond against its insurer, Progressive Casualty Insurance Company ("Progressive"), to recover for the loss of $600,000, which Jefferson had loaned to an unscrupulous lawyer engaged in a scheme to defraud Jefferson and several other Philadelphia banks. Jefferson attempted to secure repayment of the loan by taking a first mortgage on a residential property owned by the lawyer. That mortgage, however, ultimately proved worthless to Jefferson for two reasons.

First, unbeknownst to Jefferson, the mortgage was acknowledged not by a notary public but by an imposter posing as a notary. This deficient acknowledgement rendered the mortgage unrecordable and therefore incapable of achieving priority over third parties under Pennsylvania law. Second, also unbeknownst to Jefferson, its mortgage was never left for recording. This failure to record proved fatal to Jefferson because the lawyer-mortgagee was insolvent and had mortgaged the same property to a number of other lenders, including one whose mortgage was effectively recorded before Jefferson's.

The district court held that the policy did not cover the loss and therefore granted summary judgment for Progressive. Although we agree with the district court that coverage was clearly not provided under one portion of the insurance policy, we believe that Jefferson was potentially covered under another portion and that there is a genuine issue of material fact as to the existence of such coverage. We will therefore reverse the judgment of the district court and remand for further proceedings.

Late in 1988, Jefferson agreed to loan $600,000 to Frederick Shapiro, a lawyer with whom it had previous financial dealings. As security for the loan, Shapiro agreed to grant Jefferson a first mortgage lien on his real property located at 1708 Locust Street, Philadelphia, Pennsylvania, and to obtain title insurance on the property by the time of the closing.

On December 16, 1988, Shapiro met with Jefferson's agent, Charles Kerrigan, to close the loan. Also attending the closing was a woman who claimed to be "Sandra Stevenson." Shapiro represented to Kerrigan that "Stevenson" was both a notary and a vice president of Sterling Abstract Corporation, which, Shapiro claimed, was an authorized agent for Ticor Title Insurance Company ("Ticor"). Shapiro presented Kerrigan with a supposedly current title commitment issued by Ticor on the Locust Street property. During the closing, "Stevenson" acknowledged the mortgage and notarized it with a notary's seal and stamp. Kerrigan witnessed Shapiro's signature on the mortgage. Shapiro and Kerrigan agreed that "Stevenson," acting as a representative of Sterling Abstract, would have the mortgage recorded.

Unfortunately for Jefferson, Shapiro, using similar schemes and the same real estate as security, also borrowed money from six or seven other banks. Shapiro secured one loan before the Jefferson loan and the remainder of the loans afterward. Ultimately, he found himself in such a hopeless financial situation that he surrendered to the U.S. Attorney and confessed. The U.S. Attorney began contacting the banks and, in August 1989, informed Jefferson that it had been one of the victims of Shapiro's scheme.

Jefferson soon discovered the extent of the fraud perpetrated against it by Shapiro and "Stevenson." The Ticor title commitment presented at closing was counterfeit. To create this document, Shapiro had altered an authentic Ticor title commitment that he had obtained in connection with an earlier real estate transaction. In addition, Sterling Abstract, a creation of Shapiro's, was not an authentic title insurer. "Sandra Stevenson" was an imposter. She held no position with either Sterling Abstract or Ticor and was not an authorized notary public. Moreover, she never recorded the mortgage that was executed at the closing.

"Stevenson's" invalid notarization and failure to record proved particularly damaging for Jefferson because on December 15, 1988, one day before the Jefferson closing, Shapiro had borrowed money from Royal Bank. In exchange for that loan, Royal Bank apparently had obtained a validly notarized first mortgage on Shapiro's property at 1708 Locust Street.1 Under Pennsylvania law, the Royal Bank purchase-money mortgage would have had priority over other liens from the time it was delivered to the mortgagee, if it was recorded within ten days after its date. 42 Pa.Cons.Stat.Ann. § 8141(1) (Purdon 1982).2

The Royal Bank mortgage was not recorded, however, until January 26, 1989, forty-two days after the Jefferson closing. Thus, under section 8141(1), Royal's priority did not begin until the date of actual recording, January 27, 1989. Because the Royal Bank mortgage was not promptly recorded, had the Jefferson mortgage on the Locust Street property been properly notarized and recorded between December 16, 1988 and January 26, 1989, it would have had priority, and Jefferson would have been adequately secured against the loss it now claims. However, the mortgage was not properly notarized and recorded, and, by the time Jefferson discovered the fraud, the Royal Bank mortgage was recorded and had first priority as a lien on the property.3 Because the value of the Royal mortgage equalled the equity Shapiro had in the Locust Street property, Jefferson would have been unable to recover its loan through foreclosure of its subordinate lien.

Realizing that it could not recover on the Locust Street property or from Shapiro's other assets, Jefferson filed a claim under the Banker's Blanket Bond issued by Progressive. Progressive denied the claim, and Jefferson filed the present complaint. On November 19, 1990, the district court denied Jefferson's motion for summary judgment while granting Progressive's motion. Jefferson moved for reconsideration, which the district court denied without opinion. This appeal followed. The district court had subject matter jurisdiction under 28 U.S.C. § 1332, and we have appellate jurisdiction under 28 U.S.C. § 1291.II. THE INSURANCE POLICY

The Banker's Blanket Bond covers Jefferson for losses from embezzlement, robbery, larceny, false pretenses, forgery, receipt of counterfeit money and other criminal acts. Within these broad categories of covered loss, however, certain types of loss are excluded from coverage. One such exclusion, the Loan Exclusion Clause, limits recovery for losses resulting from defaulted loans:

...

(e) loss resulting directly or indirectly from the complete or partial non-payment of, or default upon, any Loan or transaction involving the Insured as a lender or borrower, or extension of credit, including the purchase, discounting or other acquisition of false or genuine accounts, invoices, notes, agreements or Evidences of Debt, whether such Loan, transaction or extension was procured in good faith or through trick, artifice, fraud or false pretenses, except when covered under Insuring Agreements (A), (D) or (E);

Pursuant to various exceptions to the Loan Exclusion Clause, however, the Bond does provide coverage for certain types of loan losses. Two exceptions are relevant in this case. First, the Fraudulent Mortgage Rider to the Bond provides an exception to the Loan Exclusion Clause for loan losses resulting from mortgages or "like instruments" that contain a fraudulently obtained signature:

Loss through the Insured's having in good faith and in the course of business in connection with any loan, accepted or received or acted upon the faith of any real property mortgages, real property deeds of trust or like instruments pertaining to realty or assignments of such mortgages, deeds of trust or instruments which prove to have been defective by reason of the signature thereon of any person having been obtained through trick, artifice, fraud or false pretenses or the signature on the recorded deed conveying such real property to the mortgagor or grantor of such mortgage or deed of trust having been obtained by or on behalf of such mortgagor or grantor through trick, artifice, fraud or false pretenses.

Second, Insuring Agreement E provides coverage for loan losses which result because one or more of certain enumerated loan documents used in transacting the mortgage are stolen or contain forgeries or alterations. Specifically, Insuring Agreement E provides coverage for:

....

(E) [l]oss resulting directly from the insured having, in good faith, for its own account or for the account of others,

(1) acquired, sold, or given value, extended credit or assumed liability, on the faith of, any original

(a) Certificated Security,

(b) Document of Title,

(c) deed, mortgage or other instrument conveying title to, or discharging a lien upon, real property,

(d) Certificate of Origin or Title,

(e) Evidence of Debt,

(f) corporate, partnership or personal Guarantee,

(g) Security Agreement,

(h) Instruction to a Federal Reserve Bank of the United States, or

(i) Statement of Uncertificated Security of any Federal Reserve Bank of the United States

which

(i) bears a signature of any maker, drawer, issuer, endorser, assignor, lessee, transfer agent, registrar, acceptor, surety, guarantor, or of any person signing in any other capacity which is a Forgery, or

(ii) is altered, or

(iii) is lost or stolen ...

In short, the policy's language makes clear by its terms that the Banker's Blanket Bond does not cover loan losses suffered by an insured bank unless (1) the loss resulted from the bank's having acted in good faith in reliance on a mortgage, or like instrument pertaining to realty, that was defective because it bore a signature that had been obtained through fraud or (2) the loss resulted "directly" from the bank's extending credit in good faith in reliance on a mortgage that bore a forged signature. We consider whether Jefferson is covered for this loss under either of these exceptions.

When faced with an appeal from an order granting summary judgment, this court exercises plenary review. Nathanson v. Medical College of Pennsylvania, 926 F.2d 1368, 1380 (3d Cir.1991). Our review entails the same consideration that a district court gives to a ruling on a motion for summary judgment under Rule 56 of the Federal Rules of Civil Procedure: (1) determination whether the record as it stands reveals any genuine issue of material fact, (2) resolution of all such issues in favor of the non-movant, and (3) determination whether the movant is then entitled to judgment as a matter of law. See id.; First Jersey Nat'l Bank v. Dome Petroleum Ltd., 723 F.2d 335, 338 (3d Cir.1983).4

As we explained above, recovery by Jefferson under the Fraudulent Mortgage Rider requires that the loss have resulted from the Bank's acting in good faith reliance on a mortgage or like instrument pertaining to real estate, which instrument was defective because it bore a signature which had been obtained through fraud. In other words, the notary signature, the potential source of defect on the mortgage, or the various signatures on the title commitment would have to have been induced by fraud to allow Jefferson coverage under the Fraudulent Mortgage Rider.5 Jefferson argues that the signatures of "Stevenson" on both the mortgage and the Ticor title commitment were fraudulently induced.

With respect to the mortgage, Jefferson contends that the forged signature of the notary is grounds for coverage because the mortgage is "defective by reason of" this signature "having been obtained through trick, artifice, fraud or false pretenses." In formulating its argument, Jefferson emphasizes that the legal effect of a mortgage is to protect against third-party claims to the same property. Jefferson contends that the mortgage in this case is defective under Pennsylvania law due to the notary's fraudulent signature and that this defect entitles it to recover under the Fraudulent Mortgage Rider.

We agree that the mortgage is defective by virtue of "Stevenson's" invalid signature, a point we will return to at length. Mere defect is not sufficient, however. The Fraudulent Mortgage Rider requires that Jefferson, in entering the mortgage, relied on a signature that, in turn, was induced by fraud. That is not the case here. Progressive correctly contends that the signature of the notary "Stevenson" on this mortgage was not obtained by fraud; rather, it was part of the fraud. The parties agree that the "Stevenson" imposter knew exactly what she was doing when she forged the signature. Her signature therefore was not induced by fraud. Hence, we affirm the district court's holding that the defective mortgage is not covered under the Fraudulent Mortgage Rider.

Because the signature of "Stevenson" was not induced by fraud, neither can her signature on the title commitment serve as a basis for coverage under the Fraudulent Mortgage Rider. Jefferson also urges, however, that it is covered under the Fraudulent Mortgage Rider because of fraudulently obtained signatures of three Ticor officers on the title commitment. The district court did note the existence of factual issues regarding whether the title commitment bore the signatures of three officers of Ticor and whether these signatures were obtained by trick, artifice, fraud or false pretenses. The court concluded, however, that it was unnecessary to resolve these issues because, regardless of which party's version of the facts were correct, Jefferson could not possibly depend upon the title commitment to recover because the title commitment was not a "mortgage, deed of trust or like instrument pertaining to realty," as required by the language of the Fraudulent Mortgage Rider. The court, relying on the principle of ejusdem generis,6 held that the phrase "like instruments pertaining to realty" meant "mortgages and mortgage-like documents, and [could] not reasonably be stretched to mean the title insurance papers in this case."

On appeal, Jefferson urges that the phrase "mortgages, real property deeds of trust or like instruments pertaining to realty" is not ambiguous and that it was therefore inappropriate for the district court to rely on the principle of ejusdem generis. See In re Donaldson's Estate, 362 Pa. 357, 67 A.2d 88 (1949) (relying on ejusdem generis to construe ambiguous phrase in a will). Jefferson apparently believes that the phrase "like" unambiguously means "as important as." That construction would place title commitments in the category of "like instruments pertaining to realty" because a title commitment is, next to a mortgage, the most important document in a loan transaction secured by real property. We reject that reading.

The term "like instrument" necessarily refers back to its immediate antecedents, "mortgages" and "deeds of trust."7 The word "like" means, in every day use, "similar" or "equivalent." In other words, a "like" instrument serves the purpose or role of another instrument. The district court was thus correct to state that "like instruments" here means "mortgage-like" instruments.8

Jefferson suggests that interpreting "like instruments pertaining to realty" to mean "mortgage-like instruments" will render the language of Insuring Agreement E superfluous. Specifically, Jefferson contends that restricting this phrase to mean "mortgages and mortgage-like instruments" would be redundant with Insuring Agreement E, which, Jefferson contends, already covers losses resulting from mortgages and mortgage-like instruments. This argument, however, completely ignores that the two provisions are drawn to cover not only specific categories of documents from which loss may result but also specific causes of loss. For example, Insuring Agreement E, as it pertains to mortgages, covers only those losses directly resulting from credit extended in reliance on a mortgage which bore a forged signature, or was altered or stolen. Jefferson apparently wanted greater coverage than this for losses from mortgage transactions, so it contracted to include the Fraudulent Mortgage Rider in its Banker's Blanket Bond. The Rider adds coverage to include loan losses caused by fraudulently induced--although not forged--signatures on mortgages. The Rider thus would cover bank losses resulting when the seller of real property fraudulently induces the mortgagor to purchase the property and sign a mortgage by promising that the mortgage will never be enforced, by misrepresenting to the mortgagor that the mortgage covers a different piece of property, or by telling the mortgagor that the document being signed was a contract to purchase rather than a mortgage.

In each of these examples, the mortgage bears a fraudulently induced signature but not a forged signature; yet in each case, the bank may find itself with a defective, unenforceable mortgage. To protect against such losses, Jefferson obtained coverage under the Fraudulent Mortgage Rider to the Banker's Blanket Bond. It is not necessary to twist the interpretation of the Rider to cover additional types of documents, not specified in the policy, in order to prevent the Rider from being redundant. Thus, Stevenson's signature on the title commitment cannot serve as a basis for coverage under the Fraudulent Mortgage Rider.

As we suggested above, recovery by Jefferson under the provisions of Insuring Agreement E depends upon Jefferson's being able to demonstrate that its loss resulted directly from the Bank's extending credit in good faith reliance on a mortgage that bore a forged signature. The district court acknowledged that the mortgage contained a forgery, namely the signature of the notary, but denied coverage because, in the court's view, the loss did not "result directly from" this defect. The court reasoned:

The notary's signature is on the mortgage to witness the signature of Shapiro. Shapiro's signature is valid.... It is not the forged signature of the notary that caused the loss to the bank, it is the fact that so much money was lent with this building as security that the mortgage was worthless.

On appeal, Jefferson contends that the district court applied an inappropriate legal standard of causation and that, regardless of the standard applied, the court erred in concluding that the forged signature did not cause Jefferson's loss.

To decide whether the fraudulent notary's signature "directly" caused Jefferson's loss, we must first ascertain the correct definition of "resulting directly from" under Insuring Agreement E. We must then decide whether, under that definition, Jefferson suffered the loss as a result of the forged signature, as a result of the worthlessness of the security, as a result of the failure to record the mortgage, or as a result of some combination of these factors.

2. Meaning of "Resulting Directly From"

The district court construed "resulting directly from" in Insuring Agreement E strictly, concluding that the phrase means that the loss must be "directly caused" by the forged signature.9 Although the precise contours of the causation standard applied by the district court are unclear, it appears that it required the plaintiff not only to prove proximate cause, but also some additional closeness in space and time between the loss and the cause of the loss. In Jefferson's view, this requirement of "direct causation" is legally incorrect. Jefferson submits that the correct standard is "broader than proximate cause,"10 or, alternatively, even if Insuring Agreement E's standard is proximate cause, that the policy only requires proof--satisfied on these facts--that an enumerated event was a substantial contributing factor in bringing about the loss.

Progressive, in contrast, argues for a stricter standard, apparently the same standard of direct causation endorsed by the district court. Nevertheless, Progressive contends that, even applying the proximate cause standard, Insuring Agreement E did not cover the loss sustained by Jefferson in this case because the forged signature of "Stevenson" was not a proximate cause of Jefferson's loss.

We begin our analysis by acknowledging that the phrase "resulting directly from" in the policy does suggest a stricter standard of causation than mere "proximate cause." Under Pennsylvania tort law, a cause is proximate if it is merely a "substantial cause" of the harm. See Whitner v. Von Hintz, 437 Pa. 448, 263 A.2d 889, 893-94 (1970). Arguably, the words "resulting directly from" suggest a requirement beyond that the cause be substantial, for the words imply that the loss must flow "immediately," either in time or space, from the forged signature.11 An analysis of Insuring Agreement E in light of Pennsylvania law persuades us, however, that conventional proximate cause is indeed the correct standard and that requiring "immediacy" is inappropriate. We reach this conclusion for two reasons.

First, we construe the policy as providing coverage when the forgery proximately causes the loss because Pennsylvania courts have construed "direct cause of a loss" in the language of an insurance policy as meaning "proximate cause of a loss." See, for example, Marks v. Lumbermen's Ins. Co., 160 Pa.Super. 66, 49 A.2d 855 (1946) (equating "direct loss by windstorm" with "proximately caused by windstorm."). See also Lorio v. Aetna Insurance, 255 La. 721, 232 So.2d 490, 493 (1970); Lipshultz v. General Insurance, 256 Minn. 7, 96 N.W.2d 880 (1959). See generally Annotation, 65 ALR 3d 1128 (1975) (noting that courts have equated "direct result" with "proximate cause of loss").12 Based on this precedent, we believe that an accurate prediction of Pennsylvania law is that "resulting directly from" means "proximately caused by" under Insuring Agreement E.

Additionally, "direct 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. See Hamil v. Bashline, 481 Pa. 256, 392 A.2d 1280, 1284 (1978). Not only has Pennsylvania failed to adopt the immediate cause standard in tort, but commentators have demonstrated the nearly universal rejection and unworkability of the standard:

It is of course obvious that if a defendant sets a fire which burns the plaintiff's house, no court in the world will deny liability upon the ground that the fire, rather than the defendant's act, was the nearest, or next cause of the destruction of the house. It is everywhere recognized that there must be some causal degree of progression into the causal sequence. There may have been considerable confusion in the distant past, but the question is certainly no longer open.

Prosser on Torts, 276-77.13 Given the difficulty and confusion that results from applying a "nearest cause" or "immediate cause" standard, we do not believe that the parties intended to contract for it. Instead, we believe that in this contract "resulting directly from" means "proximately caused by."

Having decided that Jefferson is covered under Insuring Agreement E when a forged signature on a mortgage proximately causes loss, we must determine whether the record would support the conclusion that the forged notary's signature proximately caused Jefferson's loss.14 Even assuming that the proximate cause standard controls, the parties dispute the source of Jefferson's loss and whether it is covered. According to Jefferson, the forged notary's signature on the mortgage was a substantial cause of the loss because the forgery made the mortgage unrecordable and hence prevented Jefferson from protecting itself against other liens on the property. Jefferson's argument depends in part on the predicate argument that the loss took place at the time of the transaction, that is, that Jefferson sustained a loss immediately upon receipt of a mortgage with a forged acknowledgment.

Progressive, in contrast, agrees with the district court that the cause of Jefferson's loss was not the forged notary signature but the fact that the property serving as security was worthless from the outset due to the existence of the prior mortgage lien. Progressive, in other words, argues that even if the notary's signature had not been forged, the bank still would have suffered the same loss due to the overall nature of Shapiro's scheme. Progressive argues that Royal's mortgage trumped Jefferson's and that Jefferson's security was worthless regardless whether or not it was recordable.

Neither of these versions of events correctly describes the chain of causation leading to Jefferson's loss. Each party proffers a different--and, in our view, incomplete--explanation of what "caused" the loss, largely because both parties misapprehend when Jefferson sustained the loss. Therefore, each offers an unpersuasive explanation of the cause(s) of the loss.

Jefferson is incorrect as a matter of law to assert that it sustained a loss on December 16, 1988, the day of the loan closing. We concede that the mortgage containing the fraudulent notary's signature executed on that day could not have been properly recorded as it then existed and thus could not have served to ensure the priority of Jefferson's lien against other lienors. As between Jefferson and Shapiro, however, the mortgage was perfectly valid, and no other recorded liens existed until January 27, 1989. Moreover, during the interim between December 16, 1988, and January 27, 1989, Jefferson might have taken various steps to achieve priority over other liens that would have given Jefferson's security value.

First, Jefferson might have attempted to have the mortgage containing Shapiro's signature properly acknowledged. Because Shapiro's signature was non-fraudulent, all that was required to make the mortgage recordable was that the acknowledgement be validated. Jefferson might theoretically have corrected the defective acknowledgement before January 27, 1989, either by having Shapiro (the grantor) acknowledge the mortgage in the presence of a genuine notary, see 21 Pa.Stat.Ann. § 42 (Purdon 1955 & 1991 Supp), or by seeking a decree in equity reforming the defective acknowledgement, see 21 Pa.Stat.Ann. § 279 (Purdon 1955 & 1991 Supp), which would allow the mortgage to be properly acknowledged. There is, of course, no guarantee that Jefferson would have been successful in obtaining an equitable decree validating the acknowledgement. The mere possibility of such a decree under Pennsylvania law persuades us, however, that the mortgage had potential value between December 16, 1988 and January 27, 1989.

Second, Jefferson could have obtained a judgment against Shapiro between December 16, 1988 and January 27, 1989 because, under the terms of the mortgage, Shapiro was in immediate default. Shapiro had provided a covenant in the mortgage that "the Property is unencumbered, except for encumbrances of record." That was, of course, untrue, as the property was encumbered by the then-unrecorded Royal mortgage. Because Shapiro had breached that covenant, Jefferson was entitled under the mortgage to foreclose almost immediately. Had Jefferson done so before January 27, 1989, its judgment would have had priority over the Royal mortgage. See 42 Pa.Cons.Stat.Ann. § 8141(4) (Purdon 1982).15

We recognize that both of these methods for giving the Jefferson mortgage value are theoretical. Jefferson never, in fact, received any notice that would have given it reason to seek valid acknowledgment of a mortgage it already believed was validly acknowledged. Nor did it have any reason to know that the property was already encumbered and that Shapiro had breached the mortgage's covenant of non-encumbrance. Nonetheless, our concern is determining when the Jefferson mortgage lost value, and in order to ascertain the date of loss, we must consider all of the possibilities, no matter how theoretical, that would have given the Jefferson mortgage value. Because Jefferson either could have reformed the acknowledgement and properly recorded or secured a judgment prior to January 27, 1989 with priority against Royal, the mortgage theoretically might still have had value for Jefferson. We conclude therefore the loss did not actually occur until January 27, 1989, when Royal achieved priority by recording; at all events, we cannot say as a matter of law that it did occur before that date.

Just as Jefferson's contention that the loss occurred at the date of the signing is erroneous, so too is Progressive's argument that Jefferson gave the mortgage in exchange for worthless security even before "Stevenson" affixed her fraudulent signature to the mortgage. As we have explained, the mortgage had potential value for Jefferson until Royal's January 27, 1989 recording, which rendered the security worthless. Before that date, however, Jefferson could have obtained judgment on the mortgage or could have taken steps to make it recordable. Again, although the mortgage was not all that Jefferson believed it to be, it was far from completely valueless until January 27, 1989.16

Having clarified that the loss actually occurred upon Royal's recording of its mortgage, and that Jefferson's security lost all possible value only at that time, we consider what factors proximately caused the loss. Consistent with basic principles of tort law, the Pennsylvania Supreme Court has long held that multiple proximate causes may lead to injury:

Proximate cause is a term of art, and may be established by evidence that a defendant's negligent act or failure to act was a substantial factor in bringing about the harm inflicted upon a plaintiff. Pennsylvania law has long recognized that this substantial factor need not be ... the only factor....

Jones v. Montefiore Hospital, 494 Pa. 410, 431 A.2d 920, 923 (1981).

We must decide which factors may be substantial in this case.

Initially, the failure of anyone to record the Jefferson mortgage arguably deprived it of priority and caused the loss. Obviously, but for the failure to record a properly acknowledged mortgage prior to January 27, 1989, Jefferson would have enjoyed priority over Royal, and the failure to record can fairly be said to be a proximate cause of the loss. Additionally, Shapiro's overarching scheme to defraud Jefferson and other banks was undoubtedly a substantial factor causing loss.

We must, however, also decide whether the forged notary signature itself played a significant enough role in Jefferson's loss that it might be considered a proximate cause of the loss. Jefferson presses the theory that the unrecordability of the mortgage in its existing form was a substantial factor causing the loss. We believe that there is much to commend this argument.

Under Pennsylvania law, a non-fraudulent acknowledgement is the sine qua non of an effective mortgage.17 Without such an acknowledgement, taken before an authorized notary, the mortgage cannot be validly recorded and cannot provide security against third party claims. See generally 21 Pa.Stat.Ann. §§ 291.3 and 621 (Purdon 1955 and 1991 Supp.). See also John Makdisi, ed., Ladner on Conveyancing in Pennsylvania § 18.08 at 8-9 (Bisel 1979). This principle holds true even if the fraudulently acknowledged mortgage is physically recorded by the Recorder of Deeds. See Conveyancing in Pennsylvania, § 18.08 at 8-9. See also Lancaster v. Flowers, 198 Pa. 614, 48 A. 896 (1901). In the face of the fraudulent acknowledgement by "Stevenson," the Jefferson mortgage, as it existed on December 16, 1988, was not properly recordable.

We have difficulty, however, viewing the unrecordability of the mortgage as a proximate cause of the loss on these facts. As Jefferson concedes, no effort was ever made to record the Jefferson mortgage. Thus, even had the notary's signature been genuine and the acknowledgement valid, Jefferson would still have suffered the loss because its mortgage was not recorded prior to Royal's recording of its mortgage. Had Jefferson attempted to record, there is no doubt that the fraudulent signature would have been a proximate cause of loss. But Jefferson did not even attempt to record and that failure would appear to be a superseding cause of Jefferson's loss. Therefore, as a matter of law, the unrecordability of the mortgage was not a proximate cause of Jefferson's loss.

Although we reject the theory that the unrecordability of the mortgage substantially contributed to Jefferson's loss, the fraudulent signature nevertheless may have been a substantial factor causing Jefferson's loss for a different reason. The presence of the notary's signature on the mortgage, although seemingly a small part of Shapiro's elaborate scheme to defraud Jefferson, arguably induced Jefferson in part to enter the transaction. There can be no doubt that the signature was at least a cause-in-fact of Jefferson's loss, because absent the notary's signature, Jefferson, which knew that a valid acknowledgment was necessary to protect the security against the claims of third parties, would not have entered the transaction. See generally Whitner 263 A.2d at 893 (detailing Pennsylvania's "but for" causation rule).

The closer question is whether the fraudulent signature was a substantial factor in causing the loss. At first glance, the significance of the notary's signature appears to pale in comparison to the other elements of the fraud scheme. Nevertheless, as we have pointed out, Jefferson, as a sophisticated bank, was undoubtedly fully familiar with the legal effect of a notary's acknowledgment and signature, and would have refused to enter the transaction had not an individual purporting to be a notary signed. The importance of the signature to Shapiro's scheme is evidenced by the very fact that Shapiro brought the imposter "Stevenson" to the closing and duped Jefferson's representative into believing Jefferson had the normal legal protections that accompany a valid acknowledgment.

We cannot say, as a matter of law, whether the signature was a proximate cause of the loss. In our view, there is a genuine issue of fact as to whether it was, and that decision should be left for a jury. Allowing the jury to decide the issue of proximate cause is consistent with Pennsylvania's substantive law. See Ford v. Jeffries, 474 Pa. 588, 379 A.2d 111, 114 (1977) ("The determination of whether the conduct of the defendant was a substantial cause or an insignificant cause of plaintiff's harm should not be taken from the jury if the jury may reasonably differ as to whether the conduct of the defendant was a substantial cause or an insignificant cause.").

For the foregoing reasons, the judgment of the district court will be reversed and the case will be remanded for further proceedings consistent with this opinion. The parties shall bear their own costs on this appeal.

I disagree with the majority's conclusion in Section III C of its opinion that the loss here may be covered by Insuring Agreement E. First of all, I do not agree that the security for Jefferson's mortgage had value on December 16, 1988, the day of the loan closing. Under Pennsylvania law, the Royal Bank purchase money mortgage would have had priority over other liens from the time it was delivered to the mortgagee, if it was recorded within ten days after its date; otherwise, it would have priority from the time it was left for record. 42 Pa.C.S.A. § 8141(1). Because of the priority of Royal Bank's mortgage until December 25, 1988, ten days after it was executed, Jefferson's mortgage was not a valid lien on 1708 Locust Street during this period. For this reason, I find the cases of Reliance Ins. Col. v. Capital Bancshares, 685 F.Supp. 148 (N.D.Tex.1988), aff'd 912 F.2d 756 (5th Cir.1990), Liberty National Bank v. Aetna Life & Casualty Co., 568 F.Supp. 860 (D.N.J.1983) and American Banking Corp. v. Flota Mercante Grancolombiana, S.A., 752 F.Supp. 83 (S.D.N.Y.1990), aff'd on other grounds, 925 F.2d 603 (2d Cir.1991), to be on point.

A further problem with the majority's analysis of the cause of the loss lies in the necessary interaction under the language of the policy between the Section 2(e) Loan Exclusion Clause and the Insuring Agreement E provision for loss resulting from forgery. Under the policy, most losses resulting from non-payment of a loan obtained through fraud or false pretenses are not covered. Worthlessness of the security provided for a loan is not a basis for coverage under any of the policy sections. Only if Jefferson's loss resulted directly from its having in good faith extended credit on the faith of a mortgage that bore a forged signature would the Insuring Agreement E exception come into play. As the courts concluded in the cases cited above, Reliance Ins. Col. v. Capital Bancshares, Liberty National Bank v. Aetna Life & Casualty Co., and American Banking Corp. v. Flota Mercante Grancolombiana, S.A., I also conclude as a matter of law that the bank extended credit on the faith of the value of the security and that the authenticity of the notary's signature was of minimal, if any, significance, in Jefferson's decision to make the loan. Similarly, the loss was caused by the worthlessness of that security and not by the bank's inability to record a mortgage which the bank itself never presented for recording.

If someone, not the owner of the property at 1708 Locust Street, had forged the owner's name as mortgagor, Section E would clearly cover the loss. Credit would have been extended on the faith of that signature, which was forged. However, we must conclude that the express limitation of the language of Insuring Agreement E should not be extended to a fraudulent scheme such as the present one where the validity or not of the notary's signature was in effect immaterial to the success of the scheme and where the loss would have been caused by the worthlessness of the security, whether or not the notary's signature was valid.

The mortgagor is the person who can bind the property with the lien; the notary cannot--she can do no more than verify the signature of the mortgagor. If we consider then a case in which the property has value, the mortgagee needs to have the mortgage signed by a person having a legal interest in the property; if the owner's signature is a forgery, the mortgagee gets no interest in the valuable security. I interpret the policy language in Insuring Agreement E to protect against such a risk; namely, where someone other than the person having legal right to encumber the property forges that person's signature.

The majority questions whether the bank would have made the loan without a valid notary signature on the mortgage. I agree that it would not have, but I don't find that the lack of a signature was the cause of the loss. The act of notarization is such a simple, ministerial function that, if Sandra Stevenson were to have been discovered to be an imposter, another notary could have been found immediately. Any notary public could have performed Sandra Stevenson's function.

If, on the other hand, the security had no value, it really didn't matter who signed the mortgage--whether it was the actual owner, as in this case, or a forger. The mortgagee would get no valuable security because the security was worthless.

I realize that Pennsylvania law concerning the necessity of a notarization to record a mortgage adds an additional factor to this case which is not present in the cases cited above. Nevertheless, I don't find that the forged notary signature proximately caused the loss here (even if under other fact situations it might have). The security here was worthless when the mortgage was executed. It was worthless when the fraud was uncovered. Moreover, there was no effort by Jefferson to record the mortgage during the brief period of time when the mortgage might have attached Jefferson's priority lien to the Locust Street property.1 Therefore, the scenario never arose under which Jefferson's loss could have resulted from the lack of a valid notarization.

For the above reasons, I conclude that the loss did not result directly from the forged notary's signature; it resulted directly from the worthlessness of the security. I, therefore, respectfully dissent. Under the circumstances of this case, I would affirm the decision of the district court.