This diversity suit for breach of a suretyship contract, decided in favor of the defendants after a bench trial, presents questions primarily relating to the contract-law doctrine of "novation," but more broadly to principles of contract interpretation; all the questions are governed by the common law of Indiana.

In 1991 defendant Vigo purchased Buck Creek Coal, which operated a coal mine and was required by both federal and state law to post reclamation bonds as a condition of being permitted to operate the mine. 30 U.S.C. § 1259(a); 30 C.F.R. § 800.20; Ind.Code § 14-34-6-1. In connection with the purchase, Vigo, joined by defendant Atlas and by the owners of Vigo and Atlas (the Koesters and the Piepers, respectively, who are also defendants) and by Buck Creek Coal, signed a "General Indemnity Agreement." In it they agreed to indemnify Utica insurance company for any losses that Utica might incur from issuing reclamation bonds to the Indiana state government on behalf of Buck Creek. The following year (1992) another "General Indemnity Agreement" was signed, identical to the first, except that the only signers were defendant Schulties, who had not signed the previous agreement, and the Piepers (who, remember, are Atlas's owners). Mr. Pieper signed both individually and as president; and because of his having thus signed in his official capacity, as it were, as an agent apparently authorized to bind his principal, the district judge ruled that Atlas was bound by the second agreement. E.g., Winkler v. V.G. Reed & Sons, Inc., 638 N.E.2d 1228, 1232-33 (Ind.1994); City of Gary v. Conat, 810 N.E.2d 1112, 1115-16 (Ind.App.2004); DFS Secured Healthcare Receivables Trust v. Caregivers Great Lakes, Inc., 384 F.3d 338, 343 (7th Cir.2004) (Indiana law). Atlas has not challenged that ruling, even though it appears from the 1992 agreement that Pieper actually was signing in his capacity as president of Buck Creek, not of Atlas. So Atlas is bound; and to simplify our opinion we shall assume that Atlas and Schulties were the signers of the second agreement and Vigo and Atlas the signers of the first; in other words, we'll ignore not only the district court's unchallenged error but also the companies' owners. With this simplification, the case becomes Utica versus Vigo.

The surety bonds were later "forfeited"; that is, Buck Creek proving unable to fulfill its reclamation obligations, the state required the surety, Utica, to do the reclamation. Utica then brought this suit, against all the signers of either agreement, for reimbursement of the expense — some $400,000 — of the reclamation.

The district court concluded that the signers of the 1991 indemnity agreement were off the hook (except Atlas, since it had signed the second agreement as well) because the second agreement was a "novation," that is, a replacement of the first agreement, which therefore released the obligors in that agreement. SSD Control Technology v. Breakthrough Technologies, Inc., 685 N.E.2d 1136, 1137-38 (Ind.App.1997); Rose Acre Farms, Inc. v. Cone, 492 N.E.2d 61, 68-69 (Ind.App.1986); T & N v. Pennsylvania Ins. Guaranty Ass'n, 44 F.3d 174, 186 (3d Cir.1994); Restatement (Second) of Contracts § 280, comment b (1981, 2004 supp.); 3 E. Allan Farnsworth, Farnsworth on Contracts § 11.11, pp. 141-42 (3d ed.2004); 30 Williston on Contracts § 76:1 (4th ed.2004 supp., Richard A. Lord ed.). But the judge also turned down these defendants' counterclaim, in which they sought to recover the attorneys' fees that they had incurred in defending against Utica's claim.

Unable to recover its entire loss from the signers of the 1992 agreement (Atlas and Schulties — the latter now bankrupt), and seeking therefore to enforce the 1991 agreement against Vigo, Utica challenges the finding that the 1992 agreement was a novation.

The agreement does not describe itself as a novation or a substitute, or purport to release the signers of the first agreement. The only clue in the agreements themselves that the second one might be a novation is that Atlas signed both, and if the only purpose of the second was to add an indemnitor, namely Schulties, why did Atlas sign it, having signed the identical first agreement, unless that wasn't the purpose, and the second agreement replaced rather than supplemented the first? Additional evidence, that is, evidence beyond the two agreements themselves, was presented at the trial, and that evidence, together with the anomaly we've just noted, persuaded the judge that the second agreement was indeed a novation.

That evidence revealed the following. In 1992, Vigo sold the coal mine to Atlas and Schulties and they agreed to use their best efforts to replace the existing reclamation bonds and obtain a release of Vigo's liability under those bonds. They failed to do so, but an insurance agent named Jones submitted a "reclamation bonding application" to Utica, proposing to "transfer these bonds over and have new Indemnity Agreements signed by Chuck Schulties." There was no mention of Vigo. The agent testified that the reason Vigo was left off the second agreement was that it and its owners, who had signed the 1991 agreement, "had no ownership, they had no control, they were not party to running the company." Asked at trial whether Vigo's omission from the agreement was "confirmatory of the conversations that you'd had with [Utica's Gerald Swarthout, who handled the negotiations leading up to the 1992 agreement] that they would not be indemnitors," the agent replied "of course, they were selling."

Swarthout gave contrary testimony, but the district judge disbelieved it, in part because Schulties had (at the time!) substantial assets. That fact, together with his substantial expertise in coal mining, suggested that an indemnity agreement signed by him as well as by Atlas would provide sufficient security to persuade Utica or some other insurance company to issue new reclamation bonds to replace those that Utica had issued. For although the General Indemnity Agreement (whether the 1991 or the 1992 version, since they were identical except for the signers) embraced replacement bonds, it was terminable by an indemnitor on 20 days' notice. And therefore as part of Vigo's sale of the coal mine to Atlas and Schulties, Atlas and Vigo could have withdrawn from the 1991 agreement and Atlas and Schulties, with a net worth between them of $7 million, could have persuaded Utica or some other insurance company to issue new surety bonds that Vigo, having withdrawn from the agreement, would not have been a guarantor of. Utica does not deny that its bonds could have been replaced in this fashion. Ind.Code § 14-34-6-14.6.

The consequence of replacing the bonds would have been to release Vigo. But given Schulties' financial wherewithal and mining expertise, and Vigo's natural reluctance to remain a guarantor of performance over which, as a result of the sale, it would no longer have any control, the judge was on solid ground in finding that Utica would have agreed to an express novation rather than lose to another insurance company a business relationship that yielded it significant premiums ($18,000 a year) in exchange for assuming what seemed at the time a modest risk. Modest because underground coal mines (Buck Creek's mine was an underground mine) tend to require less expense to restore the surface after the mind is exhausted than strip mines and because two substantial-seeming parties, Atlas and Schulties, had agreed to indemnify Utica for any loss.

The judge's reconstruction of the parties' deal — his conclusion that they intended the second agreement to substitute for rather than supplement the first — is not clearly erroneous. On the contrary, it makes commercial, economic, and common sense; and good sense, or such synonyms as commercial reasonableness, provides sound guidance for interpreting ambiguous contracts, in Indiana as elsewhere. Soames v. Young Oil Co., 732 N.E.2d 1236, 1239 (Ind.App.2000); F.E. Gates Co. v. Hydro-Technologies, Inc., 722 N.E.2d 898, 903 n. 4 (Ind.App.2000); Beanstalk Group, Inc. v. AM General Corp., 283 F.3d 856, 859-61 (7th Cir.2002) (Indiana law); Morin Building Products Co., Inc. v. Baystone Construction, Inc., 717 F.2d 413, 414-15 (7th Cir.1983) (ditto); XCO Int'l Inc. v. Pacific Scientific Co., 369 F.3d 998, 1005 (7th Cir.2004); Outlet Embroidery Co. v. Derwent Mills, 254 N.Y. 179, 172 N.E. 462, 463 (1930) (Cardozo, C.J.). As Judge Boudin explained recently, "Agreements, especially commercial arrangements, are designed to make sense. If one reading produces a plausible result for which parties might be expected to bargain, that reading has a strong presumption in its favor as against another reading producing an unlikely result (e.g., windfall gains, conditions that cannot be satisfied, dubious incentives)." National Tax Institute, Inc. v. Topnotch at Stowe Resort & Spa, 388 F.3d 15, 19 (1st Cir.2004).

The difficult question is whether the district judge was entitled to take evidence — to mine underground, as it were — rather than to stay on the semantic surface of the two agreements. A finding of novation is dynamite, as this case illustrates. Instead of supplementing a previous contract, it wipes it out. Parties that enter into multiple contracts with one another want protection against a trial that may convert a contract that says nothing about novation or release into a release of the obligors under a previous contract. Were it not for what we're calling the dynamite effect of a novation, there would be no reason to treat a novation any differently from any other contract modification.

One way to give obligees protection would be to have a rule that a novation or release must be explicit; that it can never be implied. Indiana rejects that approach, Winkler v. V.G. Reed & Sons, Inc., supra, 638 N.E.2d at 1233-34; Rose Acre Farms, Inc. v. Cone, supra, 492 N.E.2d at 69; no jurisdiction, to our knowledge, accepts it. See, e.g., Imperial Hotels Corp. v. Dore, 257 F.3d 615, 620-21 (6th Cir.2001); National American Ins. Co. v. Hogan, 173 F.3d 1097, 1107-08 (8th Cir.1999); Capital Nat'l Bank v. Hutchinson, 435 F.2d 46, 50 (5th Cir.1970); Kroll v. Sugar Supply Corp., 116 Ill.App.3d 969, 72 Ill. Dec. 396, 452 N.E.2d 649, 653 (1983). So we can set it to one side. There are three alternatives. One, proposed by the Williston treatise, is to require that a novation be proved by clear and convincing evidence. 30 Williston on Contracts, supra, § 76:42. The cases do not support this alternative; Johnston v. Holiday Inns, Inc., 565 F.2d 790, 797 (1st Cir.1977), rejects it outright. What the case law, including Indiana case law, does support is that proof of a novation must be "clear and definite" or "clear and satisfactory" or words to that effect. State v. Traylor, 98 Ind.App. 290, 173 N.E. 461, 464 (1930); Joslyn Mfg. Co. v. Koppers Co., Inc., 40 F.3d 750, 759-60 (5th Cir.1994); Meredith v. Rockwell Int'l Corp., 826 F.2d 463, 466 (6th Cir.1987); Hofer v. Merchants State Bank, 823 F.2d 271, 272-73 (8th Cir.1987) (per curiam). We take this to be an appropriate reminder of the fell consequences of a finding of novation.

Another alternative, which would treat the determination of novation like other issues of contractual interpretation, is that a novation can be implied only if there is an ambiguity as to whether the agreement claimed to be a novation really is one, rather than merely a supplement to the earlier agreement. This approach also has support in Indiana, as in other states. Downing v. Dial, 426 N.E.2d 416, 419 (Ind.App.1981); Boswell v. Lyon, 401 N.E.2d 735, 741 (Ind.App.1980); Calder v. Camp Grove State Bank, 892 F.2d 629, 631-33 (7th Cir.1990); see also In re Newport Plaza Associates, L.P., 985 F.2d 640, 644-45 (1st Cir.1993).

Still another alternative, supported by the Indiana case of Rose Acre Farms, Inc. v. Cone, supra, 492 N.E.2d at 68-69, and by some cases from other jurisdictions as well, Imperial Hotels Corp. v. Dore, supra, 257 F.3d at 620-21; United States v. Hastings Motor Truck Co., 460 F.2d 1159, 1161 (8th Cir.1972), appears to allow extrinsic evidence to be introduced in any novation case, perhaps on the theory (not articulated in the cases, however) that the agreement claimed to be a novation cannot be interpreted without consideration of the meaning of the contract it is alleged to have replaced. This cannot be quite right; if the second agreement states "this is a novation" or "this is not a novation," that surely is the end of the case; extrinsic evidence will not be allowed.

Probably what we have described as alternative rules come to the same thing, in the following sense; the court can always peek at the earlier contract, and if in light of what that contract says it is uncertain whether the new contract is a novation, then the court can treat the interpretive issue as one of fact, that is, can take evidence on it, but always bearing in mind the need for the proof of a novation to be clear. The earlier contract is a bit of extrinsic evidence that is always already in the case, and if it shows that the new contract is ambiguous (as to whether it is a novation), the door is opened to the presentation of additional extrinsic evidence, which may make clear that the new contract really was intended to take the place of the old.

The distinction that we are exploring is the familiar one between intrinsic and extrinsic (or patent and latent) ambiguity, the former referring to an ambiguity that is apparent just from reading the contract itself (the contract in issue, and therefore the 1992 General Indemnity Agreement), and the latter referring to an ambiguity that becomes apparent only when evidence outside the contract, that is, extrinsic evidence, is consulted. Adams v. Reinaker, 808 N.E.2d 192, 196-97 (Ind.App.2004); East v. Estate of East, 785 N.E.2d 597, 601 n. 1 (Ind.App.2003); United States v. Rand Motors, 305 F.3d 770, 774-75 (7th Cir.2002); Rossetto v. Pabst Brewing Co., Inc., 217 F.3d 539, 542-43 (7th Cir.2000); PMC, Inc. v. Sherwin-Williams Co., 151 F.3d 610, 614 (7th Cir.1998). Vigo argues that there is an intrinsic ambiguity consisting of Atlas's having signed the second agreement, when it had signed the first. If the second just added indemnitor(s) (for remember that the terms of the two agreements are identical), what was Atlas doing there? One might be tempted to answer that it couldn't hurt to have the indemnitors sign multiple agreements, just in case one of the agreements turned out to have some defect, but if so why was Vigo left off the second?

This is a genuine puzzle, raising a serious question about the meaning of the 1992 agreement, but it is not an intrinsic ambiguity. No one reading just the 1992 agreement, the agreement contended to be a novation, would attach any significance to Atlas's presence as a signatory. It is only when one reads the other agreement, and notices that Atlas signed that one too, that the meaning of the second is thrown into doubt. That is a bit of evidence extrinsic to the second agreement, but it does create an (extrinsic) ambiguity. And it is not the only evidence that does so — there is also Vigo's sale of the mine to Atlas and Schulties, the undertaking by the purchasers to obtain Vigo's release from the 1991 indemnity agreement, Schulties' expertise and financial strength, the insurance agent's testimony, and the unlikelihood that Utica would have refused to release Vigo from the earlier agreement because, had it refused, the purchasers of the coal mine would have turned elsewhere for the necessary reclamation bonds.

But the desirability of allowing such evidence to be presented in order to create a full and accurate picture of the transaction in suit, and by doing so resolve the interpretive question, must be balanced against the danger that the "four corners" rule — the hallowed rule of contract law that the court must not delve below a contract's semantic surface in interpreting it unless the contract is ambiguous, Thomas v. Thomas, 577 N.E.2d 216, 219-20 (Ind.1991); Poznic v. Porter County Development Corp., 779 N.E.2d 1185, 1189-90 (Ind.App.2002); Tri-Central High School v. Mason, 738 N.E.2d 341, 344 (Ind.App.2000); Neuma, Inc. v. AMP, Inc., 259 F.3d 864, 873-74 (7th Cir.2001); Bourke v. Dun & Bradstreet Corp., 159 F.3d 1032, 1036 (7th Cir.1998) — will unravel if extrinsic evidence can be used to demonstrate ambiguity. For that is the very evidence that the party presenting it would use to prove that its interpretation of the contract was correct, converting interpretation from a reading exercise by the judge to a factual decision by a trier of fact and thus stripping the parties of protection from the vagaries of a jury. (Neither side requested a jury in this case, it is true; but either could have.)

There are two methods of resolving such a conflict. The first, which is the older and the more conventional, is based on the parol evidence rule — the rule that if a written contract is "integrated," parol (i.e., extrinsic) evidence, whether oral or written, concerning the negotiations or other background to the contract, including preliminary agreements and understandings, is inadmissible to contradict the apparent meaning of the written contract. Franklin v. White, 493 N.E.2d 161, 165-66 (Ind.1986); Truck City of Gary, Inc. v. Schneider Nat'l Leasing, 814 N.E.2d 273, 278 (Ind.App.2004); Deckard v. General Motors Corp., 307 F.3d 556, 563 (7th Cir.2002) (Indiana law); In re Yates Development, Inc., 256 F.3d 1285, 1289-90 (11th Cir.2001). An "integrated" contract is simply one that is intended to be the parties' complete agreement. So what the parol evidence rule amounts to is that if the parties want their written contract to be treated as the sole basis for interpreting their agreement the court will honor their wish — their wish to have any contract dispute resolved by application of the "four corners" rule.

Normally parties demonstrate this wish by including an integration clause in the contract, that is, a clause that states that the contract reflects the parties' complete understanding of their deal. The omission of the clause (there was none here) is not fatal; but when it is omitted a party can ask the judge to look at extrinsic evidence, though initially just for the limited purpose of deciding whether the contract really is integrated. Franklin v. White, supra, 493 N.E.2d at 166-67; I.C.C. Protective Coatings, Inc. v. A.E. Staley Mfg. Co., 695 N.E.2d 1030, 1035 (Ind.App.1998); Betaco, Inc. v. Cessna Aircraft Co., 103 F.3d 1281, 1285-86 (7th Cir.1996); Cook v. Little Caesar Enterprises, Inc., 210 F.3d 653, 656 (6th Cir.2000); Starter Corp. v. Converse, Inc., 170 F.3d 286, 295 (2d Cir.1999). This does not mean throwing open the contract to a trial. The evidence is presented to the judge for a preliminary determination of whether the contract is integrated. Only if he decides that it is not integrated does the case go to the jury (or the judge in his capacity as the trier of fact in a bench trial, as in this case) for a determination, based on extrinsic evidence as well as the written contract — based in short on all relevant, admissible evidence — of what the contract really means. Truck City of Gary, Inc. v. Schneider Nat'l Leasing, supra, 814 N.E.2d at 278; Center State Farms v. Campbell Soup Co., 58 F.3d 1030, 1032-33 (4th Cir.1995); American Bridge Co. v. Crawford, 31 F.2d 708, 710 (3d Cir.1929); 11 Williston on Contracts, supra, § 33:19.

The other method of resolving the tension between the interest in admitting all evidence bearing on the meaning of a contract and the parties' desire to avoid a trial in which that meaning might be determined by a commercially unsophisticated jury or judge is to allow only objective evidence to be used to determine the existence of an ambiguity. Hemenway v. Peabody Coal Co., 159 F.3d 255, 259 (7th Cir.1998) (Indiana law); Mathews v. Sears Pension Plan, 144 F.3d 461, 467 (7th Cir.1998); Cole Taylor Bank v. Truck Ins. Exchange, 51 F.3d 736, 737-38 (7th Cir.1995); AM Int'l, Inc. v. Graphic Management Associates, Inc., 44 F.3d 572 (7th Cir.1995); Kerin v. United States Postal Service, 116 F.3d 988, 992 n. 2 (2d Cir.1997); Duquesne Light Co. v. Westinghouse Electric Corp., 66 F.3d 604, 614 (3d Cir.1995); Carey Canada, Inc. v. Columbia Casualty Co., 940 F.2d 1548, 1557-58 (D.C.Cir.1991); Cincinnati Ins. Co. v. River City Construction Co., 325 Ill.App.3d 267, 258 Ill. Dec. 987, 757 N.E.2d 676, 681 (2001). The idea of a judicial screening is retained, but there is an additional screen as well: the evidence that the judge uses to decide whether the meaning of the contract will have to be determined in a trial must be objective in the sense of not depending on the credibility of the parties to the contract; the evidence must be independently verifiable. The fact that a party testifies that although the contract says X he and the other party to the contract meant Y flunks the test of objectivity and so is inadmissible, whereas under the first approach, the one based on the parol evidence rule, a judge would be allowed by resolving a swearing contest to decide that the contract was not integrated and therefore that its meaning presented an issue for trial.

Under either approach, extrinsic ambiguity was adequately demonstrated here. There was no integration clause in the 1992 agreement (or in the prior one, for that matter, for remember that the terms, as distinct from their signatories, are identical). And so Judge Hamilton had to consider extrinsic evidence in order to decide whether the agreement was the complete agreement of the parties. It was not. From the sale of the mine and the other circumstances we've recounted it is apparent that the parties understood the 1992 agreement to replace rather than supplement the previous one.

Under the "objective evidence" approach, the conclusion is the same. The evidence that the 1992 agreement was a novation rather than a supplement was objective in the sense that we have just explained. The sale of the mine, Atlas's signature on the agreement, the testimony of the insurance agent — for he was Utica's agent, not Vigo's — and the implausibility of supposing that Vigo would have agreed to remain bound by the 1991 agreement as a condition of being able to sell the mine, were none of them supported only by self-serving, unverifiable testimony of a party.

So the judge's ruling that there was a novation stands. But it is does not follow, as Vigo's counterclaim charges, that by trying to enforce the 1991 agreement against Vigo, Utica violated the agreement and is therefore liable for damages for breach that consist of the attorneys' fees that Vigo expended in this case. This argument would if accepted make attorneys' fee shifting the rule in contract cases. Several states do allow the award of attorneys' fees as a matter of course in a breach of contract suit. MindGames, Inc. v. Western Publishing Co., Inc., 218 F.3d 652, 654 (7th Cir.2000) (Arkansas law); Flourine On Call, Ltd. v. Fluorogas Ltd., 380 F.3d 849, 866-67 (5th Cir.2004) (Texas law). But most — including Indiana — do not. Thor Electric, Inc. v. Oberle & Associates, Inc., 741 N.E.2d 373, 382-83 (Ind.App.2000); Shumate v. Lycan, 675 N.E.2d 749, 754-55 (Ind.App.1997); Osler Institute, Inc. v. Forde, 386 F.3d 816, 818 (7th Cir.2004) (ditto); Oscar Gruss & Son, Inc. v. Hollander, 337 F.3d 186, 199-200 (2d Cir.2003); cf. J.R. Cousin Industries, Inc. v. Menard, Inc., 127 F.3d 580, 583 (7th Cir.1997). Vigo's argument confuses a mistaken litigating position in a contract case with a breach of the contract. A breach of contract is a failure to perform a contracted-for undertaking. Utica did not fail to perform any of the undertakings that it had agreed to in the 1991 or 1992 agreements. It merely mistook Vigo's undertakings.

There is one more issue. An Indiana statute, recently repealed but conceded to be applicable to this case, provides, so far as bears on the case, that a "debtor" may not "bring an action upon an agreement with a creditor ... to forbear from exercising rights under a prior credit agreement" unless the agreement "sets forth all the material terms and conditions of the agreement" and "is signed by the creditor and the debtor." Ind.Code § 32-2-1.5-5. Many states have such statutes, as we noted in Consolidation Services, Inc. v. KeyBank Nat'l Ass'n, 185 F.3d 817, 819-20 (7th Cir.1999). They are a reaction against "lender liability" litigation, Whitney Nat'l Bank v. Rockwell, 661 So. 2d 1325, 1329-31 (La.1995), in which borrowers assert "breach of oral agreements to lend, to refinance or to forbear from enforcing contractual remedies." Id. at 1329. And indeed the Indiana law was not really repealed, but merely shifted without substantive changes to another chapter of the Indiana Code. See Ind.Code §§ 26-2-9-1 et seq.

Utica argues that the 1992 General Indemnity Agreement, if viewed as a novation, is an agreement between a creditor (Utica, the releasing party, in the novation aspect of the agreement) and a debtor (Vigo, the released party), and neither party signed the agreement and the agreement doesn't set forth all the material terms — it doesn't even set forth the key term: that it is a novation.

Vigo was not suing on the 1992 agreement, but merely interposing it as a defense to Utica's suit on the previous agreement. Transportation & Transit Associates, Inc. v. Morrison Knudsen Corp., 255 F.3d 397, 400 (7th Cir.2001); Able Int'l Corp. v. B.P. Chemicals America, Inc., 145 F.3d 67, 68 (1st Cir.1998); Phillips & Arnold, Inc. v. Frederick J. Borgsmiller, Inc., 123 Ill.App.3d 95, 78 Ill. Dec. 805, 462 N.E.2d 924, 929 (1984); 30 Williston on Contracts, supra, § 76:40. That in itself is not critical, however, as Wabash Grain, Inc. v. Bank One, Crawfordsville, NA, 713 N.E.2d 323, 326 (Ind.App.1999) makes clear, though two other courts disagree. Brown v. Founders Bank & Trust Co., 890 P.2d 855, 861-62 (Okla.1994); Fleming Irrigation, Inc. v. Pioneer Bank & Trust Co., 661 So. 2d 1035, 1039 (La.App.1995). If testimony to oral modifications of credit agreements is thought unreliable evidence, it doesn't matter whether the evidence is introduced by the plaintiff or the defendant.

Nevertheless the statute is not applicable to this case. A suretyship contract is not a "credit agreement," the key term in the statute, which defines a creditor (so far as bears on this case) as an insurance company that "extend[s] credit under a credit agreement with a debtor" and a debtor as one who "obtains credit under a credit agreement with a creditor," "seeks a credit agreement with a creditor," or "owes money to a creditor." Ind.Code §§ 32-3-1.5-2, -3. Utica did not lend money to the coal company, or to any other entity or individual involved in this case. Nor did it extend credit in the sense of committing itself under stated conditions to lend money to anyone in the future. Its obligations ran only to the Indiana state government, the entity that had required the posting of reclamation bonds on behalf of Blue Creek Coal. If Utica and Vigo come within the statute's reach at all, it is as "debtors" of the state.

And if despite what we have said the Indiana statute did render the 1992 agreement unenforceable (more precisely, not "interposable" as a defense), equally it rendered the 1991 agreement unenforceable, because the "creditor," Utica, did not sign it. Yet Utica sued to enforce the 1991 agreement, and prevailed against the signers of it. For that matter, it sued to enforce the 1992 agreement as well, and though it gained nothing because Schulties declared bankruptcy, it could have filed a claim in bankruptcy against him based on the agreement (we don't know whether it did or not). The doctrine of "mend the hold" forbids a contract party, particularly when it is an insurance company, to change its position on the meaning of the contract in the middle of litigation over it. National Hame & Chain Co. v. Robertson, 90 Ind.App. 556, 161 N.E. 851, 853 (1928); Houben v. Telular Corp., 309 F.3d 1028, 1036 (7th Cir.2002); United States v. Newell, 239 F.3d 917, 922 (7th Cir.2001); Harbor Ins. Co. v. Continental Bank Corp., 922 F.2d 357, 362-64 (7th Cir.1990). Which is what Utica has done.

AFFIRMED.