This is a diversity suit, primarily for breach of contract. The suit was tried to a jury, and the plaintiff, Autotrol Corporation, obtained a judgment of more than $1.5 million. The appeal raises a large number of issues, only a handful of which require discussion. So far as the remaining issues are concerned, it is enough to note that the district judge resolved most of them in a thoroughly satisfactory manner and that the others could not possibly have affected the outcome of the trial.

The contract provides that any dispute under it shall be resolved in accordance with the law of Texas. Neither party questions the validity of this choice-of-law stipulation but it is rather academic because there are no Texas cases on the hotly contested issues.

The contract, signed in May 1986, established a joint venture between Autotrol's controls division and Continental Water Systems Corporation to create a system for water purification based on a patented new technology, known as "electrodiarese," that Continental owned the exclusive right to exploit. Autotrol was to manufacture the control for the system and Continental the rest and both companies would sell the completed systems--Autotrol the large systems, Continental the small ones. There was an ambiguity, later to prove critical, about the dividing line between large and small.

Before production could begin, there had to be product specifications. These had not been completed when the contract was signed. Anticipating this possibility, the contract provided that when approved by both parties the specifications would be attached to, and thereby made a part of, the contract; and should the parties be "unable, in good faith, to agree upon the contents of the Products Specifications Schedule by June 30, 1986, either party hereto may elect to terminate this Agreement." On June 25 the parties agreed to extend this deadline to July 17. July 17 came and went and the product specifications had not been agreed upon but neither party exercised its right of termination. It was almost a year later, with the product specifications still not having been agreed upon, that Continental declared the contract terminated. That is the alleged breach. Meanwhile the other defendant, Olin Corporation, had acquired all the stock of Continental, and according to evidence that the jury reasonably could credit had decided that unless Autotrol acceded to the defendants' understanding of what the smallest system was that Autotrol was entitled to sell--which Autotrol refused to do--Autotrol would be encroaching on the segment of the market that Olin wanted to reserve for Continental.

The defendants argue that they were free to terminate the contract at any time after July 17, for any reason or no reason, without liability, provided only that the product specifications had not been finally agreed upon and attached to the contract as a schedule. Yet Continental encouraged Autotrol to continue working, which meant continue spending, on Autotrol's share of the project for many months after July 17--indeed, right up to the notice of termination--even though the product specifications had not been agreed upon. Autotrol does not argue that this encouragement was sufficiently promissory in form to support liability under a theory of promissory estoppel, on which see Restatement (Second) of Contracts Sec. 89(c) and comment d (1981). But it does argue that, in conjunction with Autotrol's parallel forbearance to exercise its right of termination, Continental's encouragement supports an inference that the parties had modified the contract to waive Continental's right to terminate after July 17 for failure to agree on product specifications. This is assuming Continental had such a right, which Autotrol denies, thus giving it two grounds upon which to argue that there was a breach of contract. The jury found a breach but was not asked to indicate the ground, so we must affirm if either ground is supportable; as a matter of fact both are.

Under either of Autotrol's theories, once July 17 came and went Continental could not terminate the contract without liability--ever--even if the parties never did work out product specifications. At first acquaintance the argument is implausible. The parties could not go into production without such specifications. If they reached impasse after bargaining in good faith, would not either party be entitled to walk away from the contract without liability? Apparently not. The contract is deliberately asymmetrical with respect to termination after July 17 (June 30 in the original contract, before the date was extended). Until then either party can walk if the product specifications have not been agreed upon. After that Autotrol can walk until production commences but "Continental shall have no right to terminate this Agreement except as provided in" certain paragraphs of the contract that relate to specific (and immaterial) changed circumstances, such as bankruptcy, but not to failure to work out product specifications. Not only by encouraging Autotrol to continue working after July 17, but in the contract itself, Continental seems to have surrendered the right it would otherwise have had to terminate the contract upon the failure of an essential condition.

What sense can this make? The answer lies in the asymmetry of the parties' position. Continental controlled the patent on electrodiarese and if it abandoned the project, say by failing to agree on product specifications, Autotrol would be unable to go forward and would lose its investment. But if Autotrol abandoned the project Continental could always get another partner. Making the contract terminable by Autotrol but not by Continental gave Autotrol leverage to force Continental to license the patent to Autotrol on reasonable terms if Continental lost interest in the project.

If all this is wrong and Continental somehow retained an implicit right to terminate the contract without liability, even after July 17, provided that the parties were unable to agree upon product specifications, still the jury was entitled to find that the parties had modified the contract to forbid termination on this ground at least until such time as the need to agree on product specifications was urgent and agreement impossible. That deadline had not been reached when the defendants terminated--on a ground, moreover, that had nothing to do with product specifications, but rather with the division of the market for the product. The modification was supported by consideration; both parties benefited from relaxing the deadline, inasmuch as this allowed the project to proceed. United States v. Stump Home Specialties Mfg., Inc., 905 F.2d 1117, 1122 (7th Cir.1990). Though oral, and somewhat vague about the duration of Continental's commitment, the modification was sufficiently definite to be enforceable, cf. Goldstick v. ICM Realty, 788 F.2d 456, 462-63 (7th Cir.1986), and an oral modification is enforceable under Texas law even if the contract forbids oral modifications, as this one did. Adams v. Can-Dee Oil Corp., 357 S.W.2d 808 (Tex.Civ.App.1962). The Texas approach, by no means idiosyncratic, 2 Farnsworth on Contracts Sec. 7.6, at pp. 229-30 (1990), is not so frontal a blow at freedom of contract as it may appear to be. For as Farnsworth points out, id. at p. 230, in most of the cases in which oral modifications are enforced in the face of prohibitory language in the contract the party seeking enforcement has relied on the modification; and an invitation to rely, here inferable from both statements and conduct of Continental, is the sort of waiver of a clause forbidding oral modifications that would be honored even under the Uniform Commercial Code, which makes such clauses expressly enforceable in contracts for the sale of goods. UCC Sec. 2-209(2); Wisconsin Knife Works v. National Metal Crafters, 781 F.2d 1280, 1286-88 (7th Cir.1986).

The remaining issues relate to damages. During the year in which the contract was in force, Autotrol incurred $245,000 in out-of-pocket costs of performance. There is no quarrel over the appropriateness of awarding these costs as damages for breach of the contract. But in addition the jury awarded Autotrol more than $700,000 in overhead expenses. If, for example, a salaried engineer spent 25 percent of his time on the project, then 25 percent of his salary and benefits during the period he was working on the project were considered to be damages from the breach. The defendants object, pointing out that it is merely a conjecture that if the contract had never been made in the first place Autotrol would have replaced it with a project that would have covered the engineer's salary and benefits.

Economists distinguish between a firm's fixed and variable costs. The former, as the name implies, are the same whether or not the firm does anything; a good example is the fee that a state charges for a corporate charter. The fee is paid before the firm begins operations and is utterly invariant to the firm's fortunes. It would be an improper item of damages for the breach of a contract because the breach could not have caused the expense to be incurred.

Variable costs are those that vary with the firm's activity--more precisely that are caused by fluctuations in that activity. It is easy to see how the out-of-pocket expenses that Autotrol incurred in the joint venture were variable costs--specifically, variable costs of the joint venture, that is, of the contract. Had it not been for the contract, those expenses would not have been incurred. They are recoverable as damages because the breach deprived Autotrol of the opportunity to recover them by making and selling the water-purification systems envisaged by the contract.

It is more difficult to see how any part of the salary of the engineer in our illustration is a variable cost of the contract. His salary would presumably have been paid, for a time anyway, whether or not he worked on the electrodiarese project or, for that matter, on some substitute project. His salary was an expense but not an expense "caused" even in part by the project and hence, it might seem, was not a loss when the project collapsed because of Continental's breach and Autotrol was as a consequence unable to recoup the money that it had expended on the project.

But this analysis is superficial. Note, Home Office Overhead as Damages for Construction Delays, 17 Ga.L.Rev. 761, 776-80 (1983). If the contract had never been signed, the engineer would have had more time to devote to some other project of Autotrol's. If we assume that the project would have been sufficiently profitable (which does not mean grandly profitable) to cover the salary and benefits of Autotrol's salaried employees who worked on it, it follows that the electrodiarese project turned this expense into a loss by preempting the substitute project that would have enabled the expense to be covered. Autotrol was left at the end of the day--thanks to Continental's breach--with no customer to charge 25 percent of the engineer's salary and benefits to.

It is a question of fact whether, as the example assumes, salary and other overhead items allocated as a bookkeeping matter to a broken contract would in fact have been recovered in a substitute contract. If the victim of the breach of contract is a growing firm--as Autotrol was--it is quite likely that they will be recovered. Growth implies alternative uses for the company's workforce. If not this contract there would have been another, and it would not have been broken--or if it would have been broken, there would have been the same entitlement to damages. But if the firm is declining, then it might well have had to pay its overhead expenses out of its own pocket had this contract not been signed, although before concluding this one would have to consider the possibility that the firm could have economized on those expenses by layoffs and other adjustments.

It might be a useful simplification of the law of damages to have a flat rule that any firm whose business was growing in real (i.e., inflation-adjusted) terms over the period of the contract is entitled to recover the overhead on the contract that was broken; for in all likelihood the firm could have substituted another contract profitable enough to cover that overhead. But this we need not decide; and naturally we are reluctant to speculate about the course of Texas law, let alone try to influence it. It is enough that a reasonable jury could conclude that Autotrol probably would have recouped its overhead expenses on other projects had the contract with Continental never been signed, instead of wasting a year of its employees' time on this barren contract. There was testimony that Autotrol's controls division was consistently overbooked with new projects and that its new projects had a consistent record of profitability--and remember that any project that Autotrol would have substituted for the electrodiarese project, had the latter never existed, need only have been profitable enough to cover the overhead expenses that Autotrol would have incurred in the substitute project, using the inputs that instead it wasted on electrodiarese, in order to justify an award of those overhead expenses to Autotrol as damages.

Of course if this substitute project were the last of Autotrol's promising projects, then the only thing the breach would have done was to accelerate Autotrol's having to swallow its overhead expenses. But this is merely a variant of the declining-firm scenario, one in which the award of overhead expenses will sometimes be inappropriate. Sometimes, perhaps often, but not always. To repeat a previous point, a declining firm may be able to slash its overhead expenses; and if it did not do so because they were necessary to the performance of the contract, and now it is left holding the bag because the contract has been broken, then those expenses are a loss caused by the breach and are recoverable as contract damages.

The question of the proper treatment of overhead expenses arises more frequently in cases in which the plaintiff is seeking not the expenses themselves--they have not been incurred--but the price of a contract that has not yet been fully performed, and the defendant asks that the overhead expenses assigned by the plaintiff on the uncompleted portion of the contract be deducted, on the theory that they were saved by the breach and therefore that their inclusion would exaggerate the plaintiff's loss. The proper analysis of that case is symmetrical with the proper analysis of our case. Houston Chronicle Publishing Co. v. McNair Trucklease, Inc., 519 S.W.2d 924, 932 (Tex.Civ.App.1975); Taylor v. Meirick, 712 F.2d 1112, 1121 (7th Cir.1983); 3 Farnsworth on Contracts, supra, Sec. 12.10, at p. 212. If the plaintiff can either cut his overhead expenses or recover them in a substitute contract, then he indeed has not lost them as a result of the breach and they should not be figured in his damages. But if he cannot do either of these things--if in other words these really are fixed costs--then the breach gives him no scope to economize and there should be no deduction.

The breach in this case of course occurred after the overhead expenses were incurred. This sequence meant that the breach, far from enabling those expenses to be covered in a substitute contract, converted them from a bookkeeping entry into a loss, because it was too late for the plaintiff to make the substitute contract that would have enabled the plaintiff to recover the expenses. If there had been a contract price, therefore, the jury would not have been entitled to subtract the overhead expenses from it in figuring Autotrol's damages. The only thing to be subtracted would be expenses not yet incurred, and therefore saved by the breach. But there was no contract price. Autotrol was not selling something to Continental; they were joint venturers. That made figuring profits difficult and the implicit theory of the damages award is that Autotrol would have had zero profits on the venture, and is thus conservative. The defendants do not deny that Autotrol would have done well enough on the contract to cover the overhead expenses allocated to it, and with that concession Autotrol's case is complete.

The assumption of zero profits is based on the rule of Texas law "that the loss of anticipated profits from a new business is too speculative and conjectural to support a recovery of damages." Universal Commodities, Inc. v. Weed, 449 S.W.2d 106, 113 (Tex.Civ.App.1969); see also Southwest Battery Corp. v. Owen, 131 Tex. 423, 427, 115 S.W.2d 1097, 1099 (1938). Were it not for that rule Autotrol would not be seeking reliance damages, but instead expectation damages, and the issue of overhead expenses to which we have been devoting such complex attention would be simplified. Autotrol would have presented an estimate of the price at which it would have sold the water-purification systems envisaged by the contract and would have subtracted the costs that it would have incurred to complete the systems after the breach. The difference would be its damages. Suppose that the overhead and other expenses that Autotrol incurred before the breach were $1 million, the costs it would have (but had not yet) incurred to complete the systems after the breach $2 million, and the price it would have obtained for the systems $4 million. Then its damages would be $2 million ($4 million--$2 million), and this would cover the overhead expenses plus an allowance for profit. Since Autotrol was not in fact a new business, and had an established track record on new projects from which a projection could have been made of the likely success of the joint venture with Continental, we are far from certain that the Texas "new business" rule should apply to this case; but that is a matter for the Texas courts to think about; it is not an issue for us.

The defendants place great weight on a passage in Kansas City Bridge Co. v. Kansas City Structural Steel Co., 317 S.W.2d 370, 377 (Mo.1958), from which we quote the gist: "prerequisite to plaintiff's recovery of this general overhead item as part of its damages, some evidence was essential from which the jury reasonably could have found that such general overhead was not only an expense but also represented a loss to plaintiff ... [E]ven though a percentage of that fixed overhead was properly allocable to the Leavenworth job during the period of delay, nevertheless any amount so allocated could not represent a loss or damage to plaintiff unless plaintiff would have, but for the delay, obtained other work (which it did not have or which it did not in fact obtain) sufficient in amount to have absorbed the allocated portion of general overhead." (Italics in original.) We have no quarrel with this; it is our analysis in a nutshell. But since the plaintiff is being asked to prove a counterfactual (what would have happened had the contract in suit not been signed in the first place), he should not be subjected to too demanding a burden of proof. The degree of precision that can reasonably be demanded of a litigant, unless you want to doom his case from the outset, depends on the nature of the issue. The most that can fairly be asked of a plaintiff in Autotrol's shoes is to establish a reasonable probability that it would have covered its overhead expenses by means of another contract had the contract in suit never been made. A higher burden would not only be unrealistic but reward people who break their contracts. A lower burden might, as noted earlier, be justified by a desire to simplify litigation to the ultimate benefit of both plaintiffs and defendants.

The defendants argue that some of the overhead expenses that Autotrol sought to recover as damages had been incurred either before the contract was signed, or, after the breach, in preparation for the lawsuit; and neither sort of expense, whether overhead or otherwise, is as a general rule recoverable as damages for a breach of contract. Cacavas v. Zack, 43 Mich.App. 222, 227, 203 N.W.2d 913, 916 (1972) (per curiam); Hough v. Jay-Dee Realty & Investment, Inc., 401 S.W.2d 545, 551 (Mo.App.1966); Bazzini v. Garrant, 116 Misc. 2d 119, 455 N.Y.S.2d 77 (Cty.Ct.1982). The qualification, "as a general rule," could be important, though. Texas has by statute entitled the winning party in a contract suit to recover his reasonable attorney's fees from the loser. Tex.Civ. Practice & Remedies Code Sec. 38.001(8); Streeter v. Thompson, 751 S.W.2d 329, 331 (Tex.App.1988); Gerdes v. Mustang Exploration Co., 666 S.W.2d 640, 645 (Tex.App.1984). The statute could be thought to knock the props out from under the rule against recovering expenses incurred in preparation for suit. But the plaintiff does not argue this. It grants the defendants' major premise, but its witnesses testified that all the expenses that the plaintiff sought to recover had been incurred during the period when the contract was in effect. At the oral argument of the appeal the plaintiff's counsel conceded with a frankness as refreshing as it is commendable that forty hours of one of its employees' time for which the jury had awarded damages had in fact been expended before the contract was signed. But the adjustment required by this concession would be trivial, and as the computation of damages is in any event very far from being an exact science we shall let the point pass. Pace Corp. v. Jackson, 155 Tex. 179, 190, 284 S.W.2d 340, 348 (1955); Restatement, supra, Sec. 352, comment a. In this connection we remind the reader yet again of the assumption of zero profits that underlies the award of damages to Autotrol. Because of this assumption, the award of damages probably on balance understates Autotrol's loss.

The final issue is the amount of Autotrol's attorney's fees. The witness who testified to the amount of the fees was Autotrol's in-house counsel; the fees had been incurred by the law firm hired to prosecute the lawsuit. The defendants argue that the only proper witnesses to testify to those fees were the lawyers who billed them. Although we cannot understand what possible ground the argument might have, it is pressed with such force that some comment is required. It seems to us, quite to the contrary, that house counsel is the ideal witness to testify to the amount of fees incurred. Here as in all cases where a court awards fees, the fees must be reasonable in amount. It is hardly to be expected that a lawyer will testify to having submitted an unreasonable bill, whereas house counsel, who must approve the outside lawyers' bills, is in a good position to explain to the jury the measures that he (in this case, she) took to protect the client against exorbitant billing.

We find no reversible error in the judgment, and it is therefore

AFFIRMED.