Reproduced with permission of 52 American Journal of Comparative Law (Summer 2004) 513-530
James Gordley [*]
Suppose that the parties are agreed on what performance each is to make but some event makes this performance impossible, or more difficult, or its market price drastically changes. These problems have been handled by two doctrines drawn from two different sources. Roman law developed a doctrine of impossibility to deal with impossible performances. Canon law developed one of changed circumstances to deal with unexpected events that made a performance of greater or lesser value to each party. In the medieval and early modern period, jurists developed philosophical and moral explanations of these doctrines. With time, these explanations were forgotten or simplified. Today, to many modern jurists, both doctrines seem to defy rational explanation. Nevertheless, if we look at the history of these doctrines, we will find that there is a much to be learned, both negatively and positively. Negatively, we will see, the jurists were never able to develop a good explanation of the Roman texts governing impossibility. Their explanation was that no one could be bound to do the impossible - which simply did not square with the texts. On the positive side, they did develop a good explanation of the effect of unexpected circumstances. It was so good that I believe that if it is properly understood, it can also explain why relief is sometimes given when a performance is impossible.
I. IMPOSSIBILITY: A DOCTRINE THAT FAILED
There is a sense in which the doctrine of impossibility did not fail. It is recognized in one form or another by many modern legal systems. It does not enforce all contracts in which performance was or has become impossible, but it does enforce some, and as we will see, the end results are generally correct. But the attempt to explain [page 513] which contracts should be enforced in terms of whether their performance is possible or not has been a failure.
A famous Roman text contained the maxim, "there is no obligation to the impossible." A number of texts excuse a party when performance was impossible at the time the contract was made, all of which concern two types of contracts, sale and stipulatio. For example, a sale is not valid if the object sold never existed or had perished prior to the sale. It is also true that if a party's performance became impossible after the contract was made, he was excused provided he was not at fault. But one cannot conclude that in Roman law, a party was excused if his performance was or became impossible and he was not at fault in the ordinary sense. According to one text, if his performance was initially impossible, a party cannot liability escape if performance is merely beyond his own power. It must be beyond anyone's power. As later commentators put it, impossibility must be "objective" or "absolute," not "subjective" or "personal." If a party's performance became impossible, he could not escape merely because he was not at fault in the ordinary sense of the word. One who borrowed property gratuitously for his own use is liable if he failed to exercise the most scrupulous diligence (exactissima diligentia). He is not liable if the property was destroyed by invading enemies or bands of robbers. The puzzle is why a party should be liable failing to be "more diligent than the diligent." Many modern scholars believe that in classical Roman law, he was liable for custodia which meant neither negligence nor strict liability. A person liable for custodia was liable even though he used proper care for a loss that the use of proper care would typically prevent. He was not liable for vis maior, that is, for accidents that no one could have prevented. The medieval jurists classified this kind of "fault" as culpa levissima - most light fault. From there, one ascended through culpa levis, culpa lata, and culpa latior to dolus or intentional wrongdoing, the correct definition of each degree remaining a matter of continual argument.
The medieval Canon lawyers, however, turned impossibility and fault into basic principles of moral responsibility. The Canonists concluded, [page 514] after some initial hesitation, that one could not be morally obligated to do the impossible. Gratian had said that when every alternative course of action is sinful, one must choose the lesser sin. In some of the cases he was discussing, the seemingly lesser sin was to break a promise: after promising to keep a friend's secret, one learns he is planning a murder. If Gratian were right, a person might find it impossible not to sin since every possible course of action is sinful. Later Canonists rejected his position. By definition, one is morally obligated not to commit a sin. One could not, they argued, be obligated to do the impossible. If a person in these circumstances should reveal the friend's secret, then to do so is not sinful.
The Canon lawyers also concluded that a person who was at fault because he failed to use due care was not only morally guilty  but morally obligated to make compensation for any harm he had caused. Since the Canon lawyers were concerned with moral responsibility, it is not surprising that they concluded that a person can be morally responsible only when he could have acted otherwise and was at fault for failing to do so. Nevertheless, their conclusions about moral responsibility were now in tension with the Roman limitations on civil liability.
In the 13th century, Thomas Aquinas used Aristotle's theory of human responsibility to explain the conclusions of the Canonists. Choice was an act of will, and one could only choose what was possible. A promise to do the impossible was not binding. In the 16th century, this approach was followed by a school known to historians as the "late scholastics" who were attempting to synthesize Roman law with the moral philosophy of Aristotle and Aquinas.
The late scholastics borrowed the conclusion that one cannot be obligated to keep an impossible promise. They never explained how [page 515] to reconcile this maxim with the Roman rules. They were followed in turn by members of the northern natural law school of the 17th and 18th centuries. Samuel Pufendorf claimed that the seller was never liable for failing to do the impossible but, if he were at fault in making the promise, the buyer could recover any loss suffered. He had achieved consistency but only by sacrificing the Roman texts.
In discussing fault, the late scholastics admitted that they could not reconcile their moral philosophy with Roman law. In moral philosophy and theology, they said, fault means sin, a deviation of the will from right reason and the law of God. For the Roman lawyers, it meant "the omission of some sort of diligence from which something disadvantageous happens," a "deviation from that which is good and that could be provided for by human diligence."
Two leading late scholastics, Lessius and Molina concluded that the jurists were describing Roman positive law, not natural law. By natural law, one could only be liable for failure to use the care that men of the same condition use. They tried, rather unsuccessfully, to find pragmatic justifications for the Roman rules. According to Lessius, these rules prevented fraud and iniquity and by promoted diligence. According to Molina, they discouraged litigation and encouraged people to make contracts that promote commerce and human society such as loan for use, deposit, and lease.
The Roman texts would have been hard to explain by any theory. It is difficult to see, for example, why one needs two doctrines, impossibility and fault, and why the first should apply when a performance is impossible from the beginning, and the second when the performance becomes impossible. But there was a deeper reason why the natural law theory could not explain the Roman texts. The natural lawyers had a theory about when a person was at fault in the moral sense. As Zimmermann and Wollschlager have pointed out, such a theory could not explain Roman law because the Romans were really imposing liability absent fault for risks that, in their view, should fall [page 516] on the promisor. One of the weaknesses of the natural law tradition, as I have noted elsewhere, is that it never succeeded in finding a place for strict liability.
The 19th century jurists thus inherited a body of law in disarray. The principled explanations of the natural lawyers did not seem to explain the Roman law, which was very hard to explain in any case. In France, the textual problems changed with the enactment of the Code, although in a way that made principled solutions still more elusive. In Germany, where the Roman texts remained in force, the jurists developed principled explanations, and then forced the Roman texts to conform to them. In England, Roman law was borrowed in part but then confounded with another doctrine entirely which concerned implied conditions.
In the late 18th century, Thomasius in Germany and Le Brun in France attacked the traditional view that there were degrees of fault, and that the degree for which a person is liable dependin on the contract in question. Speaking for the drafting committee, Bigot-Preameneu denounced the old view as useless and overly subtle. Accordingly, Article 1137, paragraph 1 declared that when a person is charged with looking after an object, there would be one standard of care, that of a bon pere de famille.
But the older concept of culpa levissima, though rejected by this provision, reappeared elsewhere. Articles 1147-48 cast doubt on whether a party was normally liable for ordinary negligence, as Article 1137, par. 1 suggested. According to Article 1147, a party is liable "whenever he does not prove that non-performance resulted from an extrinsic cause (cause étrangere) that cannot be not be imputed to him." According to Article 1148, he was not liable if performance had been hindered by "irresistible force (force majeure) or a chance event (cas fortuit)." For Pothier, from whom the drafters had borrowed, cause fortuit and force majeure meant vis maior, the sort of event that would exonerate even the person liable for faute la plus legere or culpa levissima. Thus the drafters established two contradictory general rules: one imposing liability for ordinary negligence, and the other for what had traditionally been called culpa levissima. [page 517]
The approach of the French commentators was confused. Initially, Toullier and Duranton each espoused a clear theory, but neither theory was accepted. Toullier claimed that the Code had, in fact established only one standard of liability in contract as well as tort: liability for culpa levissima. Duranton claimed that the Code had preserved the traditional theory of degrees of fault depending on the type of contract the parties entered into.
The later French commentators rejected both views as incompatible with Article 1139, par. 1. Their own ideas were much less coherent. With variations, Aubry, Rau, Demolombe, Laurent and Larombiere concluded that the standard of care of Article 1137 par. 1 was ordinary negligence. Nevertheless, they also agreed that under Articles 1147-48, a party escapes only if he proves the sort of event traditionally called vis maior. Their lists of such events are like those of the Romans: tempest, earthquake, war, and so forth. They then sluffed over the contradiction, usually, by asserting or implying that these standards are always or nearly always the same: there is fault when there is no force majeure, cas fortuit or cause étrangere. Demolombe did observe that sometimes a party might not be at fault even though no such event had occurred, but he simply noted that the security of transactions requires that the party be held liable.
Demolombe and Laurent also conflated vis maior with impossibility. They explained Articles 1147-48 as an application of the principle that that no one is liable for the impossible. They endorsed the Roman rule that performance will only be excused when it is objectively impossible or impossible for anyone, not if it is only impossible for the particular party.
Finally, Rene Demogue found a more consistent solution which is now widely accepted. He distinguished obligations to use proper [page 518] means (obligations de moyens) from obligations actually to attain a certain result (obligations de résultat). In the former case, a party was liable for ordinary negligence. In the latter case he was liable unless he could prove force majeure, cas fortuit or cause étrangere.
The problem then is to explain which obligations fall into each category and why. The term "impossibility" and the principle that there is no obligation to do the impossible are not helpful.
Scorning such pragmatic confusion, the 19th century German jurists tried to formulate coherent definitions of impossibility and fault. Unlike the natural lawyers, the Germans did not claim to be developing moral or philosophical explanations of responsibility. They were trying to state clearly the ultimate conceptions on which the law rested. This time, however, their method led them to the same sweeping conclusion as the natural lawyers with the same unfortunate results.
They concluded that it was logically contradictory to say that a person is obligated to make an impossible performance. The obligation to do something presupposes the possibility of doing it.
They faced the same difficulties as the natural lawyers in squaring this conclusion with the Roman texts. They acknowledged that a contract was invalid only for initial objective impossibility, a conclusion that found its way into 306 of the German Civil Code. But they never found a satisfactory explanation of why it should matter that the performance was impossible from the outset. Ultimately, the German reforms of 2002 abolished the distinction between initial and subsequent impossibility.
Similarly, the German jurists concluded that, logically, fault could only mean failure to use the care that a person could reasonably be expected to use. As Puchta said, liability for any higher degree of care is liability for chance. As Windschied said, it is not liability imposed because one was negligent but liability imposed despite the fact that he was not. Again, the texts seemed to stand in their way. Thibaut and Hasse argued that the Roman texts did not impose liability for what had traditionally been called culpa levissima. Eventually, [page 519] most of the German jurists persuaded themselves that Thibaut and Hasse were right. To reach this conclusion, texts that imposed liability for custodia were said to refer merely to the kind of care a person must give, not the degree of care. As mentioned earlier, most modern scholars do not read the texts this way.
The solution of the jurists passed into the German Civil Code. While initial objective impossibility invalidates a contract, subsequent impossibility excuses the promisor only if he was not "responsible" (275). He is responsible if he acted willfully or negligently (276). Negligence is defined as a failure to use "ordinary care" (277). The German reforms of 2002 abolished the distinction between initial and subsequent impossibility. Supposedly, a party was to be liable for fault even after the reforms of 2002 eliminated the distinction between initial and subsequent impossibility. Nevertheless, although the drafters of the Civil Code had used the term "fault" in its ordinary sense, German courts were unwilling to live with that position even before the reforms of 2002. They have held a party liable because of an event which he could nor prevent by ordinary care if the event is one that is normally within a person's control: he is liable, for example, if he failed to perform because he lacked the financial resources to do so, or materials were delivered to him late, or his suppliers failed him. Moreover, the event that prevents performance must be one that the parties would not foresee or take into account at the time the contract was formed. As a result, cases in Germany come out in much the same way as in France, and, as we will see, in England and the United States. The problem remains [page 520] how to distinguish the cases in which the plaintiff is excused from those in which he is not. As in France, that cannot be done by invoking the principle that no one is liable for the impossible or that a person is only liable for fault.
Before the 19th century, English courts sometimes excused a party who could not perform. They did so, for example, when the performance was illegal, or the party obligated to perform had died, or the object bailed had been destroyed by an "act of God," or a plague suspended construction work. On the other hand, in the case of Paradine v. Jane, a lessee was not excused from paying rent when soldiers in the English Civil War made it impossible for him to occupy the property.
In the 20th century, Williston claimed that Paradine v. Jane represented the original common law principle -- that impossibility was no excuse -- and that 19th century cases had made inroads in the principle. Max Rheinstein agreed. He said the principle followed from the "essence" of assumpsit, and the cases excusing performance were exceptions to that principle. Like Williston, Rheinstein thought that impossibility first became a defense in 1863 in the leading case of Taylor v. Caldwell, when Judge Blackburn excused the owners of a music hall that had burned down from providing it for a performance. According to Williston and Rheinstein, he took this rule from the civil law.
But it is not clear why we should think assumpsit had an "essence" from which such a principle follows. Moreover, when courts do seemingly contradictory things, it is odd for an historian to claim he knows which actions reflect the true underlying principle and which are exceptions. The pre-19th century judges themselves did not say that Paradine v. Jane exemplified the underlying principle. They did not discuss the case. The first reference to it seems to have appeared only in 1802 in the notes of Sergeant Williams. Nor was the case discussed by pre-19th century treatise writers. Blackstone does not [page 521] mention it. Neither does the earliest treatise on contract law by Powell written in 1790.
Moreover, it is odd to cite Paradine for the proposition that in principle, impossibility does not excuse performance. As Pollock pointed out, the performance remained possible: the tenant could still pay the rent. The court did not even say that impossibility would never excuse the lessee from performing his obligations. It distinguished obligations expressly undertaken -- such aS paying rent -- from those imposed by law without an express promise - such as to give back the premises without "waste." The latter would be excused by a sufficiently uncontrollable event.
In any case, instead of trying to revive this distinction, both before and after Taylor v. Caldwell was decided, the 19th century treatise writers borrowed continental ideas about impossibility. Powell, W.W. Story, Leake, Pollock and Williston all said that performance was excused by events that made it absolutely or objectively impossible, but not by those that merely made it impossible for the promisor. In the United States, their view passed into the First Restatement of Contracts and is preserved, despite a change of vocabulary, in the Second Restatement. Leake and Hammon tried and failed to introduce the distinction between initial and subsequent impossibility.
Nevertheless, Blackburn imposed another requirement that must be met before impossibility would excuse performance. A party would only be excused when "the contract is not to be construed as a positive contract, but as subject to an implied condition that the parties shall be excused in case, before breach, performance becomes impossible from the perishing of the thing without default of the contractor." In doing so, he conflated the traditional doctrine of impossibility with another traditional doctrine which we will examine next: that a party may excused from performance by an implied condition. The question then becomes, in what cases a court should imply such a condition. [page 522]
Williston made an observation that has unfortunately become a commonplace. He claimed that common law and civil law take opposite positions: unlike civil law, common law treats any case of non-performance as a breach, regardless of fault, though it admits impossibility as an excuse. Actually, it is a question in France, of which contracts are subject to the cas fortuit and force majeure provisions of arts. 1347-48, in Germany, of when inability to perform counts as "fault" in the ordinary sense, and in England and the United States, of when an implied condition that a performance is possible is to be read into a contract. The differences do not reflect a disagreement in policy or principle on which contracts to enforce. As Williston said, and as Zweigert and Kötz now say, the practical results are much the same in both systems. They reflect the historical circumstance that different systems borrowed in different ways from traditional continental rules. In any event, each system faces a similar problem in deciding which contracts should and should not be enforced. This problem cannot be solved by an appeal to the principles that were once thought to explain the impossibility doctrine: that no one is liable for failing to achieve the impossible or for a result for which he was not at fault.
B. The Modern Search for Solutions
Some modern attempts to explain relief for impossibility or impracticability still seem to rely on the idea that a person should not be liable for a performance which has become impossible through no fault of his own. An example is the rule of the Vienna Convention:
Does an "impediment beyond his control" mean an impediment that he could not overcome by reasonable efforts? Does a failure to take the impediment into account when he contracted mean that, by reasonable efforts, he should have done so? If so, it would seem that a party is not liable whenever he was not at fault, a principle which cannot explain the law of any of the legal systems we have discussed.
Since we are not dealing with a doctrine based on fault, it would seem that we must be dealing with a problem of risk allocation independent of fault. That possibility is suggested by German jurists who [page 523] discuss to which party's "sphere of risk" an event belongs. The trouble is that speaking of spheres of risk is not helpful unless one can explain why a risk should be born by one party rather than the other.
Such an explanation was attempted by Richard Posner. Posner observed, correctly, that a party can better bear a risk if he can better foresee it, better insure or self-insure against it, or better prevent its occurrence. These observations are helpful in other contexts. By themselves, they are not so helpful in explaining the doctrine of impossibility. Suppose that war is declared, and so a shipment cannot be made. That is not the sort of event that is likely to be foreseen by parties who made no provision for it in their contract. It is not the sort of event for which one can readily buy insurance. If one asks which party is engaged in similar transactions, there is likely to be no clear answer. One party is repeatedly shipping and the other repeatedly receiving shipments. That leaves us with one factor: control. It is true that the results in the impossibility cases do seem related to the degree to which a party could have controlled the event that made performance impossible. No private party can control whether war is declared. One has more control over one's suppliers, one's workers, or one's financial condition. But from the standpoint of risk allocation, it would seem that the degree of control matters only because the party in control could have lessened the risk by taking reasonable precautions. In that case, why isn't a party always free from liability if he proves he took reasonable precautions? To put it another way, why isn't a party excused whenever he was not at fault as the late scholastics and natural lawyers claimed he should be?
Still another explanation is that relief should be given because an unforeseen event overturns the "equivalence" between what a party gives and what he receives. For example, according to the art. 6.2.2 of the Unidroit Principles of International Commercial Contracts, relief is given for "hardship," and "there is hardship where the occurrence of events fundamentally alters the equilibrium of the contract either because the cost of a party's performance had increased or because the value of the performance a party receives is diminished... ." The Unidroit Principles immediately add, however, that relief will be given only if "the risk of the events was not assumed by the disadvantaged party." As Helmut Kohler has noted, before one can speak of a disruption of equivalence, one has to know [page 524] which party assumed the risk of the event that occurred. In an insurance contract, nothing could seem more unequal than that a person who paid $2,000 to insure his house should receive $800,000 when it burned down. But the contract was not unequal when it was made. Indeed, the point of it was to transfer a risk to the insurance company. Therefore, before we ask whether equivalence in the value of the performances was disrupted, we have to ask what risks the contract transferred.
Consequently, another solution is to ask directly what risks were assumed by the party who failed to perform. According to the Uniform Commercial Code  and the Second Restatement of Contracts  that a party is not bound when the contract was predicated on a "basic assumption" which was false. To ask on what basic assumptions a contract was based is not much different than asking, as Judge Blackburn did, on what implied conditions it was made. As mentioned earlier, Blackburn was conflating the doctrine of impossibility with another traditional doctrine: the so-called clausula rebus sic stantibus: an implied condition that the parties will not be bound under certain circumstances which they did not envision. This rule is the ancestor of the doctrine of changed and unforeseen circumstances which is recognized in most modern legal systems with the notable exception of France. This doctrine was not part of Roman law but was developed by the Canon lawyers whence it passed into civil law. Correctly understood, however, I believe it can explain the cases the Romans and later legal systems have tried to resolve with the doctrine of impossibility.
II. UNEXPECTED CIRCUMSTANCES
According to the Canon lawyers, while it was wrong to break a promise, one could legitimately do so when circumstances had changed sufficiently. Gratian's Decretum contained a passage in which St. Augustine, following Cicero, said that one need not keep a promise to return a sword to a person who has become insane. A gloss to the Decretum explained that "this condition is always understood: if matters remain in the same state." The great medieval [page 525] jurist, Baldus degli Ubaldi. then read the doctrine into civil law. All promises were subject to such a condition.
Thomas Aquinas used Aristotle's theory of equity to explain why. According to Aristotle, whenever a law is made, particular circumstances may arise in which the lawmaker would not want it to be applied. As a matter of "equity," the law should not be applied in those circumstances. Aquinas concluded that similarly, promises are a kind of law one gives to oneself, and they are not binding in circumstances where the promisor would not have intended to be bound.
Aquinas' explanation was adopted by the late scholastics  and preserved by the 17th and 18th century natural lawyers. The promisor is not bound if the change of circumstances concerns the "unique reason" or "unique cause" for his promise  or the "presumption of some fact" on which his consent was conditioned.
In the 19th century, however, the doctrine of changed circumstances had few defenders. These few explained relief by saying the existence of certain circumstances was a tacit or implied condition of the contract. According to the French jurist Larombiere, an "error in motive" affected the validity of a contract only if the parties so wished, but a judge would determine whether they so wished by examining "according to the circumstances, if the fact alleged as a motive was taken to be the determining reason (raison déterminante) and if the consent depended on its reality." The German jurist Windscheid said that the continuation of certain circumstances could be an "undeveloped condition" of the contract, "undeveloped" in the sense that it was not expressly willed by the parties. For most 19th century jurists, the obvious objection to the doctrine was that a tacit or undeveloped condition was one that the parties never consciously willed. They had never thought about the change in circumstances, let alone agreed on what should happen if the change occurred. Thus the doctrine of changed circumstances slumbered during the 19th century.
It continues to do so in France, although, beginning in 1916, the administrative courts have developed a version of it called imprévision. The drafters of the German Civil Code ignored the doctrine, but the courts brought it back under the pressure of the First World [page 526] War and the drastic inflation of the 1920's. They gave relief under what is perhaps the Code's most general article, 242, which requires that a contract be performed in good faith. The doctrine is now enacted in 313 of the German Civil Code. It has been applied by German and Anglo-American courts to situations in which a performance is not impossible but more difficult than expected. As we will see, the doctrine can be used to explain relief in both types of cases.
1. An impossible performance
As applied to the problem of performances which are impossible, this approach requires us to ask why the parties would enter into either of two possible arrangements: an arrangement in which a party is liable for non-performance only if he is not at fault -- that is, his failure to perform is not due to causes for which he could reasonable be held accountable -- or an arrangement in which a party will be liable whether he is at fault or not. That is simply another way of asking whether his ability to perform is a tacit condition of the contract.
It would seem at first sight that to agree to be held liable even absent fault is to make a bet -- equal to the amount of damages one must pay -- on an event that lies outside one's control. It seems no different in principle that a clause that allows the amount that one party owes another to turn on a throw of dice. As the economists tell us, however, risk averse parties do not gamble unless the odds are tilted in their favor. Suppose the damages occasioned by non-performance are $100,000 and the odds of failing to perform, despite reasonable efforts to do so, are one in a hundred. The one party would not agree to be held liable unless he were offered more than an extra $ 1000, an amount which the other party, who is also risk averse, would never be willing to offer.
It would seem, then, that, to use the terminology of the French, all contracts should be obligations de moyens or agreements to use reasonable efforts and none should be obligations de resultat which commit a party to achieving a certain result. But that clearly is not the case. On the other hand, as we have seen, the reason is not likely to be one of the those that Posner has given why one party is in a better position to bear a risk more cheaply than the other party. Here, we are dealing with risks that, in most cases, neither party can better foresee or insure against. It is true that in many of the cases do seem to turn on the degree to which a party can normally control the event in question. No private party can control whether war is declared. He has more control over his suppliers, his workers, or his financial condition. But from the standpoint of risk allocation, it [page 527] would seem that the degree of control matters only because the party in control could have lessened the risk by taking reasonable precautions. In that case, why isn't a party always free from liability if he proves he took reasonable precautions?
The answer, in my view, is that it is often difficult for one party to prove that the other was at fault for failing to take reasonable precautions. How is one to know whether he could not have dealt with more reliable suppliers, cultivated better relations with his workers, or kept his financial position viable. It will therefore make more sense if, in such cases, he is held liable simply for failure to perform and charges a bit extra to compensate himself for that possibility. If that is so, even if liability without fault is not an explicit condition of the contract, it should be an implicit one. Nevertheless, in such a contract, some event may prevent performance for which it is extremely unlikely that the party in question is at fault, either because no one could have taken precautions against such an event or because his own interests gave him every reason to take precautions. If the point of an agreement to assume liability without fault is to avoid the difficulties of proving fault, that agreement should not apply in cases like these, where fault is not seriously in question. As the late scholastics would say, if the parties agreed to liability for non-performance without fault only to avoid an eventuality, that agreement should not apply in circumstances in which it does not arise.
Modern courts have also given relief when performance is not impossible but is physically more difficult than the parties had imagined. For example, in Mineral Park Land v. Howard, a California court excused performance when the defendant had agreed to take the gravel he needed from the plaintiff's land but the costs of doing so proved to be much higher than anticipated because, unknown to the parties, the gravel was under water.
These cases are in one way analogous to those we have already considered. As we have seen, difficulties of proving fault explain why contracts are often commitments to produce a certain result rather than merely commitments to use reasonable efforts to produce it. Similarly, here, the parties have their choice between two kinds of arrangements: a cost plus contract in which the party to perform recovers an extra amount in the event of hardship and a fixed price contract in which normally he agrees to bear unexpectedly high costs. One reason for entering into a cost plus contract is like that for agreeing to produce a certain result: a difficulty of proof. In a cost plus contract, it may be very difficult to prove that the party making the performance used reasonable efforts to keep costs under control. If that is so, one can see why relief is given in cases like Mineral Park. The cost in question is one that the excavator could not possibly control. [page 528] He cannot help it that the gravel is under water or that the costs of excavating it are therefore greatly beyond the amount he expected. The rationale is the same as before: if the parties have made the contract cost-plus only to avoid certain difficulties, then that arrangement should not govern circumstances in which these difficulties do not apply.
3. Market changes
Sometimes a contract becomes more difficult or easier to perform because market prices change. Normally, such changes do not call for relief. One of the principal reasons the parties would choose an agreement in which they can claim expectation damages rather than reliance damages is to lock in what they believe to be a favorable price. That is clearly their objective in a forward contract for the delivery of fungible goods readily available on the market. One cannot argue that either of the parties entered into the contract with any other risk in mind than a fluctuation of the market price.
Nevertheless, suppose the market price falls through the floor or goes through the ceiling. German courts have sometimes given relief: for example, when the outbreak of the First World War caused a huge increase in the price of steam  or of iron wire. In the Westinghouse litigation, in a case which was settled before appeal, the Westinghouse corporation sold nuclear generators and agreed to sell uranium at a fixed price for a fixed term thereafter. The price of uranium soared with the world energy crisis so much that performing its contracts would have bankrupted the company. Because case was settled, one cannot tell whether an American court would have given relief or not.
I believe that if the approach we have taken thus far is correct, relief is sometimes warranted. We have to ask why the parties would have wished to lock in what they consider a favorable price. In the Westinghouse case, the purchasers of the generators expected to buy uranium from Westinghouse to power the generators and sell electricity at a profit. The profit would be equal to its revenues from selling electricity minus its operating costs minus the cost of the capital invested in building the generator minus the amount it pays Westinghouse for the uranium. Suppose, however, that the price of uranium rises sufficiently, the purchasers could make more money by shutting down their generators and reselling the uranium on the open market. I don't see why the purchasers should be entitled to recover more than the profit they would otherwise have made minus [page 529] the costs they saved by shutting down their generators. To allow them to recover more would be to read the contract as though it had two purposes: (1) to guarantee that a rise in the price of uranium would not cut into the profit the purchasers make by selling electricity, and (2) to allow the purchasers to profit by a rise in the price of uranium so great that they can make a higher profit reselling it. The second purpose is simply a bet on the future price of uranium which is not connected with the purchasers' use of uranium. Had the parties thought of the possibility that the price of uranium would go that high, they would not have made such a bet. Risk averse parties do not gamble unless the bet is tilted in their favor. Once again, it is outside the purposes they did have in mind when they contracted.
As we have seen, the problem of unexpected events affecting the possibility or difficulty of performance has been dealt with by two doctrines: one of Roman origin -- the doctrine of impossibility -- and one of Canon law origin -- the doctrine of clausula rebus sic stantibus or changed circumstances. The Roman law doctrine resisted the efforts of centuries of jurists who tried to make sense of it. It never meant, nor could mean, that a party was not obligated to perform simply because he found it impossible to do so. The Canon law doctrine, if properly understood, can explain the problems the Roman law doctrine confronted and more. Under the influence of Aristotelian moral philosophy, it meant that one had to ask under what circumstances the parties would have wished to be bound when they committed themselves. We can therefore distinguish several different arrangements they might have made and ask which would best have served their purposes: one in which a party is bound only to make reasonable efforts or one in which he is bound even if these efforts fail; one in which a party can charge his reasonable costs plus an allowance for profit or one in which he must charge a fixed price however inevitable a cost overrun may be; one in which he sells at a fixed price to guarantee his buyer a certain profit or one in which he guarantees his buyer any amount by which the contract price exceeds the market price. Answering these questions requires us to ask what the parties would do, as reasonable people, if they confronted in advance problems which they did not consider. But that is hardly an objection. Any attempt to interpret their contract requires us to ask the same question. [page 530]
* Shannon Cecil Turner Professor of Jurisprudence, School of Law, University of California at Berkeley.
1. Dig. 50.17.185.
2. Reinhard Zimmermann, The Law of Obligations Roman Foundations of the Civilian Tradition 687 (1990).
3. Dig. 18.1.15 pr.; 18.1.57.
4. Dig. 45.1.137.
5. Dig. 22.214.171.124.
6. Dig. 13.6.18.pr.
7. Zimmermann, supra note 2, at 192.
8. Zimmermann, supra note 2, at 192-97.
9. E.g., Bartolus de Saxoferrato, Commentaria Corpus iuris civilis, in Bartolus de Saxoferrato, Omnia quae extant opera to D. 16.3.3 (1615).
10. Gratian, Decretum dicta Gratiani ante D. 13 c. 1 in 1 Corpus iuris canonici (E. Friedberg, ed., 1876).
11. Id. D. 13, cc. 1-2.
12. Glossa ordinaria to Gratian, Decretum to dicta Gratiani ante D. 13, c. 1 (1595).
13. Gratian, Decretum, supra note 9, at D. 50, cc. 49-50; Decretales Gregorii IX 5.10.7-13, in 2 Corpus iuris canonici (E. Friedberg, ed., 1876); Bernardus Papiensis, Summa Decretalium lib. 5, tit. 10, 5-6 (ed. Laspeyres, 1956); Raimondus de Pennaforte, Summa de Paenitentia lib. 2, tit. 1, 3 (eds. Ochoa and Diez, Rome 1976). See Stephan Kuttner, Kanonistische Schuldlehre von Gratian bis auf die Dekretalen Gregors IX 213-27 (Vatican City 1935).
14. Glossa ordinaria to Decretales Gregorii IX 5.36.50 Casus to non custovit; to reddet; to 5.36.6 to voluntarie; to 5.36.9 (1595).
15. Thomas Aquinas, Summa theologiae I-II q. 13 a. 5 ad 1.
16. Id. II-II, q. 88 a. 2 (vow to do the impossible not binding); q. 89, a. 7 (oath to do the impossible not binding).
17. Cajetan (Tomasso di Vio), Commentaria to Thomas Aquinas, Summa theologiae to II-II, q. 113, a. 1 (1698); Domenicus Soto, De iustitia et iure libri decem lib. 8, q. 2, a. 1 (1553); Ludovicus Molina, De iustitia et iure tractatus disp. 271 no. 1 (1614); Leonardus Lessius, De iustitia et iure, ceterisque virtutibus cardinalis libri quatuor lib. 2, cap. 10, dub. 10 no. 70 (1628).
18. Samuel Pufendorf, De iure naturae et gentium libri octo III.vii.2-3 (1688).
19. Lessius, supra note 17, at lib. 2 cap. 7 dub. 6.
20. Molina, supra note 17, at disp. 293 no. 10.
21. Lessius, supra note 17, at lib. 2 cap. 7 dub. 6 (omissio alicuius diligentia, unde sequitur aliquod incommodum).
22. Molina, supra note 17, at disp. 293 no. 12 (deviationem ab eo, quod bonum est, quodque per hominis potuit diligentiam provideri). The conclusion that fault had a different meaning in law and in theology had been drawn by earlier writers as well. E.g., Sylvester Priarias, Summa sylvestrina quae summa summarum merito nuncupatur v. culpa no. 1 (1591).
23. Lessius, supra note 17, at lib. 2 cap. 7 dub. 6 no. 23; Molina, supra note 17, at disp. 293 no. 18.
24. Lessius, supra note 17, at lib. 2 cap. 7 dub. 8 no. 44.
25. Molina, supra note 17, at disp. 293 no. 24.
26. Zimmermann, supra note 2, at 693-93, 695-97; Christian Wollschlager, Die willenstheoretische Unmöglichkeitslehre im aristotelisch-thomistischen Naturrecht, in Sympotica Franz Wieacker 155, 178 (1970).
27. James Gordley, Tort Law in the Aristotelian Tradition, in David Owen, ed., Philosophical Foundations of Tort Law: A Collection of Essays 131, 151 (1995).
28. Christian Thomasius, Dissertatio de usu pratico doctrinae de culparum praestatione in contractibus, in 2 Christian Thomasius, Dissertationum academicorum varii inprimis iuridici argumenti 1006 (1777); Le Brun, Essai sur la prestation des fautes, printed in 2 Robert Pothier, Oeuvres de Pothier 503 (M. Bugnet, ed., 1861).
29. Bigot-Preameneu, Expose des motifs, in 13 P.A. Fenet, Recueil complet des travaux préparatoires du Code Civil 230 (1827).
30. Robert Pothier, Traite des obligations 142 in Pothier, supra note 28.
31. 6 Charles Toullier, Le droit civil français, suivant l'ordre du Code 233-34 (1824-27).
32. 10 Alexandre Duranton, Le droit civil français, suivant l'ordre du code 410-12 (1834-37).
33. 4 Charles Aubry & Charles Rau, Cours de droit civil français 308 n. 28 (1869-71) (art. 1137 par. 1 sets the general standard; other articles create exceptions that can be extended by analogy to like cases); 24 Charles Demolombe, Cours de Code Napoleon 1137 (1854-72) (art. 1137 par. 1 sets the general standard; par. 2 allows the judge to apply it flexibly); 16 Francois Laurent, Principes de droit civil français 219, 231 (3d ed. 1869-78) (art. 1137 par. 1 sets the general standard; other articles create exceptions); 1 Leobon Larombiere, Theorie et pratique des obligations 400-01 (1857) (art. 1137 par. 1 sets a maximum standard of care; other articles reduce it).
34. 4 Aubry & Rau, supra note 33, at 308; 24 Demolombe, supra note 33, at 549-52; 16 Laurent, supra note 33, at 256-7; 1 Larombiere, supra note 33, at 541-42.
35. See 4 Aubry & Rau, supra note 33, at 308; 16 Laurent, supra note 33, at 256; 1 Larombiere, supra note 33, at 541-42.
36. 24 Demolombe, supra note 33, at 550.
37. 24 Demolombe, supra note 33, at 549; 16 Laurent, supra note 33, at 255.
38. 5 Rene Demogue, Traite des obligations en general (1921-33) 1237.
39. 2 Aloys von Brinz & Philipp Lotmar 245 (2d ed. 1892); 2 Heinrich Dernburg, Pandekten 16 (4th ed. 1894); Georg Puchta, Pandekten 302 (2d ed. 1844); Bernhard Windscheid, book review, 2 Kritische Zeitschrift für die gesammte Rechtswissenschaft 106, 118 (1855).
40. 2 Bernhard Windscheid, Lehrbuch des Pandektenrechts 264 (7th ed. 1891).
41. Puchta, supra note 39, at 266.
42. 1 Windscheid, supra note 40, at 101.
43. Anton Thibaut, System des Pandekten-Rechts 252 (6th ed. 1823); Johann Hasse, Die Culpa des Romischen Rechts24 (2nd ed. 1838); Johann Hasse, Essai sur la prestation de fautes par Le Brun, 4 Zeitschrift fur geschichtliche Rechtswissenschaft 189-256 (1820).
44. Karl Arndts, Lehrbuch der Pandekten 86 n. 3 (14th ed. 889); 2 Brinz & Lotmar, supra note 39, at 267 n. 26;1 Karl Vangerow, Leit-faden fur Pandekten Vorlesungen 107 (1847); 1 Windscheid, supra note 40, at 101.
45. Arndts, supra note 44, at 86; 2 Brinz & Lotmar, supra note 39, at 267; Puchta, supra note 39, at 266; 1 Vangerow, supra note 39, at 107.
46. Before the reforms of 2002, courts and commentators had already distinguished "initial impossibility" (anfangliche Unmöglichkeit) from "initial inability" (anfangliches Unvermogen). In the case of "initial inability," the promisor is liable even without fault. The distinction was artificial. It is simply allowed courts to circumvent the Code in cases in which it seems reasonable that the promisor should bear the risk of his inability to perform. Konrad Zweigert & Kotz, An Introduction to Comparative Law 526 (2d ed. 1992).
47. Emmerich in Münchener Kommentar zum Bürgerlichen Gesetzbuch no. 3 to 285 (3rd ed. 1995); Manfred Lowisch in Staudinger, Kommentar Kommentar zum Bürgerlichen Gesetzbuch no. 12 to 285 (13th ed. 1994); Frank Peters in Staudinger, Kommentar no. 13 to 635.
48. Battes in Erman, Handkommentar zum Bürgerlichen Gesetzbuch (9th ed., 1993), no. 2 to 285.
49. Emmerich, supra note 49, at no. 3 to 285.
50. Emmerich, supra note 49, at no. 3 to 285; Peters, supra note 47, at no. 10 to 635; Herbert Wiedemann, in Soergel, Kommentar zum Bürgerlichen Gesetzbuch no. 6 to 285 (12th ed., 1990).
51. Lowisch, supra note 47, at no. 11 to 282; Peter Schlechtriem, Rechtvereinheitlichung in Europa und Schuldrechtreform in Deutschland, 1 Zeitschrift für Europisches Privatrecht 228-29 (1993).
52. Abbott of Westminster v. Clerke, 1 Dy. 26b, 28b, 73 Eng. Rep. 59, 63 (K.B. 1536).
53. Hyde v. Dean of Windsor, Cro. Eliz. 552, 78 Eng. Rep. 798 (K.B. 1597).
54. Williams v. Hide, Palm. 548 (1624).
55. H. Rolle, Abridgment 450, Cond. (G), p. 10 (London 1668).
56. Aleyn 26, 82 Eng. Rep. 897 (K.B. 1647).
57. Samuel Williston, The Law of Contract 1931 (1920).
58. Max Rheinstein, Die Struktur des vertraglichen Schuldverhaltnisses im anglo-amerikanischem Recht 162 (1932).
59. 3 Best & S. 826, 122 Eng. Rep. 309 (1863).
60. Williston, supra note 57, at 1931; Rheinstein, supra note 58, at 175.
61. He cites it in his note to Walton v. Waterhouse, 2 Wms. Saund. 420, 85 Eng. Rep. 1233 (K.B. 1684).
62. William Blackstone, Commentaries on the Laws of England (1766).
63. John J. Powell, Essay upon the Law of Contracts and Agreements (1790).
64. Sir Frederick Pollock, Principles of Contract 365 (4th ed. 1888).
65. Powell, supra note 63, at 161; William Wentworth Story, A Treatise on the Law of Contracts Not Under Seal 463 (3d ed. 1851); Stephen Leake, The Elements of the Law of Contracts 369 (1867); Pollock, supra note 64, at 356; Williston, supra note 57, at 1932.
66. Restatement (Second) of Contracts 261, Comment e.
67. Leake, supra note 65, at 358-9, 361; Louis Hammon, The General Principles of the Law of Contract 414 (1912).
68. 3 Best & S. at 122 Eng. Rep. at (Q.B.).
69. Williston, supra note 57, at 1789.
70. Id. 1979; Zweigert & Kötz, supra note 46, at 546-48.
71. Convention on the International Sale of Goods art. 79.
72. Helmut Heinrichs in Palandt Bürgerliches Gesetzbuch to 242 no. 126 (60th ed. 2001); Gunter Roth in Münchener Kommentar zum Bürgerlichen Gesetzbuch to 242 no. 745.
73. Richard A. Posner & Andrew M. Rosenfield, Impossibility and Related Doctrines in Contract Law: An Economic Analysis, 6 J. Leg. Stud. 83 (1977).
74. Helmut Kohler, Zur okonomishcen Analyse der Regeln über die Geschaftsgrundlage, in Claus Ott & Hans-Bernd Schafter, eds., Allokationseffizienz in der Rechtsordnung 148, 152 (1989).
75. U.C.C. 2-615.
76. Restatement (Second) of Contracts 261.
77. Gratian, Decretum, supra note 9, at C. 22, q. 2, c. 14.
78. Glossa ordinaria to Gratian, Decretum, supra note 12, to furens to C. 22, q. 2, c. 14.
79. Baldus de Ubaldis, Commentaria Corpus iuris civilis to Dig. 12.4.8 (1597).
80. Thomas Aquinas, Summa theologiae II-II, q, 88, a. 10; q. 89, a. 9.
81. Lessius, supra note 17, at lib. 2, cap. 17, dub. 10; cap. 18, dub. 10.
82. Hugo Grotius, De iure belli ac pacis libri tres II.xvi.25.2 (ed. de Kanter-van Hetting Tromp, 1939); Jean Barbeyrac, Le Droit de la nature et des gens ... traduit du Latin to n. 3 to III.vi.6 (5th ed. 1734); ChristianWolff, Ius naturae modo scientifica pertractatum III 504 (1764).
83. Pufendorf, supra note 18, at III.vi.6.
84. 1 Larombiere, supra note 33, at 282-83.
85. Conseil d'Etat, 30 March 1916, D. 1916.III.25, S. 1916.III.17.
86. E.g., RG, 29 Nov. 1921, RGZ 103, 177; RG, 28 Nov. 1923, RGZ 107, 78.
87. Restatement (Second) of Contracts 261, Comment d (1981).
88. 156 P. 458 (Cal. 1916).
89. RG, Sept. 21, 1970, RGZ 100, 129.
90. RG, Nov. 29, 1921, RGZ 103, 177.
91. See Joskow, Commercial Impossibility, the Uranium Market and the Westinghouse Case, 6 J. Leg. Stud. 119 (1977).
92. They have often refused to give relief when performance became more difficult because of a change in prices. 55 A.L.R. 5th 1 n. 21.