Gordley, Legal Reasoning: An Introduction

California Law Review

MARCH, 1984

71 Calif. L. Rev. 138

THE BERKELEY LAW CENTENARY LECTURES: Legal Reasoning: An Introduction.

James Gordley

In the hundredth year in which law has been taught at the University of California in Berkeley, the School of Law celebrated by inviting a judge, a philosopher, and a historian to speak about the rule of law. As it happened, the same problem seemed to haunt each speaker. Courts, it would seem, cannot decide all their cases by watertight argument from established law. How then do they decide the remainder?

The judge was Jon O. Newman of the United States Court of Appeals for the Second Circuit. n1 He described two ways in which a judge might decide a case: by "rational thought" from definite "principles," and by his personal "preferences" or "political, economic, or social values." According to Newman, either description is misleading. Rational thought from principles will not suffice in all cases, yet a judicial decision must somehow be more than an expression of personal preferences. Newman concluded that there must be a middle ground of some sort between these alternatives.

The philosopher was David Lyons of Cornell University. n2 He began by describing the view that only some cases can be decided by watertight argument from established law. According to Lyons, one who holds this view should not jump to the conclusion that because only some cases are "easy" in this sense, the "hard cases" are decided merely by the judge's preferences. Even when established law has given out, a judge can still "justify" his decision by an appeal to "principles or policies" that are "fair," "either because they themselves are sound or because there are good and sufficient reasons for taking such principles or policies as standards for determining what should or must be done." n3 If a judge can do that, then there is a difference between genuinely justifying a decision in a hard case and merely expressing a preference. Moreover, if there is a difference, and if a judge is dutybound to look to these principles and policies in hard cases, then in a sense these principles and policies are part of the law. And if that is so, Lyons concludes, perhaps the easy cases are not so easy after all. For if one can look for genuine justifications of decisions in hard cases, then one can ask whether the decisions made by following established law in easy cases are justified by any fair principles or policies, including those that underline the process by which the law was made. If not, then there is no justification for following the established law and consequently no reason why either judge or citizen is bound to follow it. For Lyons, then, the way out of Newman's dilemma is through something called "justification," and once we acknowledge its existence, it becomes the measure for decisions in easy cases as well.

The historian was Harry N. Scheiber of the University of California, Berkeley. n4 Just as Lyons has rebelled against a philosophical tradition in which whatever is not done by rule and logic must be done by preference, so Scheiber has rebelled against a historical tradition in which the key to doctrinal change lies in the preferences of judges. Professor Hurst thinks that nineteenth century judges protected the "vested rights" of private persons selectively so as to foster a "release" of private, entrepreneurial "energy." n5 Professor Horwitz believes these judges "transformed" the law to enable "emergent entrepreneurial and commercial groups to win a disproportionate share of wealth and power." n6 According to Scheiber, neither account fits the historical evidence. Nineteenth century judges developed not only a doctrine of private vested rights but a positive doctrine of public rights, and courts used both doctrines, sometimes to further economic growth and commercial interests but sometimes to stand against them. More fundamentally, however, Scheiber argues that these accounts fail to "take doctrine seriously." For Scheiber, doctrine is neither a smokescreen for preference nor a deadweight that preference drags around. Doctrine is something to which judges respond in more or less fruitful ways.

All three lecturers, then, feel uncomfortable with a conventional view in which legal decisions must be made either by a mechanical argument from established rules or else by preference. But they have moved away from this view in three different ways. Newman wants to find a middle ground. Lyons speaks of justifications that transcend established law. Scheiber regards doctrine itself as something fluid and fertile.

In this Introduction, I would like to contribute to the discussion by showing that the conventional view establishes a false dichotomy. Part I will discuss how legal reasoning should proceed when the law is as clear as it can be made. Even then, we will see, cases could not be decided by mechanical argument save in rare instances. Part II will consider methods of legal reasoning when rules are vague or lacking. We will see that even then, genuine interpretation is possible and cases need not be decided by an interpreter's personal values or preferences. Part III will consider forms of legal reasoning that do not proceed from authority. We will see that even when a judge is not guided by authority, it is misleading to say he decides a case by his own values or preferences. Finally, in Part IV we will consider one of the reasons people have found this false dichotomy attractive: it fosters the illusion that one can give a value-neutral account of law.

I. LEGAL REASONING UNDER IDEALIZED CONDITIONS: THE ILLUSION THAT CASES CAN BE DECIDED MECHANICALLY

An easy case, Lyons says, is "one in which the law is clear enough so that it can be decided in a more or less 'mechanical' way, by applying relevant rules in a logically rigorous argument" through "logically deductive methods." n7 I will try to show that such mechanical methods are unsatisfactory even when the law is as clear as a legal authority can make it: that is, even when the authority has clearly in mind what he wishes to accomplish, formulates clear rules to accomplish it, andcommunicates these rules so as to be understood with a minimum of difficulty. Even under these idealized conditions a legal authority would not want to be interpreted mechanically, and thus it would be as silly to try to do so as it would be to interpret an English-speaking authority with a Finnish grammar. At the same time I will try to identify the methods by which an authority would wish to be interpreted under those idealized conditions. By understanding those methods I believe we will be in a better position to consider legal reasoning in the less idealized case when the law is obscure than if we begin with a false dichotomy between easy and hard cases.

One reason a legal authority would not want to be interpreted mechanically is linguistic. For mechanical interpretation to work, he would have to say exactly what he meant. His interpreters must be able to reason that because he used a certain word, he meant exactly such-and-such, for had he meant anything else, he would have used a different word.Language, however, does not come stocked with expressions that mean exactly what one wishes to say. When people speak they draw on a stock of signs that are familiar to their interpreters because other people have used those signs to say other things. This stock is like a chest of all-purpose tools or a supply of lumber cut to standard sizes. To build a message, one must either use the boards and tools in stock as best one can or else follow a more cumbersome procedure, building new tools and refashioning boards. If, for example, the city council is disturbed because cars have been roaring through the park on Sundays, it will find that there is no word in the language that describes exactly what it would like to prohibit. n8 Therefore, it can either draft into service the all-purpose word "vehicles" and prohibit vehicles in the park, or it can tediously recraft the language to specifically exempt World War I tanks set up as monuments, police cars and ambulances on emergency calls, Piper Cubs making forced landings, and so forth. A city council that wishes to be easily understood will realize that its every ordinance should not sound like a bond indenture. Indeed, if master sergeants explained to privates exactly when and how to make every move, not only would wars end before the first orders were issued, but those orders would be understood only by privates whose minds worked like computers, infinitely patient and responsive to precise programming.

A legal authority who wishes to be easily understood must count on his interpreters to know the sort of thing he might have in mind, given the problem he is addressing. But then he cannot be interpreted mechanically. If he uses a word that most people use to mean one thing, the interpreter must consider the possibility that he is using the word to mean something else. If the language contains an equally easy expression that is a better approximation of that something else, then the interpreter can dismiss that possibility as long as he trusts the linguistic abilities of his authority. If not, the interpreter has to ask which is more plausible: that his authority meant what most people mean when they use the word, or that he meant something approximated by the word and used no more exact expression because of the cumbersomeness of recrafting the language. This type of interpretation is not mechanical, and an authority who wishes to be easily understood would not wish it to be mechanical.

There is a more basic reason why a legal authority would not wish to be interpreted mechanically. Unless he regards his own rules as perfectly arbitrary, he will have some reason for preferring them to other rules. He will believe his rules are justified by some criterion of justice or rightness or usefulness, or, in aberrational cases, by some criterion having to do with personal benefit or class bias. If he believes in such a criterion, however, he will not want his rules interpreted mechanically except in very special cases. A legal rule states that a certain legal result will follow when a person acts a certain way under certain circumstances. To wish to be interpreted mechanically, the legal authority would have to believe that this legal result invariably should be reached when the circumstances he has enumerated are present. That would be an odd belief, for some unenumerated circumstance might arise that would call for a different result according to the criterion that he had in mind when he initially established the rule.

Such circumstances will sometimes arise if the legal authority is making rules as a means to achieving some ends or purposes, or to put it another way, if his criteria for making rules are teleological. n9 If they are, either the rules will specify how a particular purpose is to be accomplished, or they will determine which of several purposes should be pursued at the expense of others. In neither case will he want these rules to be followed invariably.

If rules are made for a particular purpose, some obstacle can always arise that requires the rules to be varied. No set of invariable rules could even tell one the best way to make a trip downtown, for the weather might change, one's health might change, there might be strikes, blocked roads, holiday parades, and so forth. The best battle plan or marketing strategy would have to adjust to an unlistable number of circumstances, and a good general or executive knows when and how to make the adjustments. The only way rules to achieve a particular purpose could invariably be right would be if the number of circumstances requiring a deviation were finite and each could be dealt with in advance by fixed rules rather than, for example, by a general procedure which must be retailored on the spot.

Similarly, rules cannot invariably be correct if they determine which of several purposes should be pursued at the expense of others. One cannot make that determination by asking in the abstract which purpose is most valuable. One might as well decide that one can never play squash because in the abstract that activity is less valuable than listening to Beethoven. One can ask what contribution the pursuit of each purpose will make to some general goal: an ideal of the good life, in the case of squash or Beethoven; victory, in the case of a general hesitating between military objectives; or profit, in the case of a business executive whose marketing strategy requires higher production costs. But then one is conceiving of the pursuit of each purpose as a means to that ultimate goal or purpose, be it profit, victory, or an ideal of life. And as we have just seen, when one is pursuing any sort of purpose, an invariable rule will not invariably be correct. Some circumstance might make either purpose more or less valuable from the standpoint of the ultimate goal, or the attainment of either purpose more or less difficult.

Thus, a legal authority would not regard his rules as invariably correct were he making rules by teleological criteria. Nevertheless, whether rules of law normally are or should be made by such criteria has been a matter of dispute. Some people have thought that the most basic rules of law rest on supposedly neutral principles such as liberty or equality. These principles are supposed to be neutral in that they do not presuppose that one human purpose can be more valuable than anther. In that sense these principles are nonteleological.

The attempt to derive rules from such supposedly neutral principles has a notorious habit of leading nowhere. For if no human purposes were deemed more valuable than others, there would be no way to decide what sorts of liberty or equality to protect. Legal rules typically settle conflicts between one citizen's pursuit of his purposes and another's pursuit of his own. Thus any rule will limit the freedom or ability of each party to do what he wants. It makes no sense to ask whether a particular rule leaves each party with a like liberty or equal ability to pursue his own purposes. For nearly always the purposes of the parties will be different or of different importance to them or will be pursued under different circumstances. If there is no way to ask which of several different human purposes is more valuable than another, then there is no way to ask whether the freedom or ability a rule leaves one party to pursue his purposes is like, or equal to, or as important as, the freedom and ability it leaves another to pursue quite different purposes. There is no way to ask, for example, whether a rule which allows a landowner to build or prevents him from dynamiting treats equally or affords a like liberty to nonlandowners or landowners who do not wish to build or dynamite. If the law is to determine which citizens' purposes are to be pursued at the expense of others', it must do so, then, by criteria which concern the value of different human purposes. n10

To say these criteria are teleological in this sense is not tosay that the rules based on these criteria will call for a direct decision about the value of one purpose compared to another. The rules we have been discussing typically do not. Rules of welfare and taxation govern the amount of resources each citizen will have available for private purposes without asking too many questions about how he will spend them. Rules of contract allow citizens to cooperate for private purposes without asking much about what those purposes are. The same can be said of rules of property and tort, which prevent one citizen's pursuit of his purposes from interfering unduly with that of others. A property owner typically can use his property as he wishes whether or not the use is valuable in anyone else's eyes. Rules such as these, however, could perfectly well rest on teleological criteria. Legal authorities might make these rules precisely so that citizens will pursue more valuable rather than less valuable activities. They might believe, however, that citizens themselves normally are better judges of which activities are valuable than are state officials, and that these rules avoid the disputes that a direct decision about the value of an activity would occasion. Indeed, as we have seen, if legal authorities were agnostic or neutral about whether one activity or purpose could be more valuable than another, they would be unable to decide where one citizen's freedom and ability to pursue his purposes should leave off and another's begin.

If, however, the rules which govern human conduct are based on criteria concerned at least in part with the value of what human beings do, then there must be instances in which conduct is perfectly correct according to the rules and flagrantly wrong according to the criteria. The rules cannot invariably yield results justified by the criteria on which they rest. A legal authority could deal with that problem in two different ways. He might make a progressively more complex series of rules to minimize the divergence between the rules and the criteria, almost as one might approximate a circle by drawing polygons with progressively more sides. For example, while it is generally true that a property owner can do what he wishes with his own, an owner under age cannot, an owner who is mentally incapacitated cannot, an owner with creditors cannot, and an owner who merely wishes to irritate his neighbors cannot. Of course, the more of these rules one makes, the more one complicates the legal system. Alternatively, a legal authority might allow some direct consideration of which of two activities is more valuable. In property law, doctrines of necessity, nuisance, negligence, and neighborhood change do just that. Of course, the criteria will be harder to understand and apply directly and their application may lead to more disputes than a clear rule would. But the rules themselves, no matter how complex they may be, cannot invariably give results which are correct by the criteria by which the rules were made.

A legal authority might sometimes want his rules to be applied invariably even in cases where the result is wrong. He might have decided that making more rules to avoid these results would unduly complicate the law, and that to apply his criteria directly would cause too much error, uncertainty, and dispute. But it would be peculiar to assume that a legal authority wants all of his rules to be followed invariably. To have considered the wrong results that all his rules might produce and the disadvantages of deviating from the rules to avoid them, he would have had to do a most unlikely amount of considering. More implausibly yet, he would have to have always reached the conclusion that the rules should be followed no matter what. If he has not done that, however, he will not wish his rules to be followed invariably or interpreted mechanically. He will want his interpreters sometimes to supply new rules that better fit his criteria under circumstances he has not considered and sometimes simply to reach the right result by applying his criteria directly.

Indeed, his interpreters then have no right to interpret their authority mechanically and claim any validity for their result, since that is not how he wished to be interpreted. What they must do instead, when confronted with a set of rules, is to ask by what criterion the rules were made. With this criterion in mind, they can be alert for cases that would come out wrong according to the criterion though right according to the rules. When these cases arise, they can ask whether their authority considered the advantages and disadvantages of deviating from the rules to allow for the special circumstances of that case. If he did not, they can consider the matter for themselves and act accordingly. No step of this reasoning process is mechanical, but no step of it is arbitrary. It is no more arbitrary for a legal interpreter to look at rules and infer the criterion on which they were made than for an engineer to look at mechanical drawings and infer the purpose of an unknown machine or for a soldier to infer the military objective behind a certain order. It is no more arbitrary for a legal interpreter to deviate from rules to allow for circumstances the rules were not designed to take into account than for a junior engineer to modify the rough drawing of a machine which his senior has given him. Suppose a sergeant were ordered to lead his squad up what appeared from a distance to be an unoccupied hill and to set up machine guns to command a road. If the hill later proved to be a tower yet offered an equally good site for machine guns, even a minimally intelligent sergeant would see that he ought to occupy it. If, on the other hand, it proved to be a hill already occupied by an enemy division, he would see that he should not lead a suicide attack.

We have seen, then, that if the criteria for making rules are teleological, the rules will never give exactly the results called for by the criteria on which the rules are based. Therefore, interpretation of the rules cannot be purely mechanical.

In reaching that conclusion, we have constructed an account of how legal reasoning ought to proceed under idealized conditions.We supposed that the legal authority has formulated a set of clear rules tailored to criteria that he has clearly in mind. Knowing the rules, his interpreter can infer the criteria that has clearly in mind. Knowing the rules, his interpreter can infer the criteria by which they were formulated; knowing the criteria, he can discover discrepancies between the results that the rules and the criteria call for; knowing the discrepancies, he can decide whether to tolerate them, to elaborate further rules and thereby complicate the legal system, or to apply the criteria directly to individual cases despite the risk of error and uncertainty.

To say that an interpreter can draw these inferences under these idealized conditions does not mean that he can draw them correctly or easily. Indeed, he could draw them easily only if it were obvious that the criteria themselves call for one result while the rules call for another. But if it were obvious, the argument just given would make no sense, for the interpreter could apply the criteria directly and not bother with rules. The argument assumes that rules are easier to understand, apply, and follow than the criteria on which they are based. Thus, even when the rules are clear, cases may be hard or easy depending on how difficult it is for an interpreter to draw these inferences.

II. LEGAL REASONING UNDER LESS IDEALIZED CONDITIONS: THE ILLUSION THAT CASES MUST BE DECIDED BY PERSONAL PREFERENCE

To show that mechanical methods of interpretation are unsatisfactory, we assumed that the law was as clear as an authority could make it. Let us now assume the law is obscure. The rules the legal authority has provided are too fragmentary, partial, or imprecise to describe the circumstances that normally call for a given legal result; or the authority has not provided rules but has merely indicated the results he deems appropriate in particular cases. Moreover, he has acted by criteria that he himself does not clearly understand but only vaguely grasps. We will now construct a modified account of legal reasoning to explain how an interpreter ought to proceed under these less than ideal conditions. We will see that even when the law is obscure, genuine interpretation is possible and cases need not be decided by personal preference, as long as the authority has been acting by some criteria.

Our assumption is that the authority does act according to some criteria, however vaguely grasped. When he prohibits murder or enforces contracts, even though he cannot make clear rules distinguishing murder from self-defense or explaining which contracts to enforce, and even though he cannot explain the criteria on which he believes the law of crimes and contracts should rest, nevertheless he does not act by pure caprice. He acts in response to something, be it a sense of justice or, in the aberrational case, class bias or personal interest. Indeed, if he were acting by no criteria of any sort, his actions would be disconnected, random, and barely intelligible. If he has not acted by pure caprice, however, an interpreter ought to be able to get a sense of what moves him. Ultimately a sufficiently skilled and patient interpreter ought to be able to identify what his authority's criteria are. In the following account, however, we will assume that the interpreter has not been so successful, and that while he has a sense of what these criteria might be, he has not grasped them more clearly than his authority. We will consider how such an interpreter should deal, first, with the results that his authority deems appropriate in particular cases, and then with whatever fragmentary, partial, and imprecise rules his authority has laid down.

A. Reasoning from One Case to Another: Analogy and Distinction

If the interpreter has a vague sense of the criteria that underlie the result in a given case, he can reason from one case to another by two methods long employed by common law judges: analogy and distinction. He can compare a new case to a decided case and ask how likely it is that the differences between them would matter to the result according to the criteria he has vaguely grasped. If he decides that it is likely that the differences do not matter, he analogizes the two cases and decides the second case the same way as the first. If he decides that the differences probably do matter, he distinguishes the cases and does not regard the first as authority for a similar result in the second.

These methods are exemplified by the well-known common law cases of Ploof v. Putnam n11 and Vincent v. Lake Erie Transportation Co. n12 which deal with the so-called doctrine of necessity. According to that doctrine, necessity entitles one person to use the property of another. Ideally, a person interpreting this doctrine could determine the ultimate criteria on which the rules of private property rest and then explain in terms of these very criteria why the rules of property sometimes admit an exception. We would then know what constitutes "necessity." Some older jurists actually tried to construct an account of necessity along these lines. n13 But that is not how the judges in Ploof and Vincent reasoned.

In Ploof, plaintiff tied his boat to defendant's pier to escape the dangers of a storm and defendant cut him loose. The judge awarded plaintiff damages for losses suffered in the wreck of his boat, and supported that result by citing, inter alia, cases in which a defendant, to save his own property, had entered plaintiff's land, and the plaintiff had lost an action brought for trespass. The implicit premise of the judge's argument is that, whatever criteria are involved, it would not make sense to deny the landowner legal redress for trespass while permitting him the more drastic remedy of self-help. The argument works by analogy.

In Vincent, defendant tied his boat to plaintiff's pier to escape a storm, plaintiff did not cut him loose, and the pier was damaged by the boat. The judge allowed plaintiff to recover damages, noting that here, unlike Ploof, the boat owner had destroyed another's property to save his boat. The implicit premise is that by whatever criterion ultimately underlies the doctrine, that circumstance does call for a different result. Thus the argument works by distinction. The judge in Vincent then fortified his conclusion by another argument from analogy: had the boat owner taken a coil of rope belonging to plaintiff to secure his boat, he surely would have had to give back the rope when the emergency had passed, and so he ought to pay for the damage to the dock.

Arguments like these assume that when an interpreter knows the appropriate result in a particular case, but cannot identify the criteria that make that result appropriate, he can still know enough about what the criteria might be to sense whether a given circumstance should matter. If that assumption is right, then analogizing and distinguishing cases in this way are valid methods of reasoning, although they seem puzzling and disappointing because one cannot spell out the argument fully. One cannot explain why, according to criteria one cannot state, a given circumstance should or should not matter.

Another way an interpreter could work with particular cases is to use them to formulate rules, although, of course, these rules are likely to be imprecise since the interpreter does not know precisely the criteria behind them. How this might be done will be discussed as we consider how an interpreter might work with the imprecise rules he is given by his authority, since the way such rules should be interpreted has a good deal to do with they are initially formulated.

B. Formulating and Applying Rules When Criteria Are Obscure

Suppose now that the authority has given the interpreter a set of imprecise and partial rules as starting points. The interpreter is apt to have two problems: he will still have to decide whether special circumstances warrant an exception to them; moreover, he will have to decide what the normal circumstances are to which the rule does apply.

1. Dealing with Special Circumstances

The interpreter can solve the first of these problems by the methods of analogy and distinction just described. He can look for circumstances present in the case before him that are neither enumerated in the rule nor present in cases to which the rule is ordinarily applied. He can then ask whether, by his authority's criteria, these circumstances are likely to matter to the result. If not, he can apply the rule by analogy to the case before him. If so, he can distinguish that case from those falling within the rule. This kind of reasoning is exemplified by Katz v. Walkinshaw, n14 a California decision praised by Scheiber, n15 in which the court refused to apply the common law rule on percolating water that allows landowners to pump what they please. The court argued that the rule was established in the wet climate of England and would cause a subversion of justice in arid California since it would allow rich people with powerful pumps and deep wells to control the water supply. The implicit premise is that this harsh result is not warranted by whatever criteria underlie the common law rule. The argument works by distinction. The argument would be stronger, of course, if the court told us what criteria ultimately do underlie the rules that govern property in natural resources. But assuming a judge can have some sense of what these criteria are, though he cannot state them, the argument has force.

2. Dealing with Obscurity

The interpreter's other problem might be, not to identify the special circumstances under which a rule should not apply, but the normal circumstances under which it should. Under idealized conditions, a legal authority could identify these circumstances precisely. He could make a rule that stated exactly the circumstances that call for a given legal result according to his criteria absent special circumstances. When he has done so, we shall say that he has identified the "determinative circumstances." Every case in the class or category defined by the presence of those determinative circumstances should be decided as the rule dictates, unless special circumstances arise.

There are two ways, however, in which even this seemingly ideal rule could be unclear, and we should note them before considering how obscurity can creep in under less idealized conditions. First, the rule might be one rule too many. It might be one of a series of rules elaborated, as described earlier, to get a progressively better fit between the results that rules and criteria call for. Thus the rule might be confusing, not because it itself is obscure, but because having a large number of rules is confusing. If so, as mentioned, a legal authority might choose not to enact it but to allow his interpreters to apply his criteria directly. In so doing, he would purchase one sort of clarity at the cost of another, since instead of identifying the determinative circumstances -- the very circumstances that matter to a result -- he would be making general hortatory statements about what to seek or what to avoid in deciding cases.

Second, the determinative circumstances might be abstract and hard to understand. A child can understand a rule against crossing the street before he can understand one that tells him to look both ways and gauge the speed of oncoming traffic. A bridge player can understand the rule "lead fourth from your longest and strongest suit in notrump" more easily than a complex statistical formula that would identify the right lead. Thus, when the determinative circumstances are difficult to grasp or remote from the ordinary experience of those who must follow the rule, a legal authority might choose to enact a simpler rule that does not quite capture these circumstances and does not invariably yield the right result, but that is easier to apply. People may reach the right result more often following the simpler rule than trying to follow the harder one.

We have seen how rules can be somewhat unclear even under idealized conditions. We can now identify different ways in which rules can be obscure when conditions are not ideal and the legal authority has little more to go on than criteria vaguely grasped and certain results which he deems appropriate in particular cases. In that event, he will usually be unable to identify the determinative circumstances.

One way he could respond is to enact a series of rules that are merely sterotyped or generalized versions of the particular cases in which he can identify a result he deems appropriate. These rules will not identify the determinative circumstances but will merely list facts of cases which seem to him significant. Consequently these rules will be rather fragmentary. We will call them repertorial rules.

Another way he could respond is to try to approximate an ideal rule. An ideal rule would define a class of cases in which a particular result was appropriate by identifying the determinative circumstances. Now an authority might see that cases are similar without knowing how to define the class to which they belong; he might know that a certain circumstance is critical in one or more classes of cases but not know the other circumstances that would peermit him to define any of these classes; or he might know a certain criterion is relevant without being able to identify the other relevant criteria. In the first situation, he might frame what we will call a classificatory rule: a rule that groups like cases together without identifying the determinative circumstances of these cases. In the second situation, he might frame what we will call a principle: a rule that identifies one of the determinative circumstances which in combination with others would call for a particular result but fails to identify the others. In the third situation, he might frame what we will call a hypothetical rule: a rule that identifies what would be the determinative circumstances if the particular criterion he can identify were the only one relevant.

Another possibility is that an authority might not try to identify the determinative circumstances even approximately. As we saw earlier, even when an authority does know what these circumstances are, sometimes he will still allow his criteria to be applied directly, confining himself to some general hortatory statements about what to seek or what to avoid. Sometimes he will make a rule that is simple to follow even though it does not capture the determinative circumstances. If a legal authority does not know what these circumstances are, he has an additional reason for pursuing one course of action or the other since he cannot be accurate about these circumstances anyway. The hortatory statements he makes if he pursues the first course of action we will call maxims. The simple rules he enacts if he pursues the second we will call pragmatic rules. We will use these terms only when one of the reasons an authority acted as he did was his own uncertainty about the determinative circumstances. In that event, it is possible that the law is more obscure than it would need to be if those circumstances had been known, and as we will see, an interpreter must take that possibility into account.

Having identified these six types of obscure rules, we will now see how each of them could be interpreted provided the interpreter has some grasp of his authority's criteria, and also how each could be misinterpreted.

a. Repertorial Rules

We saw earlier how, if an authority had merely indicated the results he deemed appropriate in particular cases, an interpreter could reason from one case to another by methods of analogy and distinction. He would ask what circumstances do and do not seem to matter by the criteria which he vaguely grasps. These same methods would thus permit either the legal authority or his interpreter to begin with the results in particular cases and arrive at what we are calling repertorial rules. He could make a general statement of the circumstances of a particular case which seem to matter, much as a law student does when he briefs a case.

The rules at which he arrives, however, would not identify the determinative circumstances of the case that matter from the standpoint of the criteria. The circumstances of a case can be described in an indefinite number of ways. One who makes a repertorial rule lists what seems significant but does not know how to redescribe the circumstances of the case so as to capture precisely those aspects of it which matter. For example, he might be groping for rules governing strict liability in tort, and it might strike him as significant in one case that the defendant was blasting, in another, that he was storing dynamite, in other cases, that he was building reservoirs or keeping lions. By these observations, he can arrive at repertorial rules that explain strict liability by listing these activities. But if the rules of strict liability ultimately rest on common criteria, then these different activities must have something in common which is what really matters from the standpoint of these criteria. Repertorial rules fail to capture these determinative circumstances, much as the rules of a seventeenth century doctor for preventing scurvy by drinking lime juice, eating sauerkraut, and so forth, fail to capture the common circumstance that scurvy is prevented by vitamin C. Repertorial rules are therefore fragmentary.

An interpreter who has some grasp of his authority's criteria can nevertheless interpret a repertorial rule. He can treat the rule as a mere summary of circumstances noted as important in particular cases and proceed by the methods of analogy and distinction already discussed. Or, as will be seen, he can use repertorial rules as a basis for formulating other types of rules that more closely capture the determinative circumstances.

What he should not do, however, is treat a repertorial rule as though it already had picked out the determinative circumstances. He should not conclude, for example, that because his rules impose liability for "blasting," "storing dynamite," and so forth, that all people who conduct the activities on this list are liable or that only such people are liable. Until he has identified the significant feature of these activities, he cannot rule out the possibility that some blasting does not share that feature or that some activity not on the list does share it. This point rarely escapes judges working with judge-made rules. Having made the rules themselves, they are sensitive to how much and how little information they contain. They more often miss the point when a legislature has enacted a repertoire of rules. That mistake may be responsible, for example, for the refusal of German courts to extend strict liability beyond a shopping list of activities provided by the legislature n16 or the refusal of French courts to recognize more than a legislatively enumerated handful of cases in which a price term is unconscionable. n17 The fact that a legislature makes such a list is not evidence the list is exhaustive but rather that it is open ended, for listmaking is an admission that one cannot say just what significant features the listed circumstances have in common. Thus the California Supreme Court, confronted with a legislative list of seventeen circumstances in which easements can be created, wisely presumed that the legislature had not produced the complete list that the court knew it could not have produced itself. n18

b. Classificatory Rules

If our account of repertorial rules is correct, then the determinative circumstances can be the same for a number of repertorial rules or for a number of particular cases. A legal authority or his interpreter might suspect strongly that this is so without being able to identify these common features. He might therefore find it useful to make what we will call a classificatory rule: one that groups repertorial rules or particular cases together in the belief that the determinative circumstances are the same but that does not identify them with any precision. At the most, the language of a classificatory rule merely suggests what these circumstances are, the way the vague word "ultrahazardous" suggests what blasting, keeping lions, and other such activities have in common that calls for strict liability. Or the language of a classificatory rule might merely direct attention to an important group of cases without suggesting the circumstances that distinguish this group. The expressions "changed circumstances" or "frustration of purpose," for example, direct attention to groups of cases in which a change of circumstances or the frustration of a purpose will be a defense to a suit on a contract. But since circumstances will have changed and purposes been frustrated whenever anyone wants to escape from a contract, these expressions do not suggest what distinguishes these cases from others.

An authority or interpreter who arrives at a classificatory rule in the way just described will really have reasoned by analogy on a larger scale. In arriving at a repertorial rule by analogy he might have seen that the circumstance that the defendant was blasting was significant; in arriving at a classificatory rule he might have seen that the circumstance that one defendant was blasting was significant in the same way as the circumstance that another was keeping lions. Conversely, he might sort cases he initially lumped together into distinct groups and so arrive at classificatory rules by the method of distinction. Prosser sorted invasion of privacy cases in this way into the four separate torts of commercial appropriation of name or image, disclosure of embarrassing private facts, false light, and intrusion into seclusion. n19

Classificatory rules can be interpreted by merely working with the cases or repertorial rules they attempt to classify by methods of analogy and distinction. But by focusing attention on different factual situations in which the determinative circumstances may well be the same, these rules sharpen an interpreter's sense of what these circumstances are. Even though he cannot identify and describe these circumstances, by seeing cases of blasting, keeping lions, building reservoirs, and so forth, set side by side, he will have a better feel for what other activities should give rise to strict liability. Similarly, by reading through the frustration of purpose cases he may gain some insight into the sorts of frustrated purposes which should constitute a defense, though he will be very lucky if he can put that insight into words. That is what one would expect if an authority or his interpreter were proceeding by criteria vaguely grasped. He could detect similarities and differences, the significance of which he could not fully explain.

What an interpreter should not do, however, is treat these rules as more than tentative classifications. It would be a mistake to think that a rule recognizing frustration of purpose as a defense means that absent special circumstances relief should be given whenever a contracting party's purposes are frustrated. It would also be a mistake to think that if one stares hard at the phrase "frustration of purpose" or attempts to define it or qualifies it with words such as "basis" or "substantial," one will achieve greater clarity. For the phrase may do no more than direct one's attention to an important and similar group of cases. The determinative circumstances that define this group may have to be described in some quite different way.

Similarly, since these defining circumstances have not been identified, a classificatory rule may classify wrongly, and an interpreter should be willing to scrap the rule when he is convinced that it lumps cases together that require different answers or segregates those that do not. It is fortunate for the common law of negligence, for example, that courts finally gave up the attempt to distinguish direct from indirect injuries. In the modern view, at least, the distinction is artificial. Quite possibly, the distinction between easements and covenants is equally artificial, as is the distinction between the torts of disclosure of embarrassing private facts and false light and the tort of defamation. In any event, the reason for preserving these categories is no stronger than the reason for supposing that they make good classificatory rules. An interpreter who scraps them because they do not help identify the determinative circumstances would not be changing the law but reclassifying it to understand it better.

c. Principles

A classificatory rule groups cases in which the determinative circumstances may well be the same. Sometimes, looking at that group of cases, one can identify one of these determinative circumstances but not the others. One will then have arrived at what we will call a principle. Since a principle identifies one of the determinative circumstances but not the others, it does not enable one to define a class of cases in which a particular legal result is to be reached. That circumstances may appear in other groups of cases that come out differently. It may also appear in several groups that come out the same way, but in which the other determinative circumstances are different.

In this sense, "rules" such as no one should profit by his own wrong and no one should be enriched at another's expense are principles. n20 They identify clearly or less clearly a circumstance that does seem to be critical in important groups of cases. Nevertheless, these principles do not themselves identify a group of cases in which a certain result is warranted absent special circumstances. They do not mean that absent special circumstances a person who has profited by his own wrong or been enriched at another's expense is liable. If that were what they meant, they would simply be incorrect. Principles only define a group of cases when one can identify the other determinative circumstances: for example, the sort of wrong by which one has profited or the way in which a person has been enriched at another's expense. In contract law, promissory reliance is a principle in this sense. Not all promisees who change their position in reliance on a promise are given relief, and no one is quite sure what other circumstances must be present. The second Restatement of Contracts, for example, says that relief should only be given when "justice requires." n21

To interpret a principle, an interpreter may need to clarify the principle itself, which may be expressed rather vaguely. He can do that by looking at the range of cases to which the principle applies and seeing if he can find some better way to express the circumstance that the principle identifies. To apply the principle, the interpreter will also need to have some grasp of the other determinative circumstances that the principle does not mention. Even if he cannot identify these other circumstances, he can still proceed by the methods already described. He can devise classificatory rules, splitting the cases into groups in which the relevant circumstances seem to be the same, or repertorial rules, listing the circumstances of these cases that strike him as important.

One mistake an interpreter might make however would be to treat a principle as though it were a rule that did mention all the determinative circumstances. It would be a mistake, for example, to try to give a remedy whenever someone has profited by his own wrong or whenever a promisee has relied to his detriment. The opposite mistake, currently more common in America, is to think that a principle can be debunked or shown to be useless merely by pointing out that it does not mention all the circumstances that matter to a decision. It is true that we do not give relief whenever a promisee has relied, but it does not follow, as Grant Gilmore thought, that when the drafters of the first Restatement of Contracts recognized the principle of reliance, "no one had any idea what the damn thing meant." n22 They knew that reliance is one of the circumstances that matter, and that can be useful to know.

d. Hypothetical Rules

Just as it is possible to identify one but not all of the determinative circumstances that an ideal rule would specify, so it is possible to identify one but not all of the criteria on which such a rule would rest. A particular purpose or policy might strike one as desirable in itself, such as the promotion of economic efficiency.Moreover, one might see that, were we not to respect a given purpose or policy, an entire branch of law would be meaningless. For example, contract law would be meaningless unless the enforcement of private agreements were among our purposes. One can therefore begin with the criterion one can identify and make what we will call a hypothetical rule. Such a rule identifies what would be determinative circumstances if this criterion were the only one that mattered. For example, those who study law and economics often begin by supposing our one goal is efficiency, and then show what rules that criterion calls for. One traditional approach to contract law began by noting that private agreements must be kept, and tried to derive the rules of contract law from that criterion. n23

Properly interpreted, these rules mean only that if the criteria one has identified were the only ones that matter, then certain circumstances would call for a certain result. The rules are only as good as this hypothesis. One can examine the hypothesis itself by asking why these criteria should be the only ones relevant, or by seeing how much fit there is between the hypothetical rules and rules derived in other ways. A hypothetical rule might explain why a certain principle has identified one of the determinative circumstances; it might help one identify those circumstances in a group of cases isolated by a classificatory rule; or it might explain the particular results listed in a repertorial rule. To the extent it does any of these, the hypothesis seems trustworthy. Even when the hypothesis does not seem very trustworthy, such rules can still have their uses. There may be a domain of rules for which the hypothesis works quite well even though applied elsewhere it seems to result in nonsense. Moreover, these rules can alert an interpreter to the fact that by deviating from them he will be compromising the criterion on which they are based. The interpreter can then ask whether other criteria might be involved that would warrant the deviation.

The danger of hypothetical rules is that the interpreter may believe too readily that the criterion he has identified and on which his rules are based is the only one relevant. He might imagine that because the criterion he has identified is good, the rules derived from it must be good, whether or not there are other relevant criteria. If that were so, and national independence were good, the most monstrous acts committed to win a war would be good; if life were good, any act that imperils it, such as crossing the street, would be bad, and so forth.

Or he might imagine that because a field of law would be meaningless without his criterion, the rules of that field should be derived from it alone. For example, he might observe quite rightly that contract law would be meaningless unless we valued the keeping of private agreements, and then imagine that contract law must be defined as the enforcement of private agreements and that all its rules must be derived from this criterion alone. One might as well argue that because our notion of a saw would be meaningless without the notion of jagged teeth, one can define a saw as an object with jagged teeth and ignore its other parts.

Again, he might imagine that because he has found a rough fit between his hypothetical rules and those derived in other ways, he has established that his criterion is the only one relevant, or at least that it is among those relevant. A rough fit does not establish that the criterion is the only one relevant. A concern for efficiency, for example, would indeed shape the rules courts have developed if it were one concern among many, and hence produce such a rough fit. Moreover, the mere existence of a rough fit does not even prove that a criterion is among those relevant. The error lies in a failure to distinguish the criteria or purposes for which rules are made from the side effects that these rules may produce. Rules allowing freedom of speech permit wrongheaded and offensive ideas to be aired; rules that legalize prostitution make life easier for prostitutes; and rules that permit inequalities in wealth help the wealthy. Observing these effects, one can legitimately conclude that the rulemakers were willing to tolerate them and one can argue that they should not have done so. But one cannot conclude that the rules were made in the interests of the wrongheaded, the prostitutes, or the wealthy. To draw that conclusion, one would have to establish not only that the rules favor the interests of these groups but that they deviate from the rules that would have been made by other criteria so as to favor them. One has not established, for example, that the rules of nineteenth century law were made to subsidize industrialists by showing that industrialists flourished under them.

Finally, because hypothetical rules look so much more rational and precise than those derived in other ways, an interpreter might imagine that whoever has derived a set convincingly from some definite criteria should be presumed right until those who challenge his rules can do the same. He might even imagine that those who cannot do so have no views on law or ethics worthy of consideration. But there would be no reason to presume a set of hypothetical rules is correct unless there were a reason to presume that the criteria hypothesized are the only right ones. And to ignore rules that are not derived convincingly from definite criteria would be to assume that there are no other ways of arriving at sound rules.

Classificatory rules, principles, and hypothetical rules have one feature in common: they are attempts to approximate the ideal rules a legal authority could make if he had all the relevant criteria clearly in mind and derived from them rules that would capture the determinative circumstances -- the very circumstances that would call for a particular result by these criteria. Thus, as these three types of rules more closely approximate such ideal rules, they approximate each other. Classificatory rules that did not merely isolate classes of cases in which the determinative circumstances are the same and suggest what these circumstances are but that actually identified them would be identical to principles that captured not one but all the determinative circumstances, and both would be identical to hypothetical rules in which these determinative circumstances were derived knowing all the relevant criteria. Indeed, such rules would then be identical to the repertorial rules one would make if one read cases and listed their significant features already knowing the determinative circumstances and how to describe them. These types of rules, then, are different means of arriving at ideal rules. That being so, they can be used in concert. Classificatory rules, for example, can suggest principles which in turn suggest criteria, which in turn can suggest new principles or classifications. Consequently, the types of rules just discussed will often overlap and blend into each other as they converge on an ideal rule, or as the ideal rule is approximated from more than one direction.

e. Maxims

Instead of trying to approximate an ideal rule in the ways just described, a legal authority might allow his interpreters to apply his criteria directly. As we have seen, he might wish to do so even if he could formulate an ideal rule to avoid encumbering the legal system with too many rules. He has one more reason for doing so if he can only formulate rough approximations to an ideal rule which may not be very accurate anyway. If he allows his criteria to be applied directly, then even though he himself is vague about these criteria, he may still give them some general advice on what to strive for or to avoid. We will call the advice he gives a maxim. A maxim states that across an unspecified range of circumstances, some result is to be favored or to be avoided or not to be reached without further inquiry.

In this sense, "no one departs from a court of equity without a remedy" is a maxim. It does not mean no one ever departs without a remedy and does not describe the circumstances under which a person should receive one. It merely means that giving remedies in equity is something to aim at. In a monarchy, "the king can do no wrong" is a maxim. It does not mean the king never does anything wrong, and except in absolute monarchies, it does not mean his wrongs can never be inquired into. It means that if we inquire too often whether the king has acted rightly or wrongly we will no longer have even a limited monarchy. Since the range of circumstances to which a maxim applies is unspecified, there may be more than one favored result, and hence there may be contradictory maxims. Thus, children are told "haste makes waste" but "strike while the iron is hot"; executives are told to show initiative but not to exceed their authority; and courts of equity are told to deny no one a remedy but to bear in mind that "equity follows the law."

Since these old maxims are indefinite about the circumstances under which they apply, modern lawyers suspect them of being almost meaningless. But since maxim making is well nigh inevitable as long as people cannot make decisions by accurate rules, the result is merely that modern maxims are usually phrased a different way. If there is a favored result, a modern lawyer will say there is an "interest" in reaching it. If there are several favored results and hence a need for contrary maxims, a modern lawyer will say there are several "interests" which need to be "balanced." If one of these results is to be more favored than another, he will talk about "strict scrutiny" of the other. If one of these results should not be reached without further inquiry, he will talk about a "presumption" that needs to be "overcome."

Maxims are useful, not because they tell us much about what to do, but because they counter our temptation when confronted with the facts of a particular case to lose track of what is generally valuable or dangerous. Democracies need maxims about free speech to defuse the contempt inspired by what particular people have said, just as monarchies need maxims such as "the king can do no wrong" to defuse contempt for what particular kings have done. Maxims are dangerous when they are confused with rules that would actually tell us under what circumstances a given result is appropriate. They are also dangerous to the extent they distract us from the search for such rules, for a maxim may have been laid down, not because it would be less desirable to have an accurate rule, but simply because no one was too sure what an accurate rule would be.

f. Pragmatic Rules

Rather than laying down either a maxim or a rule that attempts to identify the determinative circumstances, a legal authority might make a rule that is easy to understand and follow. We have seen that a legal authority might prefer such a rule even to one that yields more accurate results but that is more abstract and hard to understand. If he cannot frame a very accurate rule, he can at least frame one that is simple. If he does so at least in part for that reason, we will say he has made a pragmatic rule.

He might do so in two ways. First, in tackling a new area of law, he may simply borrow a rule from an old area of law in which that rule, he believes, has a relatively clear meaning. He may not be at all certain that the new area is sufficiently analogous to the old to call for a similar treatment, but at any rate he now has a clear rule. The English rule about percolating water that was overturned in Katz v. Walkinshaw n24 may have originated in this way. It treated the owner's right to pump water on his own land like his right to build or plant on his own land, perhaps for no better reason than that the curts already had rules about building and planting. Similarly, the earliest American rules about oil drilling, by analogy to the rules about wild animals, gave an owner the right to whatever oil he captured on his own land by pumping it to the surface. n25 Again, perhaps there was no better reason than that courts already had rules about wild animals.

Second, a legal authority might set up what we might call a rule-generating machine, or paradigm: that is, an abstract method of determining rules that seems to have little relationship to any criterion on which the rules should rest but that can crank out clear rules to govern a variety of situations. Whatever its origins, the common law notion that a lease was an instantaneous transfer of an interest in land played this role for a good many years. That notion can be used to generate the rules that a landlord has no duty to repair, that he need not mitigate damages by reletting if the tenant moves out, and many others. The rule against perpetuities plays a similar role. One can crank out an astonishing number of rules by beginning with the notion of "a life in being plus twenty-one years."

If an interpreter is confronted with the first sort of pragmatic rule, one borrowed from another field of law, he need not follow it any more than any other rule when special circumstances arise that the makers of the rule did not contemplate and that must make a difference by whatever criteria are relevant. Thus, as mentioned earlier, the Katz court did well not to apply the English rule on percolating water in the arid circumstances of California. But more than that, if a rule was laid down in a novel area, in ignorance of how that area ought to be governed and primarily in the interest of having some clear rule, an interpreter has no business assuming that the rule has any merit other than its clarity. He should not assume his authority meant that the rule should be followed absent special circumstances, for all his authority decided and therefore all he could have meant is that the rule should be followed faute de mieux absent special circumstances. If later experience indicates, for example, that the problems of drilling oil are not like those of catching wild animals, and a more plausible yet clear rule suggests itself, an interpreter should adopt it.

These same considerations apply to the second situation in which the authority has set up a sort of rule-generating machine, though here the interpreter faces an additional problem. It can sometimes be very difficult to discard one of the rules the machine has cranked out without calling all of them into question, thereby destroying the clarity and predictability the authority was seeking. The traditional and sometimes exaggerated response of the courts to this danger has been to modify the rules by what are often called legal fictions. The court declares that an appropriate result was generated by such a machine, even though it patently was not, and thereby announces its intention to respect the other results generated by the machine. Examples in land-lord-tenant law include the fictions that a landlord who resumes possession after his tenant abandons the premises accepts an implied offer to reconvey, that if he relets the premises he accepts an implied offer to act as the tenant's agent to relet them, that if he evicts the tenant he is somehow undoing the original conveyance, and that if he allows the premises to run down, he may have constructively evicted his tenant although he begged him to stay. Fictions like these are useful to the extent they preserve a legal paradigm or rule-generating machine while modifying its most egregious outputs. At length, however, when fiction has been so piled upon fiction that the machine has lost much of its value, and experience has suggested a set of clear rules to replace it, an interpreter would do well to scrap the machine since the very reasons his authority created it have ceased to apply.

3. The Possibility of Genuine Interpretation

We have constructed this account of legal reasoning under less than ideal conditions by assuming that an authority starts with criteria he vaguely grasps and a set of results he deems appropriate in particular cases. We saw that even if an interpreter had nothing more to go on than those starting points, he could reason from the result in one case to that in another by methods of analogy and distinction, as long as he had some grasp of his authority's criteria. We have now seen how an authority himself or his interpreter could begin from these starting points and formulate any of six types of imprecise or partial rules. These rules do not capture the determinative circumstances -- those that call for a particular result according to the authority's criteria. Yet each contains some information about the implications of these criteria that a mere recitation of the facts and results of particular cases would not. Thus each type of rule can be interpreted with some degree of accuracy as long as attention is paid to the amount of information it contains. The mistakes in interpretation that have been described arise either from dismissing a rule because it is imprecise and hence neglecting the information it contains, or from assuming that a rule contains more information than it does.

It follows that genuine interpretaon is possible even when cases are hard in the sense that the rules and criteria are unclear. As long as an interpreter, by observing his authority, has acquired some grasp of what his authority's criteria are, he need not interpret according to his personal values or preferences. He can interpret according to the criteria of his authority.

III. LEGAL REASONING AND AUTHORITY

Thus far we have discussed one way of reasoning about law: reasoning from authority. We have seen that an interpreter who wishes to can follow authority by methods that, though not mechanical, do not depend on his own values or preferences. To say that he can do so is not, however, to say that he will or that he should. He may respond, not only to his authority, but to his own conclusions about what the law should be. We will now consider how a person might reason to such conclusions without starting from authority, and discuss two ways in which his reasoning from authority might be related to his reasoning without it: they might be mutually supportive ways of reaching common conclusions; or they might clash, lead to different conclusions, and hence lead to dissent. In either case, we will see, an interpreter's own conclusions should influence how he responds to authority. We will also see, however, that in neither case is it helpful to say that the interpreter is reading his own values or preferences into the law.

Let us consider first how a private citizen could decide whether a given law is as it should be. Imagine that this citizen is not willing to begin his reasoning from any of the starting points that legal authorities have laid down, but that he, like most people, has a variety of opinions of his own about how the law should be in general or how particular cases should be decided. None of these opinions, however, happens to cover the particular question he is asking himself about that particular law. Let us assume further that this citizen, again like most people, distinguishes propositions about what should be from propositions about what he happens to want. He recognizes that he might want the law to be one way when it really should be another. What we are assuming here is a kind of basic moral literacy.

Now a person who fits this description ought to grant that there are criteria, although he cannot state them, that govern how the law should be, independently of whether he believes in these criteria or not. For if there were no such criteria his effort to decide how the law should be in the case he is considering would be meaningless; there would be no criteria that would make one decision more correct than another. Moreover, he already does have opinions about what the law should be in general or in a variety of particular situations. If he believed there were no criteria by virtue of which these opinions are more correct than the contrary opinions, he would have to grant that the contrary opinions are as correct as his own. But that would be nonsense. An opinion is a view about what is correct. One cannot have an opinion and believe the opposite view is equally correct. The person we have described ought to conclude, then, that there are criteria that he may grasp only vaguely but that inform the opinions he already holds and that should inform his decision in the new situation he confronts.

Once this citizen has drawn that conclusion, he will be in much the same starting position as the interpreter in our modified explanation of legal reasoning. The only difference is that the will be starting his reasoning from opinions he holds already rather than those given him by authority. Our modified explanation assumed nothing more than that there are criteria, vaguely grasped by the authority and his interpreter, that inform the authority's general views of law and responses to particular cases. Consequently, the same modes of reasoning that enable an interpreter to progress from his authority's starting points will enable the citizen to progress from his own starting points. To the citizen, they will appear to be ways of reasoning about what the law should be; to the interpreter they will appear as ways of reasoning about what the law is.

That conclusion should not seem too surprising. If methods of interpretation were very different from methods of reasoning from one's own convictions about what is fair and just, legal interpretation would consistently produce unfair results even when legal authorities had laid down fair starting points. One would then wonder why a process which consistently distorted what these authorities were trying to do should be called interpretation, and why a society interested in fairness would allow its judges to interpret.

Reasoning from one's own convictions and reasoning from authoritative starting points can sometimes be parts of a mutually supportive process leading to common conclusions. The two can also clash and divide those in authority from those in dissent. In light of the first possibility, we will have to qualify the account of interpretation given earlier. In that account, the interpreter worked from the authority's starting points by trying to grasp his criteria. We can now see that the authority himself might not want the interpreter to be guided exclusively by authority. The authority might have been trying to lay down fair starting points, which, he hopes, will be interpreted to yield fair results. He might believe, moreover, that the interpreter has a grasp of and sympathy for the criteria of fairness by which he hopes he has been guided, and the interpreter might believe the same of this authority. Under those conditions, the interpreter will not distinguish, and the authority will not expect him to distinguish, whether he is interpreting according to his authority's criteria, his own criteria, or the criteria that really do determine what is fair. For the interpreter and his authority are assuming that these criteria are the same.

In that event, the authority will not want the interpreter to use only the starting points that he has laid down. He will believe that his interpreter can see certain rules and results are fair without having to be told. Nor will the authority want his interpreter to accept all the starting points he has laid down. He knows that some of them could be mistaken and that the interpreter could correct them. Under these conditions, then, interpretation in the narrower sense described earlier will be inextricably intertwined with reasoning from one's own convictions about what is fair.

Conversely, the private citizen who is trying to reach his own conclusions might be willing to accept some of his starting points from authority. He might have sufficient respect for the wisdom of those in legal authority to believe what they say is apt to be right, at least when he personally sees nothing wrong with it. For that matter, he might have this same respect for persons not in legal authority -- people he knows and people who have written books that have impressed him. If so, he might accept what they say as a starting point for his own reasoning, and in this sense, they will be his authorities. In either case, the way he reasons to his own convictions will be interwoven with the interpretation of what others have said.

We have seen, then, that reasoning from authority and reasoning from one's own convictions can be mutually supportive, so much so that they get entangled and one cannot say where one leaves off and the other begins. The other possibility is that the two clash, and there is a conflict between those in authority and those who dissent. We will now consider what a judge or citizen is supposed to do when he finds himself in dissent. According to Lyons, he need not always obey. n26 From what has been said, we can see that this conclusion is correct, although not perhaps in quite the way Lyons claims.

A judge or citizen could be obliged to obey a rule requiring him to obey authority if there are criteria of fairness or justice on which this rule itself is based. If we can identify those criteria, we may be able here, as with any other rule, to identify special circumstances to which the rule does not apply. To do so, we must ask why there should be authorities, and for that matter authoritative interpreters such as judges. Suppose instead that each person reached his own conclusions about the rules he ought to obey, reasoning from his own starting points. What would be unfair or unjust about a system like that?

The account we have given suggests two answers. The first is that individuals might be wrong about their conclusions more often than are recognized authorities and interpreters. We have seen that a person's ability to reach correct conclusions depends on his grasp of the criteria that matter, his starting points, his reasoning ability, and his knowledge of the situation or circumstances that call for a decision. To the extent that some people have deeper moral insight that allows them to grasp the right criteria and find the better starting points, and to the extent that others have more subtle powers of reasoning that allow them to proceed from these starting points, those who care about fairness will recognize them as authorities and interpreters. Their rules will tend to be fairer than those one might arrive at on one's own. Similarly, the process of reasoning from starting points and learning about situations to which rules are to be applied takes time and thus lends itself to a division of labor. Were we equally endowed, it would still behoove us to allow some people to build, some to plant, and some to discover subatomic particles, while others turn moral insights into rules and adapt rules to circumstances with which we are not all familiar, such as those that involve the needs of society as a whole.

Second, fairness sometimes requires that there be a common rule by which we all act, not because that rule is any fairer than others one might think of, but because we are trying to accomplish something together or because we need to know in advance how others will act. That must be the case sometimes, if, as has been stressed here, the criteria by which rules should be made are purposive or teleological. Purposive criteria often require common rules. To pursue purposes together, we need a common plan to tell each what to do; to pursue purposes separately, we need to know what others may or may not do that would interfere. But purposive criteria by themselves often do not dictate which common rule to follow: we could travel to San Francisco together by several good routes, build a house or design a machine in several ways, or win a war by several plans. Therefore someone has to choose. In law, we could prevent accidents by all driving on the right side of the road or all driving on the left, but we need one rule or the other. Therefore, there is a reason for recognizing authorities who make the choice and interpreters whose decisions we all accept.

If these are the reasons for having authorities and official interpreters, and for respecting their decisions even when one would decide differently oneself, then, as Lyons says, there must be special circumstances under which one should not obey. It could be that a judge or private citizen has considered in humility the severe limitations of his own moral insight, his powers of reasoning, and his knowledge of the circumstances, and has also considered the possibility that the duly constituted authorities may be wiser than he, and has then concluded quite correctly that a rule is unjust. It could be that he has then considered how his disobedience could be unjust by defeating some common plan or interfering with those who have expected that the rule will be kept, and he has nevertheless concluded correctly that those injustices will be small compared with the one he will do if he obeys. It could be, finally, that he has considered whether his disobedience will undermine a process that generally does produce fair rules and has correctly concluded either that the process is not as fair as what would emerge in its stead were it undermined, or that the little he will do to undermine it is less unjust than what he will do if he obeys. Assuming all that, it is hard to see how there could be criteria of justice that obligate this person to obey.

One possible answer is that there is some nonteleological criterion of justice on which an absolute duty of obedience can be founded. I have briefly sketched my reasons for thinking that is not so. A more common answer, however, is that the world is full of people who do not care about justice, and who will not consider the limitations on their own wisdom or the need for a common rule or the danger of undermining authority. If these people disobey rules they dislike, the result will be an unjust world indeed. The reason authorities are constituted is not merely to produce wiser rules nor to produce common rules, but to restrain these people. And to be restrained they must obey.

My response is that these people should be restrained if they have disobeyed a just law because they have acted unjustly. Whether they have done so because they do not care about justice or because they have not considered its demands or because they have considered them and arrived at the wrong conclusion, the fact that they act unjustly is a sufficient reason for restraining them. If they have disobeyed an unjust law, however, they should not be restrained, and the law itself never should have been enacted. The reason for restraining people who disobey just laws lies in the injustice these people do, not in the fact that they disobey.

Moreover, it is an oversimplification to say that the reason legal authorities are constituted is to restrain such people. Such people do need to be restrained to prevent them from acting unjustly.But the reasons for committing that task to legal authorities are the two already mentioned: the authorities may be wiser in the way they use their power to coerce; moreover, coercion is usually applied more fairly by a uniform and common rule, both because people know in advance what may happen to them, and because the activities of many people, such as police, sheriffs, judges, and legislators, can be coordinated to that end.

Lyons is right, then, that one should not obey an unjust law. Lyons claims further that a citizen should not presume that a law is just merely because it has been enacted. It is this claim, he says, that distinguishes his position from that of writers such as Professor Dworkin who seem to think that a presumption needs to be overcome. We can now see that Lyons' claim may or may not be right, depending on what he means by a presumption. As we have seen, a citizen must consider the possibility that those in authority may be wiser than he, that his disobedience could thwart some worthwhile common plan or the plans of those who expect the rules to be kept, and that it may undermine a process which generally produces fair rules. If by a presumption that the law should be obeyed, one means that he should consider these possibilities, then there is such a presumption; if one means that he should regard these possibilities as probabilities for no reason or for a poor reason, then there is no such presumption. In any event, it would be better not to talk about presumptions but about what a citizen ought to consider.

We have now seen that if a judge sympathizes with his authority's criteria, he may work in partnership with his authority, introducing his own starting points and occasionally correcting those his authority has laid down. If he does not sympathize with those criteria or if he and his authority have irreconcilable views about how they should be applied, then he will dissent and sometimes he must disobey. In either case, we can now see, it is misleading to say that an interpreter is deciding cases by his personal values or preferences. If he supplements or corrects his authority's starting points by reference to criteria he believes they both share, then he is making an honest effort, at least, to do exactly what his authority would have him do. One can no more say he is acting by his personal values than a junior executive who shares his seniors' desire that their company be profitable and acts, correctly or mistakenly, in response to that criterion.

Suppose, then, we were to say an interpreter acts by his personal values only in cases when he dissents, adopting his own conclusions instead of those he would have reached by reasoning from authority. That would also be misleading. For an interpreter who does not dissent may also decide cases by criteria in which he personally believes. To use the phrase "personal values" only of a dissenter is to obscure what he has in common withthe nondissenter and to suggest that he alone has "values" which are "personal." If all one means is that he is in dissent, it would be less confusing simply to say so. Moreover, to speak about preferences or values suggests a misleading contrast of the kind Newman has drawn between rational thought on the one hand and values on the other. Newman says, for example, that even a judge who decides a case by his own "values" must be "rational" enough to "identify those issues of sufficient legal uncertainty to be vulnerable to resolution on extralegal considerations." He must be able to "reason" from "accepted legal principles" in order to persuade his colleagues. n27 One gets the impression here that the only way to be rational is to reason from authority. But that is not true. As we have seen, one reasons the same way whether one is working from one's own starting points or from those the authority has provided.

In Part I of this Article we saw that even when the law is clear, a judge cannot follow authority mechanically. In Part II, we saw that even when the law is obscure, a judge need not decide a case by his own values or preferences. We have now seen in Part III that even when a judge does not merely follow authority but draws his own conclusions from his own starting points, it is misleading to say that he decides a case by his own values or preferences. We can conclude, then, the dichotomy we have been attacking is false.

Newman saw that neither mechanical decisionmaking nor decisionmaking by preference is an adequate explanation of how judges decide cases. His mistake was to look for a middle ground between them. Once the problem has been framed that way, there is no middle ground. It is like the boy on the burning deck looking first at the fire and then at the sea and then trying to stake out a middle position. The solution is to recognize that the alternatives presented are false. Perhaps, however, Newman's major point is simply that neither alternative will do. After all, he is not a boy on a burning deck but an experienced judge piloting what he knows to be a seaworthy craft. If our tradition of legal philosophy can present him with only two alternatives, either of which makes nonsense of his experience, then he will reject them. It is very hard for a philosopher to persuade a pilot against his better judgment that the ship is going down.

IV. NORMATIVE AND DESCRIPTIVE ACCOUNTS OF LAW

Having seen that the dichotomy between mechanical decisionmaking and decisionmaking by preference is false, we will now consider one reason why people might be attracted to it. They might be trying to give accounts of law that are value-neutral or nonnormative, that do not concern themselves with whether a judge or a legal authority has acted justly or unjustly or otherwise. They might therefore prefer to think that a judge decides cases either mechanically or by his own preferences, since in either case, it seems that one can describe what the judge did without considering whether he acted justly.

The positivist philosophers with whom Lyons has broken do claim that they can give an account of what law is without concerning themselves with whether it is just or unjust. This claim has traditionally led them to see the hallmark of law in the obedience of citizens to authority. Consequently, they have been attracted by mechanical interpretation since then it is most obvious how authority can be followed. The historians Scheiber criticizes want to explain legal change. Since it is hard to see how judges can change the law if they merely interpret authority mechanically, these historians have seen the causes of change in judges' social and economic preferences rather than in their responses to the demands of justice. Indeed, they have not investigated or explained what these demands might be in order to see if judges have been responding to them or not. While they do not explain their premises as clearly as the positivists do, they must believe that a historian can account for what judges have done without discussing what the demands of justice are, for such discussions are noticeably absent from their work.

We will now see that one cannot build either a philosophical or a historical account of law that is agnostic about whether there are criteria of justice to which judges and legal authorities respond or that leaves these criteria unexamined and unexplained. That belief is as false as the dichotomy to which it may have led people.

The Positivists

One way to criticize positivism is to show that not all cases can be decided mechanically from authority. If some cases must be decided by something else, it is a challenge to positivist accounts, which emphasize authority, to explain that something else. If the something else were criteria of justice, then positivist accounts would seem to fail, for they could not explain law and legal interpretation without explaining these criteria.

Thus, Lyons argues that if there are decisions that cannot be justified by watertight argument from established rules, these decisions are either invariably arbitrary or must be justified by something else. n28 Since by hypothesis established rules have given out, the something else must be "principles or policies" that are "fair," since it is hard to see what else could justify a decision. If, however, judges are dutybound to appeal to these principles and policies, then the positivist account is in difficulty. For then we must acknowledge, first, that these decisions can be "genuinely justified" by principles or policies that are "fair," and second, that these principles and policies are part of the law in the sense that judges are dutybound to appeal to them. In that event, it seems hard to leave fairness out of an account of law.

We can now see that this basic criticism of positivism is correct. The reason, however, is not that established rules give out at some point so that there is nothing left to which to appeal except fairness. Lyons makes that assumption for purposes of argument without ever committing himself to it. The assumption, we have seen, is false. As long as a legal authority acts by some criteria that the interpreter can grasp, the interpreter can interpret with some degree of accuracy. Our discussion in Part II made no assumption about whether these criteria were just. Thus the slave of a tyrannical but not wholly arbitrary master could use the methods of reasoning we described to interpret his master's commands. One could use them to interpret the Nuremberg laws. Therefore, the problem with positivism is not that it is impossible to interpret or follow rules without at some point invoking fairness or justice.

Positivists themselves have encouraged their critics to suppose that this is a problem by failing to distinguish clearly between rules that are based on some criteria and rules that are perfectly arbitrary and based on no criteria at all. Perfectly arbitrary rules could only be interpreted mechanically, if at all, but rules based on some criteria need not be. Positivists have tended, then, to embrace one alternative of the false dichotomy we have been attacking.

This alternative may have attracted them because once it is recognized that an authority must be interpreted by the criteria by which he made his rules, one can see a hole in their theory of law that criticisms such as Lyons' expose. For there might be true criteria of justice by which an authority acts and judges interpret him. If so, then these criteria are part of the law at least as much as Lyons' fair principles and policies.

To the extent an authority and his interpreter are guided by such criteria, the criteria will influence the content of the law. Positivists have usually not tried to account for the content of the law but have merely described how rules are made and followed. One wonders why the omission is legitimate. In any event, we can now see that if the authority and his interpreter are guided by criteria of justice, laws will not only have a different content but will be interpreted in a different way. An interpreter will supplement and correct his authority's starting points and interpret them according to criteria of justice that he does not distinguish from those of his authority. A slave interpreting a tyrannical master would be ill advised to adopt any such procedure. Moreover, if an authority, his interpreters, and citizens generally are guided by criteria of justice, those criteria will shape the roles that authorities, interpreters, and citizens play. As we have seen, if there are criteria of justice, there are two reasons why there ought to be authorities and official interpreters: their wisdom or moral insight may qualify them to make better rules, and society sometimes needs a common rule. If an authority and his official interpreters believe they are only charged to expound the demands of justice for those reasons, that belief will influence what they do. For example, they may limit what they say simply because they are not certain what is right and they do not think a common rule necessary. Should an authority act in ways that cannt be justified by either of these reasons, then, as we have seen, official interpreters and citizens generally who are guided by criteria of justice will disobey him. It turns out, then, that if there are criteria of justice to which people are responding, those criteria will affect not only the content of the law but the methods of interpretation and the very notions of authority and obedience. If so, one cannot give an account of law that fails to mention these criteria. They will be a part of the law itself.

A positivist might respond that one could still give an account of law that does not mention justice. The account need only mention the possibility that authorities and interpreters act by criteria, just or unjust, that they believe they share, and that this belief would alter methods of interpretation and their own conceptions of their roles in the way just described. It is true that in constructing our account of how authorities and interpreters would respond to criteria of justice, we assumed little more than that they were both responding to the same criteria. If the criteria are shared, whatever they may be, an authority would expect his interpreter to supplement or modify his starting point in their light. Moreover, those who obeyed authority for the sake of those criteria would see the same two reasons why they should obey: the authority might know better what to do, or he could provide a common rule so people could coordinate their actions. Thus, in war where the criterion to guide actions is victory, or in business where it is profit, soldiers and executives are not expected to take all their starting points from authority but to respond directly to what victory and profit seem to demand. Armies and business firms have authorities for the reasons we have discussed: because they may have a better grasp of military or business strategy, and because the actions of many need to be coordinated by common rules. A soldier or executive who is acting for victory or profit will disobey a plainly aberrational order to dynamite the army's own military base or the firm's own plant.

It is possible, then, to construct an account of how shared criteria will modify methods of interpretation and conceptions of authority and obedience without mentioning whether these criteria are justice, victory, profit, or anything else. That possibility, however, does not vindicate positivist accounts of law but undermines them. For if the same account of authority, interpretation, and obedience can be given whenever criteria are shared, whatever those criteria may be, one cannot give an account of law in terms of authority, interpretation and obedience that would be any different than an account of war, business, or any other activity in which human beings act by shared criteria. The only thing left to distinguish these activities is the fact that the criteria, the ends sought, are different. If we imagined a society in which all rules were made to achieve one purpose only, victory in war, we would be imagining not a society but an army. If we imagined a system of state capitalism in which all rules were made to achieve profit, we would be imagining a gigantic business firm.What distinguishes rules of law from rules made to achieve other purposes can only be the pursuit of a distinct set of purposes. If the pursuit of justice is among them, then one cannot account for law by merely describing the phenomena of authority, interpretation, and obedience. One must account for the peculiar demands that criteria of justice, and any other criteria there may be, make on the way authorities act, on the way they are interpreted, and on the way they are obeyed. An account of law must explain how people behave differently when they are responding to these criteria.

Not only must an account of law mention justice as a goal people pursue; it cannot remain neutral as to whether the goal of justice is possible the way victory is possible in war or profit in business. It cannot remain neutral, that is, as to whether justice is something real and attainable like victory or profit or whether it is an illusion that people happen to believe in the way the Greeks once believed in Apollo. Since victory and profit are real and attainable objectives, the behavior of generals and executives can be influenced by them, and no one would try to give an account of war or economics without describing the demands which these objectives make on those who pursue them. Since communication with Apollo at Delphi was an impossible objective, though one people once believed to be possible, Apollo himself never affected anyone's behavior, although the belief in him did. Consequently, no one now tries to account for the rituals at Delphi by asking how the god actually could be best heard and interpreted, the way an economist asks how profit can actually be made. The rituals at Delphi must be explained by psychology or tradition or in any way except by saying that there really was an Apollo. If law must be described, in part, as an attempt to pursue justice, then an account of law will have to take a position one way or the other. For if justice is something real which can make demands, an account of law must describe these demands. If justice is an illusory goal, it cannot make demands, people cannot respond to them, and an account of the way people behave when they pursue justice through law will be considerably different.

B. Historians

If legal philosphers cannot build accounts of law that are agnostic about whether there are criteria of justice, by the same token legal historians cannot be agnostic about whether such criteria have influenced how law develops. For if there are such criteria and people have responded to them, these criteria are themselves a force in history. One cannot neglect them and still explain how people have behaved. To the extent people have not responded to these criteria, a historian can identify their responses as the product of class bias or self-interest or anything else only if he has an implicit standard of how unbiased, unselfinterested judges would behave and thus a sense of what justice would require. Thus, even a historian writing about the most biased of judges could not be agnostic about what these criteria demand. It follows that a historian's history cannot be value-neutral. The validity of his historical analysis will depend on the validity of a prior and usually implicit legal and philosophical analysis of what doctrine or justice requires.

This point is illustrated by the varying claims of Scheiber, Horwitz, and Hurst. According to Scheiber, judges respond to doctrine itself, and historians such as Hurst and Horwitz have neglected the extent to which they do. His evidence is not limited to instances in which courts stood by established rules despite the pressure of entrepreneurial interests. Scheiber believes that doctrinal innovation can itself be a response to doctrine. One of his leading examples is Katz v. Walkinshaw, in which the California Supreme Court overturned the common law rule governing percolating water on the ground that it would be a "subversion of justice" in the dry climate of California to allow landowners to pump what they pleased. In order to show that the Katz court was not merely being pragmatic but was indeed responding to doctrine, Scheiber shows how neatly the court reconciled doctrinal concerns which haunted its other opinions and sometimes led to very unpragmatic results. The court was able to reach the result it did in part because its reasoning was brilliant, reminding Scheiber of "a dazzling double-feint, backhanded flying layup shot by a basketball immortal." n29 In "taking doctrine seriously," then, Scheiber has not only described what judges did but evaluated it. And indeed he must, for if doctrine is truly a force to which judges can respond, one can only see whether they have done so by seeing what the demands of doctrine are. At that point, one will necessarily be evaluating a judge's opinion as one describes it, for one will be characterizing it as a true response to the demands of doctrine or as a neglect of these demands through ignorance of them or through some other motive. It follows that a historian who wishes to take doctrine seriously cannot be agnostic about whether there are criteria of justice which underlie doctrine and to which a judge is supposed to respond in responding to doctrine. For if there are, then these criteria will influence what judges do.One reason the Katz court rejected the common law rule about percolating water might indeed have been that the rule would work a "subversion of justice" in California.

A historian cannot be agnostic about the demands of doctrine and consequently of justice even if, like Hurst and Horwitz, he believes that judges respond, not to them, but to social and economic preferences of their own. Horwitz claims that nineteenth century contract and tort doctrines were not "neutral." n30 They resulted in a "legal expropriation of wealth," in "forced subsidies to growth coerced from the victims of the process," and consequently enabled an emergent commercial class "to win a disproportionate share of the wealth." n31 Now there is no point in talking about a disproportionate share of the wealth unless one has some idea what a proportionate share would be. To conclude that the rules of contract and tort doctrines were not neutral, one must have some idea of what they would have looked like if they were. The historical analysis is dependent on a set of prior conclusions about what doctrines are fair.

Hurst claims that "what we did in the name of vested rights had less to do with protecting holdings than it had to do with protecting ventures." n32 As evidence, he cites our willingness to abolish primogeniture and entail, to disestablish the few established churches we had, to allow married women to control their property, and to admit new states on an equal basis with the old. n33 As a possible objection to his claim, he notes that in eminent domain cases, nineteenth century courts did "protect property simply as a claim to hold onto what one has." n34 But, he says, those cases rest "on a kind of insistence on equal protection of the laws." n35 This piece of historical analysis depends on a prior legal analysis which Hurst never presents. The legal analysis would presumably explain why primogeniture, entail, the right to have one's church established, the right of a husband over his wife's property, and the right not to have one's political power diluted by the admission of new states were private vested rights every bit as much as those that the courts protected. Only if that is so does Hurst's historical conclusion follow that judges chose what to protect by some other principle than whether a right was vested. The same prior legal analysis would presumably explain why eminent domain cases really involve a problem of equal protection rather than one of vested rights. Indeed, more conclusions about the demands of justice and sound legal doctrine seem to be built into the work of Horwitz and Hurst than into that of Scheiber, the one who urges us to take doctrine seriously. Horwitz and Hurst are claiming that judges consistently deviated from these demands in pursuit of something else, while Scheiber merely claims that judges wrestled with them.

The idea that there could be a value-neutral history grew up in the nineteenth century at the same time that positivist philosophers were trying to concoct a theory of law that made no mention of justice. We can now see that these attempts founder on the same rocks. If there are criteria of justice, these criteria will make their own demands, and just as a legal philosopher will have to consider what those demands are, so a legal historian will have to consider how people have responded to them.

C. Conclusion

We tried to establish in Parts I and II of this Article that there are methods for interpreting an authority that are not mechanical, do not depend on personal preference, and will yield correct results whether the authority's criteria are just, unjust, or otherwise. By itself, that conclusion might suggest that there are value-neutral means of interpreting authority and hence that a judge could decide cases without asking what is just, that a philosopher could account for law without mentioning justice, and that a historian could give a value-neutral account of what judges have done. In Parts III and IV of this Article we have seen that this is not the case. If there are criteria of justice to which an authority is responding, he will expect judges to respond to them also; and legal philosophers and historians will have to allow for the possibility that judges do so.

We have argued that no one, judge, philosopher, or historian, can be agnostic about the existence of those criteria. For if they are real and something that human beings can learn to understand, something to be sought but something that can shape our actions to the extent they are found, then judges, philosophers, and historians cannot talk as though they were not real.

We have offered only one argument as to why these criteria might be real. Anybody who believes that some actions are just or unjust, that some rules however vague are more or less wrong or right, should believe there are criteria of justice that inform his opinions. That may be the only argument that can be made. If someone doubts that the table in front of him exists, I can ask him what distinguishes that table from other objects, the existence of which he does not doubt. If someone asks why a particular action is just, I can ask him what distinguishes that action from others, the justice of which he does not doubt. But if someone claims that no actions are just or unjust, further argument is impossible, just as it would be with someone who claimed that nothing exists. If we are to discuss how legislators, courts, or private citizens ought to behave, we must assume there are criteria of justice.

If we did proceed from the assumption that there are criteria of justice that are as objective as the law of gravity though perhaps even harder to explain, then judges, philosophers, and historians might see their problems rather differently than they have in the last century and a half. Were it not for a tacit assumption in all three disciplines that justice is somehow less an objective consideration than profit is in business or victory in war, it would be odd for us to have a series of Centenary Lectures at Berkeley in which a judge argues that cases need not be decided mechanically or by preference, a philosopher argues that decisions might rest as a matter of law on policies or principles that are fair, and a historian argues that historians should take doctrine seriously. One can only feel grateful to all three lecturers for attempting to lead us out of the dark wood in which we have found ourselves.

FOOTNOTES:

n1 See Newman, Between Legal Realism and Neutral Principles: The Legitimacy of Institutional Values, 72 CALIF. L. REV. 200 (1984).

n2 See Lyons, Justification and Judicial Responsibility, 72 CALIF. L. REV. 178 (1984).

n3 Lyons, supra note 2, at 184.

n4 Scheiber, Public Rights and the Rule of Law in American Legal History, 72 CALIF. L. REV. 217 (1984).

n5 J.W. HURST, LAW AND THE CONDITIONS OF FREEDOM IN THE NINETEENTH-CENTURY UNITED STATES 7 (1956).

n6 M. HORWITZ, THE TRANSFORMATION OF AMERICAN LAW, 1780-1860, at xvi (1977).

n7 Lyons, supra note 2, at 180.

n8 See Fuller, Positivism and Fidelity to Law -- A Reply to Professor Hart, 71 HARV.L. REV. 630, 662-63 (1958) (replying to Hart, Positivism and the Separation of Law and Morals, 71 HARV. L. REV. 593 (1958)).

n9 Our account owes an obvious debt to the Aristotelian idea of "equity," as later refined by Thomas Aquinas, according to which laws are made for an end and therefore must suffer exception in particular cases. See ARISTOTLE, NICOMACHEAN ETHICS V, at 1137a-1138a, in THE BASIC WORKS OF ARISTOTLE 927 (R. McKeon ed. 1941); THOMAS AQUINAS, SUMMA THEOLOGICA II-II, q. 120 (Biblioteca de Autores Cristianos 3d ed. 1962). This is only one debt among many that this account will owe to these authors.

n10 Realizing that one must speak about purposes in order to give an account of liberty and equality, John Rawls speaks of "primary goods" which are those one would want to have as a means to whatever ultimate purposes one wished to pursue. J. RAWLS, A THEORY OF JUSTICE 62, 92-93 (1971). In his "initial position," these goods are allocated in ignorance of what these ultimate purposes might be and hence without making judgments about which purposes are more valuable than other. The trouble with this account is that unless one knew what these ultimate purposes are, one could not judge which "goods" are means to them. Buddhists, Confucians, Mongol war lords, medieval monks, and various eighteenth century English gentlemen and nineteenth century socialists have entertained different views of the ultimate purposes a person should pursue, each of which calls for different sorts of liberty and physical property.

n11 81 Vt. 471, 71 A. 188 (1908).

n12 109 Minn. 456, 124 N.W. 221 (1910).

n13 E.g., H. GROTIUS, DE IURE BELLI AC PACIS LIBRI TRES II, ii, 6 (B.J.A. de Kanter-van Hettinga Tromp ed. 1939).

n14 141 Cal. 116, 74 P. 766 (1903).

n15 Scheiber, supra note 4, at 235-37.

n16 See, e.g., H. KOTZ, DELIKTSRECHT 150 (1976).

n17 See, e.g., B. STARCK, DROIT CIVIL OBLIGATIONS § 1592 (1972).

n18 Jersey Farm Co. v. Atlanta Realty Co., 164 Cal. 412, 415-16, 129 P. 593, 594 (1912).

n19 W. PROSSER, LAW OF TORTS § 117 (4th ed. 1971).

n20 RESTATEMENT (FIRST) OF RESTITUTION § 1 (1937).

n21 RESTATEMENT (SECOND) OF CONTRACTS § 90 (1981).

n22 G. GILMORE, THE DEATH OF CONTRACT 64-65 (1974).

n23 See Gordley, Equality in Exchange, 69 CALIF. L. REV. 1587, 1594-95 (1981).

n24 141 Cal. 116, 74 P. 766 (1903).

n25 See, e.g., Hammonds v. Central Ky. Natural Gas Co., 255 Ky. 685, 75 S.W.2d 204 (1934).

n26 Lyons, supra note 2, at 194-98.

n27 Newman, supra note 1, at 205.

n28 Lyons, supra note 2, at 190-93.

n29 Scheiber, supra note 4, at 236-37.

n30 M. HORWITZ, supra note 6, at xii-xiii.

n31 Id. at xvi.

n32 J.W. HURST, supra note 5, at 24.

n33 Id.

n34 Id. at 25.

n35 Id.