Reason and Precedent in the Law
1. The Problem
I propose to resolve a paradox of legal reasoning. I think it is appropriate to discuss it here as it bears on some topics in which Tibor Machan has had a life-long interest: the nature and value of reason, and its role in a society of ordered liberty. The paradox can be formulated as consisting of the following initially plausible claims: 1.) Judges, in reasoning as they do, are not irrational. 2) But judges, in reasoning as they do, follow precedent. 3.) To follow precedent is to accept an idea presented by a judge in an earlier case in place of whatever idea one might have on the same issue oneself. 4.) To accept another person’s idea in place of one’s own is irrational. 5.) By (2), (3), and (4), judges are irrational in reasoning as they do. 6.) But (5) contradicts (1). Clearly, (1)-(4) constitute an inconsistent belief set, one of which must be abandoned. An extreme “legal realist” would reject (2), on the grounds that judges really apply their own ideologies, class interests, and so forth, instead of following precedent. I will argue for abandoning (3), on the ground that the dichotomous distinction which underlies (3), between following precedent and using one’s own judgment, is wrong.
2. Narratives of Accidents, Injuries, and Death
The point I wish to make is probably best made by taking a fairly close look at some actual legal cases. The cases upon which I will focus come at the end of a long series of cases. To understand the cases that conclude the series, one must understand something of series that gave rise to them. This particular series has been discussed many times, and typically as a classic specimen of legal reasoning, which of course is precisely the light in which I mean to examine it. I trust that, if I say something about it that is somewhat different from what the others have said, I will be pardoned for telling this tale one more time. The story is, at any rate, interesting in itself.
The culminating case in the series is the literally epoch-making one in the New York Court of Appeals (the state’s highest court) of MacPherson v. Buick Motor Company (1916). By all accounts, the story leading up to MacPherson begins with an English case, Winterbottom v. Wright, which was decided in 1842. Briefly, the facts that gave rise to the suit in Winterbottom were these. A certain carriage driver was hired by one Nathaniel Atkins, who contracted with the local Postmaster General to supply drivers to carry the mail. As this driver was taking the mail from Hartford to Holyhead, the mail coach broke down beneath him; he was thrown from the driver’s seat and suffered injuries that permanently disabled him. He sued the contractor who had supplied the coach to the Postmaster General, arguing that the coach had collapsed due to defects which the driver himself could not have detected, and which were due to negligence on the part of the contractor. The contractor, he argued, had promised to provide coaches that were safe and fit to carry the mail. The court ruled that the contractor was not liable to the driver for damages, even supposing he really had been negligent. Several reasons for this ruling are given by the judges involved, but the one that is most strongly emphasized in subsequent case law and commentary is suggested by one Baron Rolff, who points out that the contractor’s duty to provide a safe carriage would have to “have arisen solely from the contract” by which he agreed to provide the carriage. If someone has a right to ride in a safe carriage, it could only be because the contractor agreed in the contract to provide safe carriages. The problem for the driver’s case is that the contract is one that the contractor had with the Postmaster General, and not with the driver. A contract can only create rights for people who are parties to the contract: this idea, the so-called “privity of contract,” erects a wall between the contractor and the driver which the driver’s considerable claims on our sympathy cannot breach.
Though it was an English case, Winterbottom was soon followed as a precedent in the United States as well as Great Britain. This might seem odd to us today, as it is perhaps more obvious to us now than it was then that it is a potential source of problems. As case law develops, at any rate, potential problems tend to become actual ones, and when they do they must be dealt with in one way or another. That, of course, is exactly what happened as American courts attempted to follow Winterbottom.
In Thomas v. Winchester, the New York Court of Appeals faced a difficult problem in the aftermath of a nearly fatal accident that occurred the year after the Winterbottom decision was written. In 1843, Mary Ann Thomas, a resident of Cazenovia New York, was ill, and her doctor prescribed an herbal medicine, extract of dandelion. Her husband purchased from their druggist a jar labeled “½ lb. dandelion, prepared by A. Gilbert, No. 108, John Street, N. Y.,” and she took the required dose. Her condition dramatically worsened at once and, though she eventually recovered fully, for a while it seemed to be a very real possibility that she might die. It turned out that the bottle contained, not harmless extract of dandelion, but extract of belladonna, a deadly poison. Seeking compensation, she sued a druggist named Winchester, who was the employer of the A. Gilbert mentioned on the label. It was in Winchester’s establishment that the bottle had been mislabeled. However, Mrs. Thomas had not bought the mislabeled bottle from Winchester. He had sold it to another New York City druggist, who in turn had sold it to the Cazenovia druggist, who had sold it to Mr. Thomas. Mrs. Thomas and her husband were as it were twice removed from any contractual relationship with the people whose negligence had injured her. The authors of her brush with death based their case on Winterbottom. In effect, the argument was that Mrs. Thomas could not sue them because they had never promised not to poison her.
Not surprisingly, the court was reluctant to accept this argument. As Chief Justice Ruggles points out in the Thomas decision, it would imply that the only party in the world that has protection against the negligence of Winchester and his employees is the first party to whom they sold the mislabeled poison: which in this case was someone who bought it in order to sell it again, and so was not about to swallow it and be hurt. Everyone who might actually be injured or killed by it was left unprotected. However, such implications could only be avoided by either rejecting theWinterbottom rule or by recognizing an exception to it. How could this be done without violating the idea of privity of contract? In a decision that provided the foundation for a number of subsequent decisions, the court solved this problem by placing the issue outside the realm of contract law altogether, a move which rendered the privity rule irrelevant to Winchester’s obligations in this case. The source of the defendant’s duty of care, and the reason his conduct violated this duty, was not contract at all. It lay in the fact that his
negligence put human life in imminent danger. ... The defendant’s duty arose out of the nature of his business and the danger to others incident to its mismanagement. Nothing but mischief like that which actually happened could have been expected from sending the poison falsely labeled into the market; and the defendant is justly responsible for the probable consequences of the act.
The source of the duty here is the same as that of the duty that would be violated by putting a loaded gun in the hands of a child, or leaving a horse and cart unattended in the street (which were two of Chief Justice Ruggles’ examples): it is based on the fact that the object involved is “imminently dangerous” and not on any relationship between the injured person and the one who causes the injury. It rests on the fact that “death or great bodily harm of some person was the natural and almost inevitable consequence of the sale of belladonna by means of the false label.” The principle involved is not contractual but a principle of tort law: a duty of care that one owes to people in general, and not merely those with whom one has made special arrangements.
Thomas established that privity of contract, as enunciated in Winterbottom, is no barrier to liability if the object that causes the injury is “imminently dangerous.” Obviously, it is now important to know which things are imminently dangerous and which are not. Ruggles says, as we have seen, that a mis-labeled bottle of poison, an unattended horse and a loaded gun left in the care of a child are all imminently dangerous, and he also says (in a nod of agreement with Winterbottom) that a carriage is not. In the following decades, there was a series of New York Court of Appeals cases that raised the question of which side of this line various other objects belong. In the 1870 case of Loop v. Litchfield, for instance, the court ruled that a cracked flywheel, which flew apart and killed someone after five years of use, was not imminently dangerous, partly on the grounds that the five years of non-lethal use show that the injuries it caused were not “the natural result or the expected consequence” of manufacturing the flawed wheel. Many of the objects that cause injury and death, however, are not so easy to place on one side of the line or the other. One such object gave rise to a case that later proved to be a turning point.
Mary Devlin’s husband was a painter whose employer had contracted to paint the inside of a courthouse dome. This required the construction of a tall scaffold to support the workers. The painting contractor hired an experienced scaffold-builder to do this important job. For some reason, the builder chose to fasten a vertical support beam to the top horizontal member of the scaffold by nailing it, instead of lashing it with ropes as was customary. The springing motion of the horizontal members sheared through the nails, causing the scaffold to collapse while Mr. Devlin was working on it. He fell ninety feet to his death. His widow sued the scaffold builder. Her case would have faced no serious problems but for the fact that (as the attentive reader may have noticed) the scaffold-builder’s contract – which promised a “first-class” scaffold – was with Mr. Devlin’s employer, and not with Mr. Devlin. Should the court treat the faulty scaffolding as it had treated the perforated flywheel, or should it treat it as it had treated the mislabeled poison? In Devlin v. Smith, the court held that such a scaffold, like the falsely labeled bottle of poison in Thomas, is indeed imminently dangerous:
Stevenson [the scaffold-builder] undertook to build a scaffold ninety feet in height, for the express purpose of enabling the workmen of Smith to stand upon it to paint the interior of the dome. Any defect or negligence in its construction, which should cause it to give way, would naturally result in these men being precipitated from that great height. A stronger case, where misfortune to third persons not parties to the contract would be a natural and necessary consequence of the builder's negligence, can hardly be supposed, nor is it easy to imagine a more apt illustration of a case where such negligence would be an act imminently dangerous to human life. These circumstances seem to us to bring the case fairly within the principle of Thomas v. Winchester.
Though the key phrases here -- especially “natural and necessary consequence” and “imminently dangerous” – are borrowed from or very closely modeled on the language of Thomas, it is also true that to apply them to a faulty scaffold might be seen as extending their meaning somewhat beyond the original paradigm of a bottle of belladonna masquerading as medicine.
Torgeson v. Schultz, a decision that came a quarter-century after Devlin and also relied on Thomas, seemed to confirm that a change had taken place. In this case, a domestic servant had lost an eye when a bottle of carbonated water exploded. The bottle was warm, it being a hot July evening in New York City, and when she placed the bottle in a bucket of ice the stress on the glass due to the change in temperature was enough to cause the bottle (with 125 pounds of internal pressure per square inch) to burst and injure her. She sued Schultz, the bottler. However, she had no contractual relationship with him, the bottle having been bought from a local druggist. Writing for a unanimous court, Judge Willard Bartlett ruled that, nonetheless, Schultz was not shielded from liability. Most of Bartlett’s opinion is given over to a discussion of the evidence that tends to show that Schultz was indeed negligent. Concerning the issue of whether this really is a negligence case in the first place – concerning, that is, the problem of whether the privity rule constitutes a barrier between the plaintiff and the defendant in this case, the discussion is more or less limited to quoting with approval a British case, Heaven v. Pender (1883). That quotation reads:
Any one who leaves a dangerous instrument, as a gun, in such a way as to cause danger, or who without due warning supplies to others for use an instrument or thing which to his knowledge, from its construction or otherwise is in such a condition as to cause danger, not necessarily incident to the use of such an instrument or thing, is liable for injury caused to others by reason of his negligent act.
The language here seems on the face of it markedly different from Thomas and Devlin. Talk of the injury being natural, necessary, or inevitable, or of the danger as being imminent, is gone. The rule here seems to be that, if the object is dangerous in the first place, and if the defendant through negligence allows it to be “in such a condition as to cause” an injury it would not ordinarily cause, that is enough to create liability. Indeed, the facts in Torgeson do seem to require that the more exacting language be omitted if the principle involved is to allow the plaintiff in this case to have a chance of winning. An expert witness at the trial, a Columbia University physicist, had reproduced the conditions in which the plaintiff lost her eye. Of seventy five of the plaintiff’s seltzer bottles thus tested, five exploded. Obviously, this is far too many exploding bottles, but it would be hard to say that this shows detonations were “necessary” or “inevitable.” During the next year, the same court dealt with another exploding container of fluid – a large coffee urn in a hotel restaurant, which badly scalded two people and killed a third – in Statler v. Ray Manufacturing Co. Torgeson is cited and the same quotation from Heaven is repeated, with no discussion of how likely the explosion was, given the condition of the container. Some would say that the stage was now set for the case of MacPherson v. Buick Motor Co. To many observers at that time (1916), however, MacPherson must have seemed an earthquake that no one could have predicted.
The facts of MacPherson are very simple. One day, as Donald MacPherson was driving his new Buick automobile, taking a friend to the hospital, one of its wheels suddenly collapsed – the wooden spokes having crumbled to pieces under him – and MacPherson was thrown from the car and seriously injured. Evidence presented at the trial showed that he was driving the car prudently at the time (traveling only eight miles per hour) and that the wheel collapsed because it was made of defective wood. He sued the manufacturer of the car. However, he had no contract with the manufacturer, having bought it from a retail dealer in Schenectady, who in turn had bought it from the manufacturer in Detroit. Hence the problem. Mr. MacPherson could only have a case if these facts fall within some exception to the privity of contract defense. Clearly, things did not look good for him. A short while before, a federal court, in a case with virtually identical facts, had declared that those facts did not constitute an exception, because an automobile is not an inherently dangerous article. Indeed, the facts in MacPherson read curiously like a twentieth century version of the Winterbottom case, with an automobile playing the role of the horse drawn mail coach: in all relevant respects, it resembled the case that had long been the paradigm of a manufacturer seemingly negligent but shielded from liability by the privity rule.
Nonetheless, Judge Benjamin Cardozo, writing the opinion of the majority of the court, declared that the manufacturer was liable to MacPherson for damages. His reasoning was based on a magisterial overview of the whole series of cases I have described here. One main point of Cardozo’s overview is that, over the course of this series of cases, the law has changed. This of course is obvious enough. (It is also rather obvious that the law jolly well ought have changed.) But he presents a very distinctive account of the nature of this change. The turning point, as he sees it, is Devlin, the scaffolding case:
It may be that Devlin v. Smith and Statler v. Ray Mfg. Co. have extended the rule of Thomas v. Winchester. If so, this court is committed to the extension. The defendant argues that things imminently dangerous to life are poisons, explosives, deadly weapons--things whose normal function it is to injure or destroy. Whatever the rule in Thomas v. Winchester may once have been, it has no longer that restricted meaning.
As he sees it, the early cases regarding possible exceptions to the privity of contract as a barrier to liability (specifically, the Thomas and Loop cases) “suggest a narrow construction of the rule,” while the later cases (beginning with Devlin) “evince a more liberal spirit.” The focus of Cardozo’s argument is on the facts of the cases and the outcomes (ie., whether the defendant won or the plaintiff), rather than on the arguments presented by the judges in defending their decisions. The basic idea is here is straightforward:
A scaffold (Devlin v. Smith, supra) is not inherently a destructive instrument. It becomes destructive only if imperfectly constructed. A large coffee urn (Statler v. Ray Mfg. Co., supra) may have within itself, if negligently made, the potency of danger, yet no one thinks of it as an implement whose normal function is destruction. What is true of the coffee urn is equally true of bottles of aerated water (Torgeson v. Schultz, 19 2 N Y. 156).
Given the facts of these cases, the principle cannot be what it was in Thomas, that the exceptions are occasioned by objects that are “inherently dangerous,” where the paradigms of inherent danger are bottles of poison and loaded guns. The time has come, he tells us, to state the principle behind the new cases. According to Cardozo, it must be this: “If the nature of a thing is such that it is reasonably certain to place life and limb in peril when negligently made, it is then a thing of danger.” That is, an object is dangerous in the relevant sense if it is something that would pose a risk if negligently made. This would mean that privity of contract is no barrier to liability in precisely those cases where Winterbottom said that it was a barrier: namely, where the manufacturer has been negligent. If this decision becomes influential, then the Winterbottom has been entirely overthrown. In fact, Cardozo’s reasoning in MacPherson was almost at once very influential, quickly becoming part of the foundations of products liability law as we have it today. With MacPherson, the era of Winterbottom came to an end.
There is, however, one important complication in the MacPherson case, one that, aside from its intrinsic dramatic interest, is important for what it can teach us about the nature of legal reasoning. There is a spirited dissent, written by none other than Chief Justice Willard Bartlett, author of the Torgeson decision, which Cardozo cites in support of his position. Defending his lone dissenting vote, Bartlett declares, as directly as courtesy will allow, that his colleagues have flagrantly misread Torgeson and the whole series of decisions that Cardozo claims to be interpreting. Speaking of the lower court rulings that the majority has upheld, he states categorically that
these rulings ... extend the liability of the vendor of a manufactured article further than any case which has yet received the sanction of this court. It has heretofore been held in this state that the liability of the vendor of a manufactured article for negligence arising out of the existence of defects therein does not extend to strangers injured in consequence of such defects but is confined to the immediate vendee. The exceptions to this general rule which have thus far been recognized in New York are cases in which the article sold was of such a character that danger to life or limb was involved in the ordinary use thereof; in other words where the article sold was inherently dangerous.
The principle that Cardozo claims is forced upon him by the Devlin decision is one that Bartlett believes is simply not to be found in the text of Devlin at all:
It is said that the scaffold if properly constructed was not inherently dangerous; and hence that this decision affirms the existence of liability in the case of an article not dangerous in itself but made so only in consequence of negligent construction. Whatever logical force there may be in this view it seems to me clear from the language of Judge Rapallo, who wrote the opinion of the court, that the scaffold was deemed to be an inherently dangerous structure; and that the case was decided as it was because the court entertained that view. Otherwise he would hardly have said, as he did, that the circumstances seemed to bring the case fairly within the principle of Thomas v Winchester.
Bartlett’s comments suggest that what we have here is not merely two different views on a question of the law (one that is by now quaintly obsolete) but, much more importantly, two different and opposed views of the process by which judges reason from precedent. His own approach, he seems to suggest, is one that might well collide with “the logical force” (to use his phrase) of one’s own private views. Cardozo’s view, he seems to insinuate, is not so much following the law as it is, as supplanting it with one’s own view of what is right. Actually, as I would like to show, the two approaches are not as different as they look. The conflict between Bartlett and Cardozo can be understood as a clash between two ways of employing the same method. That method is the method of the law itself, insofar as it arises from the practice of following precedent. Before I can attempt to show this, however, I will need to step back and make some more general comments about the nature of legal reasoning.
3. Following Precedent
Both of the judges in MacPherson claim that they are following precedent. What is it to follow precedent? And what is the precedent, which the judge is following? The answer to the latter question, at any rate, seems obvious. The precedent is an earlier case, which is embodied in a certain legal document which is part of the public record. This document can be seen as a written narrative, recounting a meaningful series of concrete events. The earliest of these events, typically described very tersely and in abstract language, is the fact-pattern that eventually gave rise to the legal proceedings: someone was injured in certain circumstances and by certain means, or perhaps someone breached an agreement of a certain sort and in a determinate way, and then someone sued, seeking a legal remedy. The judge in that case made a decision regarding the merit of the suit. The narrative does not stop there. The precedent case is always the result of an appeal from the original judge’s decision: one of the parties involved in the original decision, either the plaintiff or the defendant, holds that the first judge’s decision was wrong and seeks from a higher court some correction of the error. Accordingly, the continuing narrative recounts to some extent the reasons why the original plaintiff and defendant disagree about whether the lower court decision was an error or not. Finally, it ends with an outcome handed down in the precedent case itself, in which the court generally either grants or withholds all or part of what the appellant was seeking: a happy ending or an unhappy one, depending on which party has one’s sympathies.
This, then, is what appellate court judges are contemplating when they review a potentially precedent case: a narrative that recounts a series of particular events. By the same token, the judge’s present, yet to be decided case is another such narrative. What, then could it mean to “follow” the earlier case in deciding the present one? Actually, we have already encountered a sort of reasoning that proceeds from one particular case to another, and that of course is reasoning by analogy. Indeed, it is rather obvious that in some sense what these judges are disagreeing about is precisely whether the present case is analogous to certain earlier cases or not. In general, the reasoning judges use in justifying a decision seems to be a sort of analogical reasoning, as a number of writers have pointed out. What is particularly interesting here is the way in which Cardozo and Bartlett argue about whether the cases in dispute are analogous or not. They do so by clearly and pointedly discussing whether there is some principle that connects the cases.
This, at any rate, is rather obviously true of Cardozo. He is trying to show that there is a principle that requires the result he has arrived at, and that this principle is supported by the cases that he claims are analogous to MacPherson. Further, one can interpret him as attempting to do this by claiming that this principle explains the analogous cases. His fundamental assumption, one could say, is that these cases were, in the main, rightly decided. But why were they right? What explains their rightness? This cannot be explained, he claims, on the basis of the principle that exceptions to the privity rule are founded upon objects whose normal function is to destroy, since this explanation only covers some of the cases. A scaffolding, a bottle of aerated water, and a coffee urn are not per se implements of destruction, as poisons are. His own principle, that exceptions are founded upon objects that are dangerous in the sense that they would be harmful if negligently made, is a better explanation because it does explain all the cases.
On the other hand, Bartlett seems on the face of it to be arguing in a completely different way. He points out that Judge Rapallo’s language in the Devlin decision includes an argument to the effect that his decision falls “within the principle of Thomas,” a principle which is obviously inconsistent with the one that Cardozo is claiming to find in Devlin. One facially plausible way to interpret what Bartlett is saying here would be to suppose he means that the way to interpret a case is to scrutinize the language in the decision to see exactly what the judge meant. One looks for linguistic clues as to the author’s intent. If the author of a decision makes an explicit statement of intent, that is extremely strong evidence. If the author says something that would, logically, grossly conflict with a given interpretation, that would also be powerful evidence as well. Of course, evidence like this is far removed from the sort that is based on explaining why the decision is right. If this is all that the reasoning that a judge uses in following precedent amounts to, then it will obviously have nothing to do (so long as it is adhered to scrupulously) with what the judge thinks is right. Such a method would require us to follow the principle that is actually in the case instead of consulting our notions of right and wrong. It also seems to generate the paradox that I formulated at the outset of this essay.
The idea that this is exactly what Bartlett believes he is doing has the additional merit that it helps to explain what seems, at any rate to me, the greater than usual heat of Bartlett’s remarks. After all, Bartlett was the author of Torgeson, a case that Cardozo had the cheek to declare agreed with his own position. It would have been an infuriating position for Bartlett to be in, particularly if he believed that all judges are supposed to do is to find the intentions of judges in precedent cases. The intentions behind Torgeson were to a considerable extent his intentions, and he surely is an expert on that. Despite the beauties of this interpretation, not the least of which is that it enhances the irony of the drama behind MacPherson, I do not think this is accounts for what Bartlett is actually doing in this case.
In the first place, it would not be consistent with what we have already seen him doing in Torgeson. In that decision, Bartlett does not interpret the series of New York cases leading up to Torgeson simply by finding the principles explicitly stated in them, nor does he linger for long over the wording of these cases in search of clues to authorial intent. Rather, he displays “the principle of law invoked” by Thomas “and similar cases,” not by quoting Thomas at all, nor indeed any of the cases which under the principle of stare decisis he was bound to follow: instead, he quotes the formulation of Heaven v. Pender – which, as I have already mentioned, is a British case. Whatever else one might say about his way of interpreting Thomas, and the related American cases, it plainly uses exegetical resources in interpreting a decision that go beyond scrutinizing the language actually used in that decision as evidence of authorial intent.
Further, if we take a second and somewhat closer look at his dissent in MacPherson, it appears that there is actually more to his method there than initially meets the eye. Notice that the point of his comment about Judge Rapallo’s actual language in Devlin is entirely negative, in the logical sense of that word, as is the point of his entire contribution to MacPherson: that is, he is concerned simply to deny the truth of a positive assertion. His contribution is, after all, a dissent. He is denying Cardozo’s interpretation of Devlin. The reason he gives is that Cardozo’s interpretation clashes with what the Judge Rapallo actually said in Devlin. In effect, he is invoking a constraint on judicial interpretation: Whatever else you do, do not put forth an interpretation that contradicts the judge’s explicit statements. Cardozo claims that the principle in Rapallo’s ruling constituted a rejection of the Thomas version of the relevant exception to the privity rule, while Rapallo (Bartlett ponts out) claimed it did not. Bartlett seems to think that, at least if a particular judge is not incompetent, this is strong evidence regarding what it is that he or she is actually ruling, and Cardozo’s reading is simply wrong because it conflicts with the clear implications of this evidence.
But this, as I say, only tells us that Bartlett believes that a certain constraint applies to judges who are interpreting cases. How does he determine what the principle of a case actually is? That is, what does he do within the limits of this constraint? No doubt he would maintain that we have already seen him doing something of that sort in Torgeson v. Schultz. In that case, he is faced with an array of cases that both affirmed that privity of contract is a shield of sorts against liability for manufactured articles and found that there are potentially exceptions to the principle based on the nature of the article that does the injury. The item in that case was a bottle of pressurized water. Obviously, such an article is less hazardous than a bottle of belladonna, the dangerous item in Thomas, the case that first established this category of exceptions. In addition, the wording of Thomas stresses the degree of danger of the bottle of poison – “death or great bodily harm” it says, are “the natural and almost inevitable consequence” of the bottle as negligently mislabeled. The bottle of pressurized water is less dangerous than that. It is also less dangerous than the faultily constructed scaffold in Devlin. However, this does not mean that the bottle cannot also be the basis of an exception to the privity rule. After all, these cases do not say that less dangerous objects do not create exceptions to the rule. He could justify including the bottle in the same category as the scaffolding and the poison if you can find a principle that explains why these earlier cases are right which at the same time is broad enough to have similar implications for the present one. Bartlett believes he finds such an explanation in the principle in the Heaven case. It maintains that if a person leaves “a dangerous instrument, as a gun” in a state or condition that will “cause danger” to others is liable for the results. The idea seems to be something like this. If someone makes some item and it then leaves their control and goes out into the world to be used and controlled by strangers, we do not ordinarily hold them responsible for injuries it might cause. However, if the thing is in itself dangerous, that indicates that they have a duty to a higher level of care that usual. In that case, if they are negligent in the way that they make it, they can be held liable if as a result of this negligence the object causes injuries it would normally not cause. That is, Bartlett can be seen as using the same method, in Torgeson, that I have found in Cardozo’s decision in MacPherson. Bartlett is building an analogy between Torgeson and the cases he claims as precedents by assuming their results were determined rightly by their authors, finding a principle that would explain why they were right, and showing that it supports his decision in the present case.
Part of the difference between Bartlett and Cardozo lies in the fact that, as we have seen, Cardozo says in so many words that a scaffolding, a coffee urn full of hot coffee, and a bottle of aerated water are not dangerous in themselves, and of course Bartlett disagrees. To some extent there is a disagreement between them about these objects, or about the correct usage of a phrase in English. But this same disagreement is also a matter of legal method as well. Bartlett holds that he is bound by the fact that these earlier decisions explicitly state that they are dangerous in themselves (even apart from negligence), while other statements make little sense unless that is what their authors think. In other words, he sees his method as constrained.
By contrast, Cardozo’s method seems comparatively freewheeling and unconstrained. As such, it bears some resemblance to Edward H. Levi’s classic quasi-realist depiction of the way in which judges reason as case-law develops. In Levi’s account, as in my account of Cardozo, the judge finds analogies between the case to be decided and potentially precedent cases. But Levi also specifies – a point that I have so far struggled to leave open – where the judge gets the principle that connects the present case and the precedent ones: in his view the judge freely chooses it. Levi variously describes the ratio decidendi that is explicitly stated by the judge in the previous case as “mere dictum” and “window dressing.” At one point, he states flatly that the present judge is “not bound [apparently meaning, not at all] by the statement of the rule of law made by the prior judge even in the controlling case.” Cardozo’s practice resembles this at least to the extent that he is happy to disregard some of the things that the earlier judges say in justifying their decisions, in order to reach what he regards as a just result.
4. Individual Judgment and Law
I should emphasize that the difference I see between Bartlett and Cardozo is only a matter of degree. The difference is not that Bartlett’s method is constrained while that of Cardozo is unconstrained. First, both Cardozo and Bartlett regard themselves as bound by the outcomes in the previous cases, insofar as they are consistent, and by the general trend of the outcomes if they are not. This much is of course consistent even with Levi’s model. In addition, even Cardozo does (contrary to Levi’s account) show a certain tendency to follow the stated rules of law in earlier cases. He at least shows them enough respect to quote what they say and show how close his rule is to what they have been saying. The issue that divides his practice from that of Bartlett is not whether there are any constraints but, rather, what those constraints are and how they are weighted, both in relation to each other and in relation to the judge’s own judgment.
On the other hand, despite appearances to the contrary, it is also true that both of them use their judgment about right and wrong. I would say that the difference between them in this respect is that, in Bartlett’s approach, the individual insight of the judge serves mainly to move into gaps in the law and fill them up, while that of Cardozo shows a greater readiness to overturn what already stands. Both, however, do use their own judgment about what is right and wrong. Further, a little reflection will show why a legal system based on case law must have both of these traits: decision-making-procedures used by judges must be constrained and they must rely on the individual judgments of the judges.
It is rather obvious that the first of these traits, constrained decision-making procedures, is a necessary attribute of any legal system at all. This follows from the mere fact that, whatever else might be true of the things that the judges say and do, they must result in a system of norms that can then govern the conduct of those to whom these norms officially apply. The decisions of a collection of judges who decide cases simply and solely on the basis of their own individual judgment would not constitute such a system. Since different individuals judge matters in different ways, such a “system” (if it could be called that) would leave its subjects in the position of repeatedly having to wait and see what the relevant judge will tell them to do. It would be a system of ad hoc judgments and not of law at all. Of course, this point has been made already by any number of people.
The necessity of the second of the these traits – that, within these constraints, judges will have to rely on their own individual judgments of right and wrong – is perhaps less immediately obvious on the face of it, but it is clearly suggested by my narrative of the series of cases that culminated in MacPherson. The decisions made by the judges in these cases have several features that seem immediately relevant. First, the “present narrative,” the one that the judge is completing by finding analogies in previous cases, is a recounting of recent events that await a resolution, which is a course of action that is yet to take place. This course of action will be taken because of pronouncements on the part of the judge, to the effect that they must be taken. Further, these pronouncements are always made in the first person. Generally, the decision is not laid down by asserting something to the effect that “Whatever you or I might think of the matter, the law says so-and-so.” By and large, the judges issue their decisions by asserting that so-and-so is true.
At this point, someone might wish to maintain that judges in such cases have simply substituted the point of view of the law for their own point of view, so that the I in these first person pronouncements is in effect the law itself. Such an interpretation of the position of the judge might be arguable (whether it ultimately holds up or not is another matter) in a legislated legal system of the sort we find in the modern continental European tradition, in which the legislature seeks to anticipate all problems of interpretation in advance and resolve them in the legal code which it enacts. What we have seen here is that it does not apply so readily to the position of the judge in the case-law tradition of England, America, and the other common law countries: the tradition in which the law to a significant extent emerges from the decisions made by judges, and from the way that individual decisions accumulate and interact.
One fully sufficient reason why it does not apply so readily is just this: a legal system that works this way will inevitably require that judges make substantive additions to it by means of an imaginative use of their own capacity to distinguish right from wrong. When Baron Rolff and his colleagues decided Winterbottom, they had before them the case of a defective carriage wheel. In enunciating the privity rule, they were not thinking of the consequences of a mislabeled bottle of poison. However, Chief Justice Ruggles in Thomas is contemplating just such a case. It would have been quite arbitrary of him to say something like “Winterbottom mentions no exceptions to the privity rule, therefor there are none.” From the fact that none are mentioned, it does not follow that none exist. On the other hand, it does not follow that any do exist, nor yet of course does the original decision tell us what these exceptions might be, if there are any. On no rational reading does the earlier case uniquely determine a single response to the present case. Further, in such a system it is inevitable that decisions that are under-determined in this way will arise unless the judges in the earlier cases are either omniscient or simply arbitrary, declaring in advance that there either will be no exceptions or that they know what these exceptions will be. Just as the mislabeled bottle of poison took the rule by surprise (so to speak), so did the cracked flywheel, the faulty scaffold, the exploding seltzer bottle, and the exploding coffee urn. Whatever the judge in the present case does, it will necessarily involve some creative use of the judge’s capacity to distinguish between right and wrong.
In addition, there are characteristics of a case-law system that will tend to increase the extent to which existing rules under-determine (some) future decisions. Everyone who has read many legal decisions has probably noticed sooner or later that courts are rather reluctant to commit themselves to positions that are more general than need be, in justifying a decision. It is of course true that the reason given in the present case must apply to other cases. If MacPherson cannot recover, and the reason he cannot is that an automobile is not an inherently dangerous item, then no other plaintiff in a case involving (in the same way) an item that fails to be (in the same sense) inherently dangerous can recover, and the only ways to avoid such an implication all involve backtracking and saying that this is not (quite, or really) the reason MacPherson cannot recover. Judges tend to judge in a way that allows additions as friendly amendments. One reason for this is that there is always the threat of being snubbed by the system later on: of being reversed on appeal, or of introducing an innovation that is then ignored by other courts on the same level, or (in the case of the highest court in the land) of having to reverse its own rulings in a case that raises sober second thoughts. Such things are at least embarrassing, and the embarrassment involved is the sort that a rational person would tend to find painful: the sort that is experienced when qualified people give reasons why one was wrong. Judges, then, will be under a certain amount of pressure to leave the law indeterminate at various points and in various respects, and there will accordingly be a presumption to read their rulings as intended to leave it this way.
In fact, a system that is to some extent indeterminate in the way I have just described is just the sort that should be desired by rational individuals who are subjects of the system. We shouldn’t want the rules to be made by people who trying to anticipate every possible contingency. Some things are best dealt with when they arise, for the simple reason that they can’t be anticipated, or can’t be foreseen in enough detail to form a sensible opinion about them. The system should be able to learn by experience, and that is just what case law does by leaving indeterminacies, lacunae to be filled as eventually as circumstances warrant.
Finally, as we have seen in the MacPherson cases, the judge in the present case can only resolve such ambiguities with a an exercise of creativity that alters in some way the legal core of the earlier decision. The aspect of the earlier case that enables it to determine or influence future decisions – the aspect of it that constitutes a contribution to the legal system – is not the concrete fact-pattern of the case, nor the concrete outcome the earlier judge arrived at, but the idea behind the outcome, an idea that is not limited to the present case but is, at least potentially, applicable to other cases as well. If the idea in the earlier case fails to completely determine a decision in the present case, this is the indeterminate element. The judge can only apply the earlier case by clarifying, extending, or qualifying this idea. This inevitably involves a judgment, inescapably individual and personal, about what is right, whether the approach one takes is characterized by Bartlettian constraint or Cardozian bravado.
For my purposes, the most important point in all this is the fact that, in instances in which the potentially precedent cases do not completely determine the resolution of the present case, a judge’s reliance on his or her own individual judgment is actually an essential ingredient of the practice of following a precedent. As paradoxical as this might sound, the independent, law-creating contribution of the present judge is part of what gives the precedent case the authority that it has. Another way to put the point is this: in such cases, following a precedent is not an instance of mindless subjection to authority, but necessarily involves the free exercise of individual judgment.
There remains a difficulty, however. It is this. So far, I have said that the practice of following precedent is compatible with the use of individual judgment, and is consequently so far compatible with any conception of rationality that makes reliance on one’s own judgment a necessary condition of it, on the grounds that using one’s own judgment to some extent overlaps with adherence to precedent: to some extent, interpreting and applying precedent cases includes such judgment. However, there is obviously a large non-overlapping area. As I have already said, both Cardozo and Bartlett are constrained (subject to fairly obvious exceptions involving consistency) to accept the outcomes of previous cases, and it is clear that their individual judgments about right and wrong are not involved in understanding, for instance, whether Mrs. Devlin won her case or not. What this result was, and that it was right, are accepted as given. Both Bartlett and Cardozo regard themselves as constrained to explain why such results were right. How can it be consistent with human reason to accept such a constraint?
I can think of several reasons, mostly suggested by things we have already seen in the series of cases leading up to MacPherson, why this can indeed be a rational thing to accept.
The most fundamental reason, from which all the others follow, is that, for a judge, the result in the earlier case does not consist, simply, in the brute fact of a Mary Devlin receiving eventually a certain amount of money. It has authority, for the judge, because it was a legal result, which means that it has meaning in virtue of its position in a system, a highly developed and artificial system, of standards and practices. As we have seen, the core of this system is the practice of giving reasons for outcomes. These reasons always have general implications that reach beyond the present result and, in virtue of these general implications are tested by other, highly qualified, givers of reasons. Moreover, these others test the reasons given by others in light of new facts, ones that in many cases are not, as I have pointed out, foreseen by the individual who originally presented the reasons. The new facts are apt to be unforseen because they are brought into the system from the world outside it. Indeed, the system has a certain built-in tendency to attract and bring to light facts that are embarrassing to the givers of reasons. Given that the reason for Mary Devlin’s victory will have implications for other cases, it will imply that other people have a chance to be awarded damages in relevantly similar circumstances. This means that other judges will indeed be asked to apply these reasons to new facts, and these facts will be brought before them by people who have no vested interest in or responsibility for showing that the reason given in Devlin’s case are reasonable ones but will have a powerful sort of interest in showing that those reasons (be they reasonable or unreasonable) are applicable to the new alleged facts. As to the question of whether what is being alleged really is factual, the system has meticulous and highly accurate mechanisms for determining the correct answer the that question, and bases its checking process on the answer it produces, and not on the allegations themselves.
Actually, I could go on here at much greater length, but you probably see what my general point is. The legal process is one that has several very pronounced characteristics: the giving of reasons, logically developing and confronting the full implications of these reasons, faithfully applying them to facts, facts that are produced by a disinterested and unbiased process and meticulously checked for accuracy. All these features are obviously characteristics of rationality itself. Not coincidentally, they are also characteristics of scientific method. My point here is not the admittedly silly one, that case law is “scientific.” I don’t see any more reason why law should be scientific than that science should be judicial. My point is rather that the method of case law, like the method of science, is (so to speak) an instance of the method of reason. What this means is that, for any rational individual who understands the legal system, the fact that what happened to Mary Devlin was a result in such a system will be evidence, potentially strong evidence, that it was the right thing, the thing that should have happened. To the extent that a legal system has the features that I have described here, it can be perfectly rational for a judge to accept a strong presumption that any given potentially precedent result is correct.
Of course, the fact remains that an individual judge might well think that the result in that case was actually wrong. Worse yet, the individual judge might think that there is a whole branch of the law that is (perhaps more and more) being decided on the wrong basis. This of course often happens. The sort of system I have described here will obviously not be static: it will change, and some judges will no doubt find that they are convinced that not only are individual cases decided wrongly, but a whole area is infected with (in the judge’s own view) the wrong ideas. Here there does seem to be a conflict between individual judgment and the law. Admittedly, this undeniable fact raises a number of deep and thorny issues but, as interesting as they are, the the thesis I have chosen to defend so far in this chapter does not require me to comment on any of them. My thesis, as you recall, is that conscience and, more generally, individual judgments about right and wrong do not necessarily conflict in the case of judges who are following the law: in such cases, to follow the law is not in itself and as such, to leave one’s private judgment aside. This is not to deny that they can conflict, and sometimes do.
 Martin P. Golding, Legal Reasoning (Peterborough, Ontario: Broadview Press, 2001), Ch. III. Edward H. Levi, An Introduction to Legal Reasoning (Chicago: University of Chicago Press, 1949), Ch. II. Golding’s discussion differs from mine in that his book is meant as a textbook and he accordingly does not present his own views in any detail. Otherwise, it supplements my discussion, in that it gives detailed treatment to some aspects that I treat more summarily. Levi’s discussion contrasts with mine in that it is written to some extent from a “legal realist” perspective, a point of view that diverges in significant ways from the one I will take here. I will comment on my differences with Levi below.
 10 Meeson and Welsby, 109 (1942).
 Thomas v. Winchester, 6 N. Y. 397 (1852).
 Loop v. Litchfield, 42 N. Y. 351 (1870). Some readers are no doubt still wondering why Ruggles wanted to hold on to any part of the Winterbottom rule, even as he reads it in Thomas. I think a brief look at the facts in Loop might help to explain this. In this case the lethal object was a cast iron flywheel which had an indentation in the rim due to a flaw in the casting process. Lyman Litchfield and an associate sold it to Leverett Collister, who requested that they modify it so that it could be used as a flywheel in a saw for cutting wood. The modifications included boring a hole in the wheel, and it was this hole that caused the fatal accident when Jeremiah Loop leased the machine from Collister five years later. Collister was aware that the modifications weakened the wheel. No doubt, one thing that probably inclined the courts to follow Winterbottom was something like the following. If the Loop court had ruled that the modified flywheel was imminently dangerous, it would have been ruling that Litchfield, and everyone else, has a legal duty never to sell such an object. But Collister (whether rightly or wrongly is a distinct question) wanted just such an article. Such a ruling would have limited the freedom of choice of the consumer (as well, more obviously, as that of the producer). If one supposes that such freedom of choice is a good thing, then the Winterbottom rule, for all its crudeness, did have one virtue: it preserved that freedom. On the other hand, no one wants to unknowingly swallow poison, and this suggests the sort of exception recognized in Thomas for the sake of consumer safety did not sacrifice any of the freedom protected by Winterbottom.
 Devlin v. Smith, 89 N. Y. 470 (1882).
 Torgeson v. Schultz, 192 N. Y. Rep. 156 (1908).
 Heaven v. Pender, 11 Q. B. D. 503 (1983). Like Devlin, which is nearly contemporary with it, Heaven concerns a collapsing scaffolding and recognizes the facts of the case as falling within an exception to the privity of contract rule.
 Statler v. Ray Mfg.Co., 195 N. Y. Rep. 498 (1909).
 MacPherson v. Buick Motor Company, 217 N. Y. 382 (1916).
 Cadillac Motor Car Co. v. Johnson, 221 Fed. Rep. 801.
 MacPherson, pp. 395 - 401.
 Both Golding and Levy (see fn. 1, above) stress the use of analogical reasoning in this series of cases.
 I have recently argued that this is what analogical reasoning is always like. See “Literature as Fable, Fable as Argument,” Philosophy in Literature ( October 2009, vol. 33 no. 2), pp. 369-385.
 Please notice that I am not denying that this is a good part of what Bartlett is doing, and of what all judges do and must do. I am only interested in denying that it is a complete description of the reasoning we have seen Bartlett doing in this case and, more generally, of what judges in general do.
 Admittedly, Bartlett is very sketchy in Torgeson about what the principle ought to be. The brief quote from Heaven is all that he says about it. The case of Statler v. Ray Mfg. Co. (195 N. Y. Rep. 498 (1909), decided by the same court the next year is less casual than Torgeson, and includes an interesting attempt to state clearly the nature of the principle involved that is consistent with Bartlett’s approach. (Though Statler was written by a colleague, Bartlett concurred with it). Immediately after repeating the quotation that Bartlett used from Heaven, the majority opinion in Statler explains: “This rule distinctly recognizes the principle that in the case of an article of an inherently dangerous nature, a manufacturer may become liable for a negligent construction which, when added to the inherent character of the appliance, makes it imminently dangerous, and causes or contributes to a resulting injury not necessarily incident to the use of such an article if properly constructed, but naturally following from a defective construction.” Talk of the article’s being imminently dangerous and of the injury following naturally has been reintroduced, but now it is perhaps clearer that these characteristics attach to the article as negligently made. However, it is also made clear that the article must also be dangerous in itself, even apart from negligence (ie., “inherently”). On the other hand, it need not be as dangerous as a bottle of belladonna, or a loaded gun.
 An Introduction, p. 2.
 An Introduction, p. 9.
 An Introduction, p. 2. On the same page, Levi describes the reasoning as a three-step process. “The steps are these: similarity is seen between cases; next the rule of law inherent in the first case is announced; then the rule of law is made applicable to the second case.” From the way in which Levi applies this model to actual cases, it is evident that the two cases are conceived as consisting of fact-patterns and results, and that the rule of law announced in the second step represents the present judge’s own inference to a principle that would explain the correctness of the the result in the first case.
 It is also relevant, I think, that Cardozo feels bound the say that he is following them, and in fact is using something very much like Bartlett’s approach. As I have already noted, he says: “It may be that Devlin v. Smith and Statler v. Ray Mfg. Co. have extended the rule of Thomas v. Winchester. If so, this court is committed to the extention.” Cardozo’s Devlin-made-me-do-it is almost comically transparent, but the fact that he feels compelled to make such a claim seems to show a sort of respect for stated rules of law. Part of this is acknowledged, and indeed insisted upon, by Levi, who says on the first page of the book that his principle theme will be that “the mechanism of the law” is completely different from “the pretense of the law.” The pretense is that “the law is a system of known rules applied by a judge,” while the actual mechanism is, in part, what I have described in fn. 18, above. An Introduction, p. 1. I can only wonder why, if the mechanism he describes is perfectly in order as it is, it is necessary to cover it with a lying pretense.
 The two people most visibly associated with this point, perhaps because they made most emphatically, are Friedrich Hayek and Lon Fuller. See Hayek’s The Constitution of Liberty (Chicago: University of Chicago Press, 1960), Ch. X, “Law, Commands, and Order,” esp. pp. 149-51, and Fuller’s The Morality of Law, rev. ed., (New Haven, Connecticut: Yale University Press, 1977), Ch. II, “The Morality that Makes Law Possible,” p. 46. I
 In the case of a court, such as the US Supreme Court, in which there are several judges and the judges vote on the outcome of their deliberations, the first person is plural, but it is the sort of we that implies an I.
 Though I cannot pretend to answer it, I would suggest that this further question – ie., that of what to do in the event that individual judgment and law do conflict – is really a problem in the professional ethics of judges. It seems obvious that, when the conflict surpasses a certain point, the judge has an obligation to resign. To administer a system one thinks is evil is to violate one’s own integrity. In that event, not only is it not necessary to endure a conflict between conscience an law, it is necessary not to do so. On the other hand, the question of what to do when the conflict falls short of this point is more complex and difficult. Fortunately, I am not committed to answering it here.