What is most striking about difficult proximate cause cases is that there is a straightforward sense in which the defendant’s conduct did cause the plaintiff’s injury. Negligence law famously utilizes a necessary-condition criterion, or “but-for” test, for actual causation: the question is whether but for the defendant’s conduct, the plaintiff would not have been injured. In Wagon Mound, we see that the defendant’s negligent conduct plainly qualifies as a cause of the plaintiff’s injury under the but-for test, leading one to the question of whether the defect in the plaintiff’s suit in that case (assuming there was a defect) could really be a defect falling within the rubric of “causation.”

The concept of proximate cause specifically targets that aspect of the concept of causation that is not captured by the but-for test. The standard lawyer or judge feels in a case like Wagon Mound that there may be something too attenuated about the connection between the tortious conduct and the injury. In this way, the proximate-cause requirement seems to presuppose a conception of a normal or standard causal connection, as opposed to one that is too attenuated or freakish. Courts commonly say that the harm for which plaintiff seeks compensation must be a “natural and probable” result of the negligent conduct. Yet it has long been evident that this phrase is not particularly helpful, for it is clearly not a necessary condition that the causal process leading to the injury be natural in the sense of being unmediated by additional human agency, and it is just as clearly not a necessary condition that negligent conduct rendered the injury probable in the sense of being more likely than not to occur. The degree of intervening human agency and the degree of probability are typically probative of the intuitive notion of proximate cause, and lawyers and judges certainly bring these factors into consideration when making a proximate-cause determination. However, this only deepens the puzzle, because we are led to wonder what the concept is that makes these considerations seem so clearly relevant.

Proposed Approaches

The classic “directness” case is Polemis (In re Arbitration Between Polemis and Furness, Withy & Co., Ltd. [1921] 3 K.B. 560). Plaintiff chartered a ship to defendant (a company). Stevedores hired by the defendant were loading cargo through a hold in the boat when one of them negligently caused a plank of wood to fall into the hold. When the dropped plank landed on something in the hold below, the friction between the two pieces of wood generated a spark. The spark ignited benzene gas that had leaked out into the hold, and a fire ensued, destroying the entire ship. The ship-owner plaintiff argued that the negligent dropping of the plank caused the fire and that the defendant should therefore be held liable for the ensuing damages. The defendant responded that this sequence of events was unforeseeable and therefore proximate cause was missing. The court sided with the plaintiff, reasoning that the linkage between the harm and the negligent act was actually quite direct, and that directness was the touchstone of proximate cause.

A third approach to proximate cause—the risk rule—has achieved less recognition among judges but has proved quite attractive to academics. The facts of Larrimore v. American Nat’l Ins. Co. nicely illustrate the idea ((1939) 89 P.2d 340 (Okla.)). A defendant landlord had left a can of rat poison in the kitchen area of the commercial tenant’s space. The rat poison, which was flammable, exploded because of the heat of the stove in the kitchen, causing serious injury. The plaintiff argued that it was negligent to leave rat poison in a kitchen, where food was prepared, and that this negligent conduct caused the injury. The defendant, in turn, argued that regardless of the imprudence of leaving rat poison in the kitchen, there was no liability unless it was foreseeable that rat poison was explosive. Otherwise, the causal chain is too attenuated. The court agreed with the defendant, couching its decision in the language of foreseeability. Subsequent commentators have pointed out a different feature of Larrimore—the absence of a nexus between the risk that made it negligent to leave rat poison in the kitchen—poisoning—and the harmful event that actually occurred—an explosion. The risk rule, first articulated as such by Harvard Law Professor Robert Keeton, says that proximate cause fails unless the injury for which the plaintiff seeks to impose liability is a realization of the risk that the defendant negligently took (Keeton 1963).

Andrews’s Palsgraf Dissent and the Reductionistic Challenge

A practicing lawyer asked to remember a proximate cause case from law school will typically recall the famous case of Palsgraf v. Long Island Railroad ((1928) 162 N.E. 99 (N.Y.)). Palsgraf’s notoriety as a proximate cause case is ironic, to say the least, given that Chief Judge Benjamin Cardozo’s majority opinion expressly disclaims all reliance on causation in overturning the plaintiff’s verdict (“[t]he law of causation, remote or proximate, is thus foreign to the case before us”). It is, in part, the eloquent dissenting opinion in Palsgraf that has led to its fame as a proximate cause case. And it is in great part the remarkable facts of the case that have left an indelible impression upon those who have studied it. A quick overview of those facts will help us focus on our topic.

As we have said, we cannot trace the effect of an act to the end, if end there is. Again, however we may trace it part of the way. A murder at Serajavo may be the necessary antecedent to an assassination in London twenty years hence. An overturned lantern may burn all Chicago. We may follow the fire from the shed to the last building. We rightly say the fire started by the lantern caused its destruction.

A cause, but not the proximate cause. What we do mean by the word “proxi mate” is, that because of convenience, of public policy, of a rough sense of justice, the law arbitrarily declines to trace a series of events beyond a certain point. This is not logic. It is practical politics.

(352 (Andrews, J., dissenting))

As Andrews points out, the need for a proximate cause requirement seems quite obvious: it is simply that but-for causation alone would yield far too much liability, and so courts must draw a line and then give a name to their line-drawing activity. Yet Andrews’s pragmatic and reductive approach appears to be inadequate for legal scholars in three respects. One is that it provides far too little guidance to trial judges, lawyers and legal actors, such as insurance companies, who must be able to estimate potential liabilities. A second, as we will see, is that pragmatic line drawing is really very different from what courts seem to be saying and doing. And a third is that most students, judges and lawyers actually come to believe that there is something which the idea of proximate cause is trying to get at—that there are cases that should qualify and cases that should not.

For all of these reasons, legal scholars have been engaged by the enterprise of trying to figure out a nonreductive analytical approach to proximate cause that is justifiable from a normative point of view. Such an analytical explanation appears important if one is to justify the proximate cause restriction, if one is to select a particular concept or concepts for elucidating and applying the restriction, and if one is to fend off the legal realist attack generated by Andrews. Incidentally, Cardozo himself analyzed the legal issues in Palsgraf in a way that did not require him to tangle with proximate cause, and ultimately led him to deny recovery (he reasoned that even if one assumes the conductors were negligent, their negligence did not involve any want of care in relation to Mrs. Palsgraf, and therefore did not constitute any breach of a duty owed to her, and therefore could not support a tort claim by her).

In what follows, I briefly survey the rationales that have been offered for each of the three leading approaches and the problems faced by each.

Proportionality

Two justifications are typically offered for limiting liability to the foreseeable consequences of negligent conduct. The first draws from the idea of proportionality. As Viscount Simonds wrote in Wagon Mound, “it does not seem consonant with current ideas of justice or morality that, for an act of negligence, however slight or venial, which results in some trivial foreseeable damage, the actor should be liable for all consequences, however unforeseeable and however grave, so long as they can be said to be ‘direct’” (422). The general idea would appear to be that norms of justice do not simply govern whether an actor shall be subjected to sanction or to liability, but also what mag nitude of sanction or liability is appropriate. Relatively greater liability may be imposed for relatively greater fault, but if the fault is relatively lesser, then the liability must not be too great. Broadly speaking, this is a cousin of the principle that the punishment must fit the crime.