Concept
Apportioning Harm to Causes
Joint and several liability versus apportionment of harm to causes
Apportioning Harm to Causes
Whether a plaintiff's damages are divided among the tortfeasors or charged in full to each depends on one question: can the harm be divided? Section 433A of the Restatement (Second) of Torts supplies the test. Damages for harm are apportioned among two or more causes where there are distinct harms, or where there is a reasonable basis for determining the contribution of each cause to a single harm.1 Damages for any other harm cannot be apportioned among two or more causes.2 When the proof of shares runs out, the common law makes each defendant whose breach was a substantial factor in producing the damage jointly and severally liable for the total loss.3
Divisibility is therefore the gate in every multiple-tortfeasor problem. A harm that passes through it is split among its causes. A harm that does not pass is the single indivisible injury for which, as the preceding section explained, each contributing tortfeasor answers in full.
The inquiry at the gate is causal. It asks how much of the injury each cause produced, not how careless each actor was. Fault percentages answer the second question, and they enter only where a state has abolished or partially abolished joint and several liability and makes each tortfeasor liable for a comparative-fault share of an indivisible injury.4 The distinction matters because a defendant with a small share of the fault can still be a proximate cause of the whole injury.5
The law charges each defendant with the harm that defendant produced. When the evidence ties a cause to one injury, or to one identifiable portion of an injury, charging that defendant with more would charge for harm the defendant did not cause. When the evidence cannot trace each cause's share, the same principle runs the other way. A defendant has no equitable claim against an injured plaintiff to be relieved of liability for damage the defendant proximately caused simply because another tortfeasor's negligence may also have caused the same harm.6 The second tortfeasor's arrival earns the first no discount, since the first is no less at fault and has caused no less harm.7 Full liability for the whole is thus a fallback for failed proof, and it leaves the commitment to causation intact.
The next three sections take up the grounds in turn: distinct harms, a reasonable basis for dividing a single harm, and the indivisible residue that § 433A(2) leaves behind.
Distinct Harms
Section 433A(1)(a) directs apportionment where there are distinct harms.8 A harm is distinct when it is a separate injury that the evidence ties to a particular cause and that does not overlap with the injury another cause produced. Each tortfeasor then answers for all of the injury it caused and for none of the other.9
A single injury that two actors cause together shows why distinctness matters. Fermi, a passenger in Joule's car, breaks a collar bone when the car collides with a truck driven by Edison, and both drivers were negligent. Edison and Joule did not act together and may not have known of each other's presence, yet each one's negligence is a but-for cause of the one injury, so each answers for all of it.10 A defendant whose negligence played no part in an injury is not a but-for cause of it, and has no reason to pay for it.
Body part can separate injuries. When two hunters in separate parties negligently fire at the same moment, and one wounds the plaintiff's arm and the other the plaintiff's leg, each is liable only for the harm traceable to that hunter, absent concert of action.11 Time can separate injuries too. In the illustrations to § 433A, a passenger breaks a right leg on a negligently maintained railroad platform and a month later breaks the left leg when a negligently driven bus collides with another vehicle. The railroad and the bus company are each liable only for the fracture of the one leg they caused.12
These examples show that the timing of the defendants' conduct does not control. The hunters fired at the same moment and still caused distinct harms, while successive impacts can join in a single harm, as the backhoe example below shows. The question in every case is whether the injuries are separate.
Successive wrongdoing produces a lopsided result. A tortfeasor whose conduct injures the plaintiff and foreseeably exposes the plaintiff to further injury is liable for the aggravation a later actor adds, but the later actor is not liable for the first injury.13 Driver A runs down P and leaves P helpless in the roadway, where B, also negligent, strikes P again. A answers for P's entire damage, and B only for the portion B caused.14 The same pattern governs a physician who aggravates an injury through negligent treatment: the original tortfeasor is liable for all of the damages, including those the malpractice caused, and the physician is jointly and severally liable for the malpractice damages.15 The illustrations to § 433A reach the same result when one driver fractures a pedestrian's skull and a second driver then breaks the pedestrian's leg. The first answers for both fractures and the second only for the leg.16
The explanation is causal. The original tortfeasor's conduct created the situation in which the later harm became possible and foreseeable, so it helped cause the aggravation as well as the first injury. The later actor entered after the first injury was complete and had no hand in it. Liability therefore overlaps on the aggravation and does not overlap on the original injury. Product cases follow the same line. A truck rams an otherwise nondefective car and the car rolls over, breaking the driver's arm in the first collision and his neck when the roof collapses in the second. The truck driver may be liable for all of the foreseeable damages under ordinary principles, while the manufacturer's exposure turns on the injury attributable to the second collision.17 A frequent mistake is to run the rule in both directions, either excusing the original tortfeasor because a physician or second driver intervened or making the later actor answer for the whole.
The backhoe problem shows the rule at work. Bethune negligently backs a backhoe into Maltby, a passing pedestrian, breaking her leg and knocking her down beside an excavation. DeWolfe, a worker in the pit, negligently throws a large stone out without looking, and it gives Maltby a severe concussion.18 Bethune and DeWolfe are joint tortfeasors for the concussion. Maltby would have avoided the stone had Bethune not put her in its path, and also had DeWolfe not thrown it, and DeWolfe's foreseeable negligence does not insulate Bethune.19 They are not joint tortfeasors for the leg, because DeWolfe's negligence had no part in causing it.20 If the jury finds both negligent and sets Maltby's total damages at $60,000, the judgment against Bethune is for the full amount, but DeWolfe owes only the concussion portion. A general verdict does not show how much that is. A special verdict form that asks the jury to state the leg damages and the concussion damages separately supplies the figures the court needs.
Reasonable Basis for Division
Section 433A(1)(b) supplies the second ground: damages are apportioned among two or more causes where there is a reasonable basis for determining the contribution of each cause to a single harm.22 This ground reaches losses that look undividable, such as a ruined crop or a contaminated site. Each is one loss, so the distinct-harms ground finds no separate injuries to assign, and the question becomes whether the evidence lets a fact-finder estimate how much each cause contributed. If the trier of fact can reasonably attribute portions of the loss to each defendant, damages ordinarily must be apportioned on that basis, and only where no fair and practicable method exists does the common law make each defendant liable for the total.23
Two features of the rule deserve separate attention. First, dividing by contribution matches liability to causal responsibility. Courts make each cause answer for the whole only when they cannot trace each cause's share, and where the evidence does trace the shares, holding one defendant to another's contribution charges that defendant for harm it did not produce. Second, the standard is reasonableness, not exactness. Nobody can say which sheep ate which blade of grass, and a demand for precision would leave every defendant liable for the whole. Requiring only a reasonable basis keeps apportionment available wherever an honest estimate exists.
The Restatement's own illustration involves cattle. When the cattle of two or more owners trespass on the plaintiff's land and destroy the crop, the aggregate harm is one lost crop, yet it may be apportioned among the owners according to the number of animals each owned, on the reasonable assumption that the harm each herd did is proportionate to its size.24 The headcount is a proxy, and it works because of that assumption. If each animal eats about the same, the number of animals stands in for the amount eaten, and the number is something the evidence can establish.
A textbook problem follows the same pattern. Farmer Jones's field of Kentucky bluegrass is eaten clean when sheep belonging to Shepherd and Herder escape their pens. Both owners were negligent, and 30 of Shepherd's sheep and 10 of Herder's got into the field.25 If the animals ate at comparable rates, the headcount supports an inference that Shepherd's sheep ate about three-quarters of the field and Herder's about one-quarter. Under the cattle illustration, damages divide on that basis, three-fourths to Shepherd and one-fourth to Herder, and neither owner is jointly and severally liable for the whole.26 The proxy is only as good as its assumption. If the evidence showed that one flock's animals ate far more per head than the other's, the headcount would lose its footing.
Courts set the bar for proof well below mathematical precision. In McAllister v. Pennsylvania R.R. Co., the plaintiff injured each leg six months apart, with some combined effect on the back, and the doctor who treated the first injury had died. The Pennsylvania court held that a reasonable basis of apportionment exists when some evidence supports the apportionment made, even though the proofs do not allow an exact dividing line between the injuries.27 A Hawaii case involving four accidents went further and held that the jury should be told to make a rough estimate and, if even that proved impossible, to divide the damages equally among the accidents.28 Some evidence and an honest estimate therefore suffice, and a gap in the proof does not by itself return the case to full liability. Treat the equal-division fallback as one court's answer to a particular proof problem, which shows how strongly some courts favor apportionment.
Pollution cases place both poles side by side. Where several riparian owners pollute a stream, each is liable only for the portion of the damage caused by that owner's contribution.29 Where two tortfeasors separately but simultaneously release contaminated water into a lake, and either release alone would have caused all the damage, no apportionment is possible and each may be held liable for the whole loss.30 Each situation involves a single harm, and what differs is whether the evidence measures contribution. The stream owners' separate contributions can be measured. In the lake, each release was sufficient by itself, so no figure shows what part of the damage either release produced.
The clearest modern application comes under CERCLA, the federal statute that lets the government recover hazardous-waste cleanup costs from the parties responsible for a contaminated site. Courts and the EPA long imposed joint and several liability on those parties, and divisibility defenses rarely succeeded.31 In Burlington Northern & Santa Fe Ry. Co. v. United States, the EPA spent $8 million cleaning up a chemical distributor's facility, part of which sat on land leased from two railroads, and sued the railroads for reimbursement. Looking to § 433A(1)(b), the Supreme Court held that a contamination event causing a single harm may still be divisible if the defendant demonstrates a rational basis for apportionment, and it upheld the district court's 9 percent allocation to the railroads.32 The evidence was geographic, chronological, and volumetric. The railroads' parcel was 19 percent of the facility, it was used for 45 percent of the relevant time, and it received 66 percent of the injurious spills. The district court multiplied those measures and added a 50 percent margin of error, and the Supreme Court found the result sufficiently grounded in fact.33 Each measure answered a separate question about where, when, and how much, and the margin of error openly acknowledged that the figure was an estimate. The defendants carried the burden of supplying that evidence, a point the next section develops.
The same standard failed at the Bluff Road waste site. The generators who had shipped waste there conceded that the environmental damage was a single harm, but argued that liability should follow the volume each had deposited compared with the total volume disposed of at the site. Applying § 433A, the court of appeals held that, in light of the conditions at Bluff Road, it could not accept volume as a basis for apportionment.34 The reason lies in how waste behaves. Waste that leaks from its containers mixes with the waste around it and diffuses through the soil, so removing one generator's waste would require removing all of it.35 A volume figure tells how much waste a generator sent, but once the waste has commingled, that figure no longer shows how much of the contamination or the cleanup cost the generator's waste produced. The headcount of animals and the volume of waste are the same kind of proxy. The first rests on an assumption the evidence supports, and the second rests on one the conditions at the site undermine.
Burlington Northern and Bluff Road both involve a single harm, and § 433A(1)(b) governs both, so the text of the rule does not separate them. The evidence does. At Burlington Northern the railroads' contribution could be measured along three independent dimensions of the site's history, while at Bluff Road the generators offered one proxy whose meaning the site's conditions had emptied out. The analysis turns on whether the offered measure traces each cause's actual contribution, and a conclusion drawn from the single-harm label alone is incomplete whichever way it comes out.
Indivisible Harms
Section 433A(2) completes the test with a residual rule: damages for any harm that is neither distinct nor reasonably divisible among its causes cannot be apportioned.37 The subsection operates as a default. When the evidence satisfies neither ground of § 433A(1), no division is available and the traditional rule of full liability takes over, so each tortfeasor whose conduct was a cause answers for all of the injury. Two negligent drivers collide, and one car bounces onto the sidewalk and injures a pedestrian. The pedestrian's injury is indivisible, and each driver is liable for all of it.38 If two separate accidents combine to kill the plaintiff, the death is likewise an indivisible result for which each is liable.39
The reason for the rule comes from what apportionment demands. A court can split damages only when some principle links a portion of the loss to a particular cause, as an arm links to the hunter who shot it. Where each cause was necessary to the whole injury, or would have been enough to produce it alone, no such link exists, and any division would be arbitrary. The law then has two options. It can leave a plaintiff who caused none of the harm to absorb whatever portion no one can assign, or it can hold each wrongdoer to the whole. It takes the second.
The point appears in its starkest form when two causes would each have produced the whole loss. In Kingston v. Chicago & N.W. Ry., the court, quoting an earlier decision, stated the rule for two causes, each attributable to the negligence of a responsible person, that concur in producing an injury, either of which would produce it regardless of the other. Whether the concurrence be intentional, actual, or constructive, “each wrongdoer, in effect, adopts the conduct of his co-actor,” and it is “impossible to apportion the damage.” The whole loss “must necessarily be considered and treated as an entirety.”40
That reasoning shows why the but-for test cannot decide these cases. Ordinarily a defendant is a cause only if the harm would not have happened without the defendant's conduct. Two negligently set fires, each enough to destroy the plaintiff's property, defeat that test, because each defendant can truthfully say the property would have burned anyway. If that answer ended the inquiry, both defendants would go free and the plaintiff would recover from no one. Courts call this duplicative causation, treat the but-for rule as inapplicable, and may hold both defendants jointly and severally liable for the whole loss.41 Neither defendant's fault nor causal impact is less in any degree merely because another tortfeasor also participated.42 Or, as the point has been put, “if two people kill a third, it is not the case that each has half-killed him.”43 The Third Restatement treats duplicative causation in its Apportionment of Liability § 27.44
One variation changes the culpability of the second cause and leaves the result alone. Suppose one of two sufficient causes is blameless, as when lightning sets the second fire. The preferred rule still holds the negligent defendant liable, since a fortuitous natural event should not excuse a defendant who would have been jointly and severally liable had the other cause been another person's culpable act.45
Indivisibility does not require that every cause be sufficient alone, and a single lawsuit can hold harms of both kinds. In the backhoe problem, the concussion is a harm for which Bethune and DeWolfe are joint tortfeasors, while the leg is Bethune's alone.46 The rollover example has an indivisible side as well. Suppose the evidence shows that a properly designed roof would also have collapsed somewhat, so that the driver's neck would have been injured to some extent anyway, but it cannot show how much injury a sound roof would have allowed. Where causation of the neck injury by the initial collision and the defect is “indivisible,” § 16(c) of the Restatement (Third) of Torts: Products Liability makes the manufacturer responsible for the defect fully liable for the indivisible neck harm.47
Indivisibility is one reason a defendant answers for more than the harm traceable to that defendant. Concert of action is the other, and the next section takes it up. Mere presence at the commission of a tort does not make an observer liable as a participant.48
The most frequent error in this area is to read a jury's percentage findings as if they divided the injury. A fault percentage compares how careless the actors were, while § 433A asks which part of the harm each cause produced. The California Supreme Court drew that line in American Motorcycle Ass'n v. Superior Court. The fact that a percentage figure can be assigned to one negligent defendant's relative culpability, the court reasoned, does not suggest that the defendant's negligence is not a proximate cause of the entire indivisible injury.49 A defendant whose fault is small by comparison with the others' is no less a cause for being small, and may be liable for the whole.50
The percentages do real work in the stairwell facts, though only where a jurisdiction gives them a role. A state that has replaced joint liability with several liability makes each tortfeasor liable for a comparative-fault share of the indivisible injury, so the landlord there would pay 20 percent of $200,000, or $40,000.51 That is a policy choice about charging defendants, and it leaves the injury indivisible, since each percentage is measured against the total damages for the whole injury. Modern reform of joint liability, taken up later in this chapter, returns to the distinction.