Ravon Owens v. Sherwin Williams Company

• Court: Court of Appeals for the Seventh Circuit
• Decided: April 15, 2021
• Docket: 20-1783
• Status: Published

Opinion

In the

United States Court of Appeals For the Seventh Circuit ____________________ Nos. 20-1774, 20-1776, 20-1777, 20-1780, 20-1781, 20-1782, 20- 1783, 20-1784 & 20-1785 GLENN BURTON, JR., RAVON OWENS, AND CESAR SIFUENTES Plaintiffs-Appellees,

v.

E.I. DU PONT DE NEMOURS AND COMPANY, INC., THE SHERWIN- WILLIAMS COMPANY, AND ARMSTRONG CONTAINERS, INC. Defendants-Appellants. ____________________

Appeals from the United States District Court for the Eastern District of Wisconsin. Nos. 07-CV-0303, 07-CV-0441, 10-CV-0075 — Lynn Adelman, Judge. ____________________

ARGUED DECEMBER 9, 2020 — DECIDED APRIL 15, 2021 ____________________

Before WOOD, SCUDDER, and ST. EVE, Circuit Judges. ST. EVE, Circuit Judge. These sprawling toxic-tort cases take us into the weeds of Wisconsin products liability law. The product at issue is white lead carbonate—a dry white powder historically used as the “pigment” in many lead-based paints. Paint has two essential components: the pigment, which hides 2 Nos. 20-1774 et al.

and protects the painted surface, and the liquid “vehicle,” which allows the pigment to be spread across the surface. Over time a consensus developed that lead-based paints were toxic and posed especially great dangers to young children, who were prone to chewing paint flakes or putting scattered lead dust into their mouths during critical stages of brain de- velopment. These dangers led the federal government to ban lead-based paint for residential use in 1978. Wisconsin fol- lowed suit two years later. Even after these bans, however, lead-based paint remained on the walls of many homes throughout the United States. The plaintiffs in these consolidated cases are three young men who grew up in Milwaukee homes that had lead-based paint on the walls. They were diagnosed with lead poisoning as young children in the 1990s or early 2000s. Years later, they filed these lawsuits against several manufacturers of white lead carbonate, seeking compensation for brain damage and other injuries resulting from their ingestion of lead paint par- ticles. The plaintiffs identified the paint pigment in their childhood homes as white lead carbonate, but they could not identify the specific company responsible for manufacturing the white lead carbonate that they ingested. To overcome this failure of proof, they relied on Thomas ex rel. Gramling v. Mal- lett, 701 N.W.2d 523 (Wis. 2005), in which the Wisconsin Su- preme Court adopted a “risk-contribution” theory of liability for plaintiffs suing manufacturers of white lead carbonate. The risk-contribution theory modifies the ordinary rule in tort law that a plaintiff must prove that a specific defendant’s con- duct caused his injury. It instead seeks to apportion liability among the “pool of defendants” who could have caused the injury. Nos. 20-1774 et al. 3

After years of pretrial litigation, the plaintiffs went to trial against five manufacturers of white lead carbonate. The jury found three of the manufacturers liable and awarded the plaintiffs $2 million each. The three defendants found liable (E.I. du Pont de Nemours and Company, Inc., the Sherwin- Williams Company, and Armstrong Containers, Inc.) now ap- peal. They challenge a long list of the district court’s pretrial, trial, and post-trial rulings. We see no error in many of these rulings, and we commend the district court for its thoughtful attention and diligent effort throughout this complex case. Nonetheless, we hold that the court committed three signifi- cant legal errors about the scope of Wisconsin products liabil- ity law. These errors shaped the trial and impermissibly ex- panded the defendants’ potential liability. Along with a sepa- rate error in the admission of certain expert testimony, they compel us to reverse the judgments and remand for further proceedings. I. Background Wisconsin’s risk-contribution theory is at the heart of this appeal. 1 In this section we trace the development of the risk- contribution theory to provide context for the plaintiffs’ law- suits. We then describe the facts and procedural history giv- ing rise to this appeal. A. Legal Background 1. Collins The risk-contribution theory has its origins in Collins v. Eli Lilly Co., 342 N.W.2d 37 (Wis. 1984). In that case, Therese

1 The parties agree that Wisconsin law governs these diversity cases. 4 Nos. 20-1774 et al.

Collins’s mother took diethylstilbestrol (DES) while pregnant with Collins in 1957–58. DES was a miscarriage-prevention drug that was on the market from 1947 to 1971. In 1971, med- ical research established a possible statistical link between DES and the later development of vaginal cancer in children exposed to DES in utero. After developing vaginal cancer in 1977, Collins sued various drug companies that had produced or marketed DES while her mother was pregnant. Under traditional tort law, Collins faced an “insurmount- able obstacle”: she could not identify which drug company had produced or marketed the DES that her mother had taken. Id. at 45. She could not do so because DES was a “fun- gible drug produced with a chemically identical formula.” Id. at 44. Moreover, hundreds of different companies had pro- duced or marketed DES. And, owing to the passage of time, records and evidence pertaining to drug companies’ produc- tion or marketing of DES and Collins’s mother’s prescription were largely unavailable. The Wisconsin Supreme Court was “faced with a choice of either fashioning a method of recovery for the DES case which will deviate from traditional notions of tort law, or permitting possibly negligent defendants to escape liability to an inno- cent, injured plaintiff.” Id. at 45. While acknowledging that “DES cases pose difficult problems,” the court concluded that, “as between the plaintiff, who probably is not at fault, and the defendants, who may have provided the product which caused the injury, the interests of justice and fundamental fairness demand that the latter should bear the cost of injury.” Id. at 49. Relying on its authority under the Wisconsin Consti- tution to fashion an adequate remedy, see Wis. Const. art. I, § 9, the court proceeded to adopt a risk-contribution theory of Nos. 20-1774 et al. 5

recovery for DES plaintiffs, adding that “this method of re- covery could apply in situations which are factually similar to the DES cases.” Collins, 342 N.W.2d at 49. The rationale for the risk-contribution theory was that the defendant drug compa- nies had, at the very least, “contributed to the risk of injury to the public” by producing or marketing a drug that turned out to have harmful side effects. Id. Moreover, the “possibly re- sponsible” drug companies were in a better position than the “innocent plaintiff” to absorb the cost of, and protect against, injuries from DES. Id. at 49–50. As for how the risk-contribution theory would apply: a plaintiff would need to sue at least one defendant and prove “that the plaintiff’s mother took DES; that DES caused the plaintiff’s subsequent injuries; that the defendant produced or marketed the type of DES taken by the plaintiff’s mother; and that the defendant’s conduct in producing or marketing the DES constituted a breach of a legally recognized duty to the plaintiff.” Id. at 50. While a plaintiff only had to sue one de- fendant, there were incentives to sue as many as possible, and named defendants were free to implead other drug compa- nies as third-party defendants. The court made clear that the risk-contribution theory was available for both negligence and strict liability claims. A plaintiff proceeding on either theory would have to prove the traditional elements of the claim, with one exception. Instead of proving that a defendant drug company caused the plain- tiff’s injuries, the plaintiff would only have to prove that the defendant “produced or marketed the type of DES taken by the plaintiff’s mother.” Id. at 51. If a plaintiff succeeded in making this prima facie show- ing, the burden would shift to the defendant to prove “that it 6 Nos. 20-1774 et al.

did not produce or market the subject DES either during the time period the plaintiff was exposed to DES or in the relevant geographical market area in which the plaintiff’s mother ac- quired the DES.” Id. at 52. To prevail on these defenses, a de- fendant would have to prove that “the DES it produced or marketed could not have reached the plaintiff’s mother.” Id. The goal of this burden-shifting procedure was to create “a pool of defendants which it can reasonably be assumed could have caused the plaintiff’s injuries.” Id. at 52. The court acknowledged that the risk-contribution theory could result in liability for innocent defendants, but it accepted this possi- bility as “the price the defendants, and perhaps ultimately so- ciety, must pay to provide the plaintiff an adequate remedy under the law.” Id. Finally, the court added that the doctrine of comparative negligence would permit the jury to equitably “apportion lia- bility among the defendants that have been unable to excul- pate themselves.” Id. at 53. In allocating liability, the jury could consider various factors, including: whether a defend- ant tested DES for safety and efficacy or took steps to protect or warn the public; the defendant’s role in securing govern- ment approval of DES; the defendant’s market share in the relevant area; and whether the defendant produced or mar- keted DES even after it knew the relevant risks. Id. at 54. 2. Thomas In Thomas, the Wisconsin Supreme Court considered whether to extend Collins’s risk-contribution theory to white lead carbonate pigment. 701 N.W.2d at 523. Steven Thomas suffered lead poisoning as a young child after he ingested lead paint particles in two of his childhood homes. Years later, Nos. 20-1774 et al. 7

Thomas sued his former landlords and several manufacturers of white lead carbonate seeking compensation for his injuries, which included brain damage and a heightened risk of future medical complications. From a legal standpoint, Thomas was in the same boat as Collins: he could not identify which company had manufac- tured the white lead carbonate that he had ingested as a child, given the “generic nature” of white lead carbonate, the large number of producers, and the lack of available records and evidence. Id. at 532. As such, he sought to hold the pigment manufacturers liable under Collins’s risk-contribution theory. The lower courts rejected Thomas’s proposed extension of Collins because Thomas was not without a remedy—he could, and did, recover from his landlords. The Wisconsin Supreme Court reversed, holding that Thomas’s case was similar enough to Collins’s case to warrant application of the risk-contribution theory. Id. at 557. As in Collins, the pigment manufacturers contributed to the risk of injury to the public by producing or marketing a harmful product. They were also better positioned to absorb the costs of the injuries. And, like Collins, Thomas could not identify “the precise manufacturer of the white lead carbonate that caused his injuries.” Id. at 559. On this point, the court rejected the manufacturers’ argument that white lead carbonate— which came in three different chemical formulas—was not fungible. It held that chemical identity was not necessary for fungibility. The court reasoned that a product could be “fun- gible” in three different senses: (1) functionally interchangea- ble; (2) physically indistinguishable; or (3) uniform in risk level. Id. at 560–61. The court concluded that, when viewing 8 Nos. 20-1774 et al.

the factual record in the light most favorable to Thomas, white lead carbonate was fungible in all three senses. The court proceeded to reject the manufacturers’ other ar- guments as to why Collins should not apply. First, the “dras- tically larger” window of liability—potentially several dec- ades or more, depending on when a house was built—was not a reason to let the pigment manufacturers off the hook. “[T]he Pigment Manufacturers’ argument must be put into perspec- tive: they are essentially arguing that their negligent conduct should be excused because they got away with it for too long.” Id. at 562. Next, the court rejected the argument that Collins should not apply because lead poisoning had multiple poten- tial causes and did not produce a “signature injury.” While it agreed with the argument’s premises, the court reasoned that these were issues for the jury. Id. at 563. Finally, the court re- jected the argument that Collins should not apply because the pigment manufacturers were not in exclusive control of the risk that their product created: [A]s doctors were the ones who prescribed the dosage of DES, so too were the paint manufacturers that mixed the amount of white lead carbonate in the paint. How- ever, the paint did not alter the toxicity of the white lead carbonate anymore than the pharmacist did by filling a prescription. To the contrary, at best, the paint manufacturers actually diluted the white lead car- bonate’s toxicity. In other words, the inherent danger- ousness of the white lead carbonate pigment existed the moment the Pigment Manufacturers created it. Id. Nos. 20-1774 et al. 9

The Wisconsin Supreme Court then set forth the elements of both negligence and strict liability claims for white lead car- bonate plaintiffs. As in Collins, Thomas’s burden was relaxed “only with respect to establishing the specific type of DES the plaintiff’s mother took, which, in this case, translates into the specific type of white lead carbonate Thomas ingested.” Id. Thomas only had to “prove that the Pigment Manufacturers produced or marketed white lead carbonate for use during the relevant time period: the duration of the houses’ exist- ence.” Id. at 564. If Thomas could make out a prima facie case, the burden would shift to the defendant to prove “that it did not produce or market white lead carbonate either during the relevant time period or in the geographical market where the house is located.” Id. Unlike in Collins, moreover, the pigment manufacturers might have additional defenses, “including that lead poisoning could stem from any number of sub- stances (since lead itself is ubiquitous) and that it is difficult to know whether Thomas’s injuries stem from lead poisoning as they are not signature injuries.” Id. at 564–65. Two Justices dissented, expressing dismay that the pig- ment manufacturers could be “held liable for a product they may or may not have produced, which may or may not have caused the plaintiff’s injuries, based on conduct that may have occurred over 100 years ago when some of the defendants were not even part of the relevant market.” Id. at 567–68 (Wil- cox, J., dissenting); see also id. at 590 (Prosser, J., dissenting) (“[T]his court has now created a remedy for lead paint poi- soning so sweeping and draconian that it will be nearly im- possible for paint companies to defend themselves or, frankly, for plaintiffs to lose.”). 10 Nos. 20-1774 et al.

3. Godoy In Thomas, the Wisconsin Supreme Court made clear that it was not reaching the merits of Thomas’s claim, but only de- ciding whether he was eligible to rely on the risk-contribution theory. See id. at 528–29 nn.2 & 4. A few years after Thomas, the court had occasion to consider the merits of a white lead car- bonate claim. See Godoy ex rel. Gramling v. E.I. du Pont de Nemours & Co., 768 N.W.2d 674 (Wis. 2009). Like Thomas, Ru- ben Baez Godoy sustained lead poisoning after he ingested white lead carbonate in his childhood home. Proceeding un- der Thomas, Godoy sued several manufacturers of white lead carbonate, asserting both negligence and strict liability claims. His legal theory was that white lead carbonate was defectively designed. As an initial matter, the court rejected Godoy’s argument that the product in question was actually “residential paint pigment.” Id. at 682. For one thing, Godoy’s complaint fo- cused exclusively on “white lead carbonate.” Id. Plaintiffs in products liability cases, the court explained, “must—at mini- mum—identify the product alleged to be defective.” Id. “Do- ing so puts the defendant on notice and allows the defendant to begin building a defense.” Id. More fundamentally, the product at issue could not be residential paint pigment be- cause the risk-contribution theory applied only to “fungible and identically defective” products. Id. In Thomas, that prod- uct was white lead carbonate pigment. The court made clear that it had never applied the risk contribution theory to “res- idential paint pigment,” or to “paint containing white lead carbonate.” Id. at 683 & n.10. On the merits, the court rejected Godoy’s argument that white lead carbonate was defectively designed. The court Nos. 20-1774 et al. 11

explained that Wisconsin courts had “discussed three catego- ries of product defects—manufacturing defects, design de- fects, and defects based on a failure to adequately warn.” Id. at 683. Godoy’s negligence and strict liability claims rested on a design defect, meaning “the design itself is the cause of the unreasonable danger.” Id. at 683–84. Godoy’s theory of design defect was that the presence of lead rendered white lead car- bonate defectively designed. The court disagreed: “A claim for defective design cannot be maintained where the presence of lead is the alleged defect in design, and its very presence is a characteristic of the product itself. Without lead, there can be no white lead carbonate pigment.” Id. at 685. Although Go- doy’s negligence and strict liability claims were “separate av- enues of recovery,” they both failed on this ground. See id. at 681 n.7. 4. Wisconsin Statute § 895.046 & Gibson In 2011, the Wisconsin legislature passed Wis. Stat. § 895.046, which effectively overruled Thomas while preserv- ing Collins. The statute reinstates ordinary causation princi- ples in all products liability cases, while carving out a narrow exception for plaintiffs like Collins who have no other remedy and whose injuries stem from a “complete[ly] integrated product” produced in “chemically and physically identical” forms and sold in generic packaging. § 895.046(4). In 2013, the legislature amended § 895.046 to make it retroactive. See § 895.046(2). The statute “assures that businesses may con- duct activities in this state without fear of being sued for in- definite claims of harm from products which businesses may never have manufactured, distributed, sold, or promoted, or which were made and sold decades ago.” § 895.046(1g). To 12 Nos. 20-1774 et al.

that end, the statute also includes certain time limits and join- der requirements. § 895.046 (4)(b), (5). In Gibson v. American Cyanamid Co., 760 F.3d 600 (7th Cir. 2014), we considered whether applying § 895.046 retroac- tively to extinguish already pending white lead carbonate claims would violate the Wisconsin Constitution’s due pro- cess guarantee. Applying the Wisconsin Supreme Court’s two-step framework, we held that it would. Id. at 609. First, Gibson had “a ‘vested right’ in his claims under Thomas’s risk- contribution theory.” Id. Second, § 895.046 lacked a rational basis because the private interest at stake outweighed the public interest that the statute promoted. Id. at 609–10. At both steps of the analysis, we held that Wisconsin Supreme Court precedent “dictate[d]” our conclusions. Id. (citing Matthies v. Positive Safety Mfg. Co., 628 N.W.2d 842, 861 (Wis. 2001) (hold- ing unconstitutional the retroactive application of a statute ex- tinguishing the right to recover on a common law negligence claim under an unmodified doctrine of joint and several lia- bility); Martin by Scoptur v. Richards, 531 N.W.2d 70, 93 (Wis. 1995) (holding unconstitutional the retroactive application of a statute extinguishing the right to recover an unlimited amount of non-economic damages)). We went on to hold that Thomas did not violate the Federal Constitution. Gibson, 760 F.3d at 627. Following our decision in Gibson, the Wisconsin Supreme Court split 3–3 (with one Justice not participating) on the first issue we addressed in Gibson: whether retroactively applying § 895.046 to white lead carbonate claims would violate the Wisconsin Constitution’s due process guarantee. Clark ex rel. Gramling v. Am. Cyanamid Co., 877 N.W.2d 117 (Wis. 2016). Nos. 20-1774 et al. 13

B. Factual and Procedural Background This legal background sets the stage for this case. The net effect of Thomas, § 895.046, and Gibson was a six-year window (2005–2011) in which plaintiffs in Wisconsin could rely on the risk-contribution theory to sue white lead carbonate manufac- turers for injuries arising from their ingestion of white lead carbonate as children. Approximately 170 such lawsuits were filed. The three plaintiffs here were the first to go to trial. The parties selected their cases as “bellwethers” for the other cases, which remain pending before the district court. 1. The Plaintiffs’ Claims The sad sequence of events leading to the plaintiffs’ claims will by now look familiar. As young children in the 1990s or early 2000s, Glenn Burton, Jr., Ravon Owens, and Cesar Sifuentes lived in Milwaukee homes (built between 1899 and 1922) that had paint containing white lead carbonate. While they were young children, the plaintiffs had elevated blood lead levels ranging from 31 (Burton) to 48 (Sifuentes) to 53 (Owens) micrograms per deciliter at their peak. For context, Wisconsin currently defines lead poisoning as “a level of lead in the blood of 5 or more micrograms per [deciliter] of blood.” Wis. Stat. § 254.11(9). (This number has steadily decreased over time.) Sifuentes and Owens were hospitalized and re- ceived chelation therapy to reduce their blood lead levels. Relying on Wisconsin’s risk-contribution theory, the plaintiffs—who could not identify the manufacturer of the white lead carbonate in their childhood homes—sued six white lead carbonate manufacturers for both negligence and strict liability, seeking to hold them liable for brain damage and other injuries resulting from their ingestion of white lead 14 Nos. 20-1774 et al.

carbonate as young children. The plaintiffs filed their lawsuits separately, but the district court consolidated them for trial. 2. The Defendants’ Production of White Lead Carbonate Only three of the defendants below are parties to this ap- peal: DuPont, Sherwin-Williams, and Armstrong. The plain- tiffs sued Armstrong in its capacity as successor-in-interest to the John R. MacGregor Lead Company. The three defendants below who are not parties to this appeal are: American Cyan- amid Company, NL Industries, Inc. (formerly known as the National Lead Company), and the Atlantic Richfield Com- pany. As explained below, the district court dismissed Amer- ican Cyanamid for lack of personal jurisdiction, National Lead settled, and the jury found Atlantic Richfield not liable. DuPont, Sherwin-Williams, and Armstrong each manu- factured white lead carbonate for a period in the twentieth century. DuPont manufactured white lead carbonate at a fac- tory in Philadelphia from 1917 to 1924. It incorporated its white lead carbonate into in its own paint products, which in- cluded ready-mixed paint and white lead-in-oil—a paste commonly sold to “master painters” who combined it with other ingredients to make their own paint. During this period, DuPont sold some of its ready-mixed paint products in Mil- waukee through a wholesaler. After it stopped making white lead carbonate in 1924, DuPont continued making and selling paint products containing white lead carbonate manufac- tured by other companies until 1966. From 1925 to 1946, Na- tional Lead manufactured white lead-in-oil for DuPont. As part of this arrangement, DuPont supplied National Lead with the ingredients and specifications for manufacturing white lead-in-oil. DuPont then sold the finished white lead- in-oil under its own brand name. Nos. 20-1774 et al. 15

Sherwin-Williams manufactured white lead carbonate from 1910 to 1947 and used its white lead carbonate as an in- gredient in its own paint products. Sherwin-Williams had a presence in the Milwaukee market during this time. Apart from its own white lead carbonate manufacturing, Sherwin- Williams sold products containing other companies’ white lead carbonate pigment from 1880 to 1969. In 1955, Sherwin- Williams began including warnings about the dangers of in- gesting lead on some of its lead-based paints. MacGregor (Armstrong’s predecessor-in-interest) manu- factured white lead carbonate at a plant in Chicago from 1938 to 1971. MacGregor used its white lead carbonate as an ingre- dient in its “Scotch Laddie” line of paint. Scotch Laddie paint appeared in advertisements and telephone directories in Mil- waukee between 1957 and 1971. MacGregor also sold its white lead carbonate to other paint manufacturers for use in their paint products. 3. Pretrial Rulings The district court made three significant pretrial rulings that shaped the trial. First, the court granted summary judg- ment to Sherwin-Williams and Armstrong on the plaintiffs’ negligent failure-to-warn claims. The court reasoned that the plaintiffs’ caregivers already knew that lead pigment was dangerous and needed no further warning. The court pointed out that Sherwin-Williams and other paint manufacturers started issuing product warnings in 1955—long before the plaintiffs were exposed in the 1990s or early 2000s. Meanwhile federal, state, and local governments began warning of the risks of lead-based paint in homes in the 1970s. The plaintiffs’ caregivers even testified at their depositions that they knew, before the plaintiffs’ exposure to lead-based paint, that 16 Nos. 20-1774 et al.

children should not eat paint chips because of the risk of lead exposure. For these reasons, the defendants had no duty to warn as a matter of law. Still, the court ruled that the plaintiffs could pursue negligence claims “based on the general duty of ordinary care.” Second, the court largely denied Sherwin-Williams’s and Armstrong’s motions for summary judgment on the plain- tiffs’ strict liability failure-to-warn claims. The court began by distinguishing between duty to warn in the negligence and strict liability contexts, reasoning that negligence requires a duty to warn the plaintiff whereas strict liability requires a duty to warn the “ordinary consumer” who purchased or used white lead carbonate or paint containing white lead car- bonate when the defendants put those products on the mar- ket. The court found a genuine issue of material fact as to whether the defendants had a duty to warn these ordinary consumers. At the same time, the court ruled as a matter of law that the defendants had no duty to warn in their capacity as pigment manufacturers who merely sold pigment to other companies or master painters. The court explained: “It makes little sense for the [white lead carbonate] manufacturer to owe a duty to warn the consumer at the moment that the pigment is delivered to the paint manufacturer, because the [white lead carbonate] manufacturer has no effective means of com- municating that warning.” The court thus granted summary judgment to the defendants on the plaintiffs’ strict liability failure-to-warn claims to the extent that the defendants sup- plied white lead carbonate to other companies. Third, less than six weeks before trial, the court denied DuPont’s motion to exclude evidence of its post-1924 con- tracts to purchase white lead-in-oil from National Lead. It also Nos. 20-1774 et al. 17

denied Sherwin-Williams’s motion to exclude evidence of its production or promotion of paint products containing an- other company’s white lead carbonate. The court reasoned that Thomas contemplated liability for producing or “market- ing” white lead carbonate and selling products that contained another company’s white lead carbonate amounted to “mar- keting” white lead carbonate. The defendants moved for re- consideration, arguing that the court’s ruling contradicted Thomas, Godoy, the plaintiffs’ longstanding theory of the case, and the court’s prior rulings. In the alternative, they moved to continue trial and reopen discovery so that they could pre- pare a defense according to this broader theory of liability. The court denied these motions, reasoning that Thomas put them on notice that they could be liable for marketing paint products containing another company’s white lead carbonate. The court also made several other relevant pretrial rulings: It granted summary judgment to the plaintiffs on the issues of whether white lead carbonate was “fungible” and whether Armstrong was the successor-in-interest to MacGregor. It de- nied summary judgment to Sherwin-Williams and Armstrong on the issue of whether § 895.046 barred the plaintiffs’ claims. It denied the defendants’ motion to exclude the testimony of two of the plaintiffs’ experts: Dr. Idit Trope and Dr. James Be- sunder. And it denied Sherwin-Williams’s motion to exclude evidence and arguments about its constitutionally protected product advertisements and association with industry groups. National Lead settled before trial. This was a significant development because, as the district court explained, “[p]laintiffs and defendants alike understand National Lead to have been a leading manufacturer of [white lead carbonate] 18 Nos. 20-1774 et al.

pigment with a significant presence in the Milwaukee market during the first half of the Twentieth Century.” National Lead settled pursuant to a “Pierringer release,” according to which “a tort plaintiff settles with a tortfeasor, reserves its right to pursue claims against other joint tortfeasors, and agrees to in- demnify the settling tortfeasor for any claims for contribution that non-settling tortfeasors might bring against the settling tortfeasor.” Bruner Corp. v. R.A. Bruner Co., 133 F.3d 491, 494 (7th Cir. 1998) (citing Pierringer v. Hoger, 124 N.W.2d 106 (Wis. 1963)). On its own initiative, and over the defendants’ objections, the court bifurcated the trial into two phases: a liability phase and an apportionment phase. Relying on Collins, the court ex- plained that phase one would yield a “pool of defendants” that could have caused the plaintiffs’ injuries. Phase one would also determine the plaintiffs’ overall damages. Phase two would focus on allocating the damages based on equita- ble factors like market share and the defendants’ respective roles in the industry. See Collins, 342 N.W.2d at 52–53. The court then made the related decision to exclude evidence of National Lead’s market share and role in the Milwaukee mar- ket from phase one, finding that such evidence would be con- fusing, irrelevant, and prejudicial. As the court explained, “National Lead’s unusual position as a settled defendant whose liability and proportional responsibility remained live issues presented significant legal and evidentiary issues.” While National Lead’s large market share in Milwaukee was relevant to the allocation of damages, it was irrelevant to whether other defendants were present in the Milwaukee market. Meanwhile, it would be all too easy for other defend- ants to point the finger at National Lead during the liability phase in hopes of exculpating themselves entirely. Nos. 20-1774 et al. 19

4. Trial The cases proceeded to trial against the remaining five de- fendants. The plaintiffs went to trial on the two claims that had survived summary judgment: ordinary negligence and strict liability failure-to-warn. To prove their claims, the plain- tiffs presented evidence of the dangers of lead-based paint to young children; awareness and recognition of these dangers among the medical and scientific communities, the paint in- dustry, and the defendants specifically; the defendants’ con- tinued production, marketing, promotion, and advertisement of white lead carbonate and products containing white lead carbonate; and the plaintiffs’ childhood ingestion of white lead carbonate and subsequent injuries. To prove their injuries, the plaintiffs relied largely on the expert testimony of Dr. Trope and Dr. Besunder. Dr. Trope, a neuropsychologist, performed neuropsychological evalua- tions on each plaintiff and concluded that each had neuropsy- chological impairments indicating brain damage. She ex- plained that a neuropsychological evaluation gauges brain function by testing various “domains” such as cognitive abil- ity, language functioning, visual motor functioning, memory, attention and concentration, and executive functions like judgment, organization, conceptualization, and information processing. She testified that abnormally large discrepancies across the plaintiffs’ functional domains led her to conclude that each plaintiff had brain damage. On cross-examination, Dr. Trope testified that the plaintiffs’ history of lead poisoning was a substantial contributing factor to their neuropsycholog- ical dysfunction. The plaintiffs objected during this cross-ex- amination that they had not offered Dr. Trope to prove causa- tion. 20 Nos. 20-1774 et al.

Dr. Besunder, a pediatric critical care physician and toxi- cologist who treats children with lead poisoning, testified to the fact, source, and extent of each of the plaintiffs’ injuries. Dr. Besunder did not personally evaluate any of the plaintiffs. Instead, he relied on Dr. Trope’s neuropsychological evalua- tions and general epidemiological studies on the effects of lead poisoning in children. Relying on the neuropsychologi- cal evaluations, Dr. Besunder testified that each of the plain- tiffs had brain damage. Like Dr. Trope, Dr. Besunder based this conclusion on “significant deviations” in the plaintiffs’ functional domains. Dr. Besunder further testified that lead poisoning caused plaintiffs’ brain damage. He explained that the plaintiffs’ functional deficits were consistent with patterns of brain damage in lead-poisoned children documented by the medical and scientific literature. He testified that he had reviewed the plaintiffs’ medical records to rule out other po- tential causes of their brain damage. Finally, Dr. Besunder attempted to quantify the plaintiffs’ brain damage. He testified that each of the plaintiffs had lost at least 10 IQ points due to lead poisoning. He explained that “several studies” had “attempted to quantitate the impact of lead levels up to approximately 20 to 30 micrograms per dec- iliter” and “all those studies have very similar results, that by the time your leads level is approaching the 25 to 30 range that child has lost approximately 10 IQ points.” He testified that children with blood lead levels above 30 would have lost ad- ditional IQ points, but he stuck with the “conservative esti- mate of 10 IQ points” because “the medical literature will only allow me to give an approximate quantitative estimate for lead levels up to 30.” He explained that the studies he relied on controlled for alternative causes. He conceded, however, Nos. 20-1774 et al. 21

that he had not published his methodology for quantifying IQ loss and that he did not have a baseline IQ for the plaintiffs. The defendants’ primary line of defense at trial was “chemical exculpation.” Each of the defendants called experts to testify to differences in the chemical composition of the paint samples taken from the plaintiffs’ childhood homes and the defendants’ residential paint formulas. At the close of evidence, the court dismissed American Cy- anamid for lack of personal jurisdiction. The court submitted the plaintiffs’ case against the other four defendants to the jury. The jury found DuPont, Sherwin-Williams, and Arm- strong liable on both claims. It awarded $2 million in damages to each of the plaintiffs. The jury found Atlantic Richfield not liable. The trial never reached phase two—where the jury would have apportioned the damages among the defend- ants—because the three defendants found liable chose to set- tle, assigning 12.5% of fault to National Lead and splitting the rest among themselves on a joint and several basis. 5. Post-Trial Rulings The defendants filed various post-trial motions. They each moved for judgment notwithstanding the verdict, arguing that Wisconsin’s judicial public policy factors precluded their liability. See Fandrey ex rel. Connell v. Am. Family Mut. Ins. Co., 680 N.W.2d 345, 350 (Wis. 2004). They also moved for a new trial, challenging the court’s pretrial legal rulings, the suffi- ciency of the evidence, and the court’s decision to bifurcate the trial. Sherwin-Williams and Armstrong moved for judg- ment as a matter of law on similar grounds. The court denied the defendants’ post-trial motions, with one exception: it agreed to remit Burton’s damages award 22 Nos. 20-1774 et al.

from $2 million to $800,000 because Dr. Besunder admitted during trial that, of the 10 IQ points that Burton lost due to lead poisoning, he lost six of them before he moved to the home that he focused on at trial. The defendants now appeal. II. Discussion The defendants raise a host of challenges on appeal. We focus primarily on three foundational issues concerning the scope of Wisconsin products liability law and the defendants’ corresponding potential for liability. First, all three defend- ants argue that the district court improperly extended Thomas by allowing the jury to find them liable in their capacity as paint manufacturers, rather than white lead carbonate manu- facturers. Second, Sherwin-Williams argues that the court er- roneously allowed the jury to find it liable on the negligence claims without proof of a product defect. Third, Sherwin-Wil- liams and Armstrong contend that the court erroneously al- lowed the jury to find them liable on the strict liability claims in the absence of a duty to warn or any proof that the lack of a warning caused the plaintiffs’ injuries. Our review of Wis- consin law convinces us that the defendants are correct on all three counts. We see no merit in the defendants’ remaining challenges, with one exception: we hold that the court abused its discretion in admitting Dr. Besunder’s testimony about the plaintiffs’ IQ loss. A. The Risk-Contribution Theory We begin with the defendants’ challenge to the court’s ap- plication of the risk-contribution theory. The defendants ar- gue that the district court improperly extended Thomas by al- lowing the plaintiffs to hold them liable in their capacity as Nos. 20-1774 et al. 23

manufacturers of finished paint products, and not just in their capacity as manufacturers of white lead carbonate. The district court denied the defendants’ motions in limine on this point, ruling that the plaintiffs could introduce evidence of the defendants’ production or sale of paint con- taining another company’s white lead carbonate. The court stood by its pretrial ruling when denying the defendants’ mo- tions for a new trial. We review both rulings for abuse of dis- cretion. Turubchuk v. S. Ill. Asphalt Co., Inc., 958 F.3d 541, 548– 49 (7th Cir. 2020). An evidentiary ruling that rests on a legal error is, by definition, an abuse of discretion. United States v. Chaparro, 956 F.3d 462, 474 (7th Cir. 2020). We will grant a new trial only if “the verdict is against the weight of the evidence, the damages are excessive, or if for other reasons the trial was not fair to the moving party.” Pickett v. Sheridan Health Care Ctr., 610 F.3d 434, 440 (7th Cir. 2010) (quoting Emmel v. Coca- Cola Bottling Co. of Chi., 95 F.3d 627, 636 (7th Cir. 1996)). Evi- dentiary errors warrant a new trial “if the evidentiary errors had ‘a substantial and injurious effect or influence on the de- termination of a jury and the result is inconsistent with sub- stantial justice.’” Fields v. City of Chi., 981 F.3d 534, 544 (7th Cir. 2020) (quoting Lewis v. City of Chi. Police Dep’t, 590 F.3d 427, 440 (7th Cir. 2009)). As set forth above, each of the defendants was both a white lead carbonate manufacturer and a manufacturer of finished paint products at various points in the twentieth century. They maintain that the risk-contribution theory applies to them only in their capacity as manufacturers of white lead carbonate. We agree; the Wisconsin Supreme Court has ex- pressly limited the risk-contribution theory to manufacturers of white lead carbonate. 24 Nos. 20-1774 et al.

1. Scope of Thomas In Thomas, the court invariably referred to the defendants as “pigment manufacturers.” The court used this term when setting forth the elements of Thomas’s cause of action. Thomas, 701 N.W.2d at 564. At one point, the court even drew a critical distinction between “pigment manufacturers” and “paint manufacturers.” The pigment manufacturers in Thomas had argued that they should not be liable because they “were not in exclusive control of the risk their product created ...” Id. at 563. In other words, the pigment manufacturers attempted to shift the blame to paint manufacturers—the companies that sold lead-based paints directly to consumers. The court re- jected this argument. In doing so it analogized paint manu- facturers to DES prescribers: while paint manufacturers mixed white lead carbonate into paint, they “did not alter the toxicity of the white lead carbonate anymore than the phar- macist did by filling a prescription.” Id. If anything, “the paint manufacturers actually diluted the white lead carbonate’s tox- icity.” Id. (emphasis added). “[T]he inherent dangerousness of the white lead carbonate pigment existed the moment the Pigment Manufacturers created it.” Id. (emphasis added). We noted the significance of this passage in Gibson. In re- jecting the argument that Thomas had discriminated against out-of-state corporations by imposing the risk-contribution theory against out-of-state pigment manufacturers but not in- state paint makers and retailers, we commented: “This argu- ment makes an inferential leap too far, and also ignores Thomas’s discussion of pigment manufacturers’ greater culpa- bility, when compared to paint makers and retailers.” Gibson, 760 F.3d at 627 n.11 (describing Thomas as holding that “if Nos. 20-1774 et al. 25

anything, paint makers and retailers reduced the risk of harm by diluting the lead pigment”). Finally, Godoy confirmed that the risk-contribution theory applies only to white lead carbonate pigment. The lower court had “mistakenly stated” that the product in Thomas was “paint containing white lead carbonate.” Godoy, 768 N.W.2d at 682 n.10. The Wisconsin Supreme Court was clear: “This statement misconstrues our holding in Thomas.” Id. Similarly, the court rejected Godoy’s argument that the risk-contribu- tion theory could apply to “residential paint pigment.” Id. at 683. In short, the Wisconsin Supreme Court has held that the risk-contribution theory applies to white lead carbonate manu- facturers to the extent that they manufacture white lead car- bonate. Applying risk contribution to paint manufacturers or finished paint products would extend Thomas beyond its bounds. See Thomas, 701 N.W.2d at 570 (Wilcox, J., dissenting) (“It is important to emphasize that the industry defendants are being sued in their capacity as manufacturers of white lead carbonate and not the finished product, paint.”). Admittedly, distinguishing between pigment manufactur- ers and paint manufacturers—and between pigment and paint—can be confusing because pigment does not reach res- idential consumers unless paint manufacturers sell them a fin- ished paint product. Moreover, many companies, including the defendants here, historically manufactured both pigment and finished paint products. Still, the Wisconsin Supreme Court has made clear that such companies are liable under the risk-contribution theory only in their capacity as manufactur- ers of white lead carbonate pigment. As a practical matter, this means that a paint manufacturer cannot be liable under the 26 Nos. 20-1774 et al.

risk-contribution theory for selling a finished product that contains another company’s white lead carbonate. Only the manufacturer of the white lead carbonate can be liable. Sherwin-Williams and Armstrong blur the issue by argu- ing that they cannot be held liable as paint manufacturers even if they sold paint containing their own white lead car- bonate. This argument is technically correct but inconsequen- tial. Under Thomas, a company that manufactures white lead carbonate can be liable under the risk-contribution theory if its white lead carbonate could have injured the plaintiff. 701 N.W.2d at 565. It does not matter how the white lead car- bonate reached the plaintiff—whether through the same com- pany’s paint or another company’s paint. See id. at 563 (reject- ing the pigment manufacturers’ argument that they could not be liable because paint manufacturers altered their products before they reached the residential consumers). Either way, risk contribution applies because the company manufactured white lead carbonate that could have injured the plaintiff. To the extent that Sherwin-Williams and Armstrong suggest that a company can avoid liability entirely under the risk-contri- bution theory by fully integrating its products and only sell- ing white lead carbonate as an ingredient in its own paints, that argument is nonsensical and has no basis in Thomas. The district court based its more expansive reading of Thomas on the Wisconsin Supreme Court’s use of the word “marketed.” As the district court saw it, a company that sells lead-based paint is “marketing” lead pigment, even if it is not “producing” it. Thus, the court reasoned, a paint manufac- turer could be liable for selling paint containing another com- pany’s white lead carbonate. We disagree. For one thing, this is a strained interpretation of “marketing” in this context. A Nos. 20-1774 et al. 27

bookseller does not “market” paper, even though paper is a component of books. More fundamentally, Thomas foreclosed this broad conception of “marketing” because it drew a dis- tinction between pigment manufacturers and paint manufac- turers and limited its holding to the former. If selling a fin- ished product qualified as “marketing” one of the product’s ingredients, a company that had never manufactured pig- ment could be liable under Thomas. Yet Thomas carefully lim- ited its holding to “pigment manufacturers,” and Godoy rein- forced that limitation. Thomas’s reference to “marketing” must be read in context. Thomas speaks of “pigment manufac- turers” that produce or market white lead carbonate. Thomas, 701 N.W.2d at 564. It must be the “pigment manufacturers” doing the producing or marketing—not paint manufacturers or anyone else. For these reasons, we hold that the district court ex- tended—and did not merely apply—the holding of Thomas. We further hold that the court’s extension of Thomas was a le- gal error. The plaintiffs do not dispute that the key feature of DES and white lead carbonate—fungibility—does not carry over to paint, so the core rationale of Collins and Thomas is in- applicable to paint and paint manufacturers. See Godoy, 768 N.W.2d at 682–83. Apart from that, Wis. Stat. § 895.046 for- bade the district court’s extension of Thomas. To be sure, Gib- son holds that § 895.046 may not retroactively extinguish the cause of action that Thomas recognized. But the plaintiffs’ claims in this case went beyond Thomas. The plaintiffs had a vested right to Thomas; they did not have a vested right to ex- tend Thomas. 28 Nos. 20-1774 et al.

2. Prejudice The court’s error requires a new trial because it signifi- cantly expanded the scope of the defendants’ potential liabil- ity and the evidence at trial. It also deprived the defendants of the opportunity to build an appropriate defense. See Godoy, 768 N.W.2d at 682–83. For years, the defendants had litigated under the (valid) assumption that they were liable only in their capacity as pigment manufacturers. Less than six weeks before trial, however, the defendants learned for the first time that they were also subject to liability as manufacturers of paint products containing white lead carbonate. The court’s last-minute, legally incorrect ruling left them scrambling to adapt their defenses to the plaintiffs’ newly enlarged theory of liability. Consider the prejudice to DuPont. DuPont manufactured white lead carbonate for a narrow seven-year period (1917 to 1924). The court’s erroneous legal ruling enlarged DuPont’s potential window of liability from that seven-year period to a nearly fifty-year period (1917 to 1966) during which DuPont manufactured paint products containing white lead car- bonate. 2 In keeping with the court’s pretrial ruling, the plaintiffs introduced evidence of DuPont’s post-1924 knowledge and conduct and asked the jury to find DuPont liable for its actions through 1966. They also seized on weaknesses in DuPont’s primary defense that resulted directly from the court’s erro- neous ruling. Through no fault of its own, DuPont was

2 The plaintiffs suggest that DuPont manufactured white lead car- bonate until 1946. But the only evidence they cite indicates that National Lead manufactured white lead-in-oil for DuPont during those years. Nos. 20-1774 et al. 29

unprepared to rebut these attacks. The court’s late-stage rul- ing meant that DuPont had to expand its chemical-exculpa- tion defense to cover all the paint formulas that DuPont had used during the additional 42 years (1924 to 1966) that were now on the table. At trial, however, the court precluded DuPont’s paint-chemistry expert from opining on any of DuPont’s post-1924 paint formulas because DuPont had not disclosed any of these opinions before trial. The plaintiffs took full advantage of this gap in DuPont’s defense during closing arguments. They argued that DuPont’s chemical-exculpation defense failed because its expert “came into this courtroom and offered opinions on the years 1917 to 1924” when “you saw that chart that indicated that they made white lead car- bonate products up until 1966.” The prejudice was similar for Sherwin-Williams, whose window of liability expanded from a 37-year period in the first half of the twentieth century (1910 to 1947) to a 90-year period (1880 to 1969) during which it produced or marketed paint products containing white lead carbonate. Armstrong is in a unique position because its predecessor- in-interest MacGregor was both a pigment manufacturer and a paint manufacturer at all relevant times. While it is still pos- sible that the court’s error prejudiced Armstrong, Armstrong has made no attempt on appeal to explain how that is so. In the end, we need not resolve this issue because, as we explain below, Armstrong (like Sherwin-Williams) is entitled to relief on other grounds. DuPont is the only defendant that rests its case on this issue. Given the prejudice we have described, DuPont is entitled to a new trial on both claims that went to trial. 30 Nos. 20-1774 et al.

3. Gibson & Fungibility We reject the defendants’ other challenges to the court’s application of the risk-contribution theory. First, Sherwin- Williams and Armstrong ask us to overrule Gibson. They claim that subsequent developments in the law have under- mined Gibson’s analysis. The district court denied the defend- ants’ motion for summary judgment on this point. We review that ruling de novo, viewing the evidence in the light most favorable to the non-moving party. Turubchuck, 958 F.3d at 548; see also Chemetall GMBH v. ZR Energy, Inc., 320 F.3d 714, 719 (7th Cir. 2003) (explaining that even after a trial we may review an earlier denial of summary judgment if the motion for summary judgment raised legal, and not factual, issues). We decline to revisit Gibson. When we weigh in on an un- settled issue of state law, our conclusion binds us until the state’s supreme court says otherwise. Reiser v. Residential Funding Corp., 380 F.3d 1027, 1029–30 (7th Cir. 2004). The Wis- consin Supreme Court has not said otherwise. In fact, it re- cently split 3–3 on the very question that we addressed in Gib- son. Clark, 877 N.W.2d 117. It would be hard to imagine better evidence that an issue of state law remains unsettled. Moreo- ver, the cases upon which we relied in Gibson remain good law. See Matthies, 628 N.W.2d at 861; Martin, 531 N.W.2d at 93. The defendants suggest that Lands’ End, Inc. v. City of Dodge- ville, 881 N.W.2d 702 (Wis. 2016), rejected the balancing anal- ysis that Gibson drew from Matthies and Martin. But Lands’ End did not overrule or undermine either case. Only the concur- rence questioned the “balancing test” that they applied. Id. at 733 (Ziegler, J., concurring). The defendants also cite Bank Markazi v. Peterson, 136 S. Ct. 1310 (2016), but that separation of powers case did not change Wisconsin law. Nos. 20-1774 et al. 31

Second, Armstrong argues that the district court erred in ruling before trial that white lead carbonate was fungible as a matter of law. It maintains that fungibility was a question for the jury. The court resolved this issue at summary judgment, so again our review is de novo, though we review the court’s denial of a new trial on this basis for abuse of discretion. Turubchuck, 958 F.3d at 548; see also Chemetall, 320 F.3d at 719. We easily reject Armstrong’s contention that fungibility is a fact question for the jury to resolve. If fungibility were a fact question, then Thomas would not have held that white lead carbonate was fungible; it would have held instead that there was a fact issue as to fungibility that precluded summary judgment for the pigment manufacturers. Yet the court held, on the record before it, that “white lead carbonate is fungi- ble”—not merely that the pigment manufacturers were not entitled to summary judgment. Thomas, 701 N.W.2d at 561. That holding would make no sense if only the jury could de- cide fungibility. The role of fungibility in the risk-contribution theory fur- ther confirms that it is a legal issue for the court. Fungibility is a prerequisite to applying the risk-contribution theory. The Wisconsin Supreme Court considers it when deciding whether a plaintiff lacks an adequate remedy at law, such that the Wisconsin Constitution authorizes the court to develop one. See Collins, 342 N.W.2d at 44–45; Thomas, 701 N.W.2d at 559–562. Judges, not juries, decide whether the plaintiff has a cause of action, and what that cause of action looks like. See Paul F. Kirgis, The Right to a Jury Decision on Questions of Fact Under the Seventh Amendment, 64 Ohio St. L.J. 1125, 1162 (2003) (“[T]he judge always has the exclusive authority to decide which questions should be asked.”). As a practical matter, it 32 Nos. 20-1774 et al.

would be rather bizarre if the parties had to go to trial just to figure out what legal theory the plaintiff could pursue. We acknowledge that Wisconsin’s civil jury instructions list fun- gibility as an element of a risk-contribution claim. Wis. JI— Civil 3295. Like the district court, we conclude that the jury instructions rest on an incorrect reading of Collins and Thomas. The harder question is whether Thomas established the fungibility of white lead carbonate for future cases. The Wis- consin Supreme Court has sent mixed signals on this ques- tion. Collins stated flatly that DES was fungible, without re- gard to the facts of the case. 342 N.W.2d at 44. Thomas, by con- trast, tied its fungibility determination to the facts of the case, implying that its determination may not necessarily extend to future cases. 701 N.W.2d at 559 n.47, 561. Godoy, meanwhile, proclaimed that Thomas had established the fungibility of white lead carbonate pigment as a matter of law: “In Thomas, we concluded that for the purposes of risk-contribution, white lead carbonate pigment is fungible, and all manufacturers of white lead carbonate pigment could be held jointly and sev- erally liable for injuries caused by the product.” Godoy, 768 N.W.2d at 683. We need not resolve this issue because, even if the fungi- bility of white lead carbonate hinges on the facts of a particu- lar case, Armstrong has not shown that the white lead car- bonate in the plaintiffs’ homes was not fungible under any definition that Thomas considered. Armstrong states, in con- clusory fashion, that it presented evidence in the district court that different types of white lead carbonate have different chemical compositions and physical properties that permit their identification and alter their risk level. But Thomas re- jected the argument that fungibility requires chemical Nos. 20-1774 et al. 33

identity. 701 N.W.2d at 559. Even if Armstrong presented ev- idence that the physical properties of white lead carbonate vary, it does not explain what those variances are or why they make manufacturer identification any easier. As such, the court properly resolved fungibility at summary judgment. B. Negligence 1. Product Defect Moving on, Sherwin-Williams alone argues that the dis- trict court erred in ruling that the jury could find it negligent in the absence of a product defect. In its view, the court’s grant of summary judgment to the defendants on the plaintiffs’ neg- ligent failure-to-warn claims should have been the end of the plaintiffs’ negligence claims, given that failure to warn was the only product defect that the plaintiffs alleged. The district court denied Sherwin-Williams’s motion for judgment as a matter of law on this point, ruling that Sherwin- Williams could face liability “based on the general duty of or- dinary care” even if its products were not defective. We re- view the court’s ruling de novo. Turubchuk, 958 F.3d at 548. Judgment as a matter of law is proper if “a reasonable jury would not have a legally sufficient evidentiary basis to find for the party on that issue.” Fed. R. Civ. P. 50(a)(1). In apply- ing this standard, we view the evidence in the light most fa- vorable to the verdict. Turubchuk, 958 F.3d at 548. In Dippel v. Sciano, 155 N.W.2d 55 (Wis. 1967), the Wiscon- sin Supreme Court adopted the products liability rule of the Restatement (Second) of Torts § 402A (1965). Id. at 63. Section 402A supplies the elements of a strict products liability claim, but its rule “is not exclusive, and does not preclude liability based upon the alternative ground of negligence of the seller, 34 Nos. 20-1774 et al.

where such negligence can be proved.” Restatement (Second) of Torts § 402A cmt. a; Godoy, 768 N.W.2d at 681 n.7. In recent years, many states have adopted the Third Re- statement’s formulation of a products liability claim, which “eschewed the doctrinal labels ‘strict liability’ and ‘negli- gence’” and “defined the categories functionally, according to their required elements of proof.” Godoy, 768 N.W.2d at 681; see also Restatement (Third) of Torts: Products Liability § 2 cmt. n (1998). In 2011, the Wisconsin legislature adopted the Third Restatement’s rule, but only for strict liability claims. See Wis. Stat. § 895.047. The plaintiffs here filed suit before 2011, so the parties agree that § 895.047 does not apply to their claims. Godoy explains the basic differences between strict liability and negligence under Wisconsin products liability law. Strict liability and negligence are “separate avenues of recovery” with “substantively different” elements. Godoy, 768 N.W.2d at 681 n.7. Strict liability “focuses on the nature of the defend- ant’s product, whereas liability in negligence ‘hinges in large part on the defendant’s conduct under circumstances involv- ing a foreseeable risk of harm.’” Id. (quoting Green v. Smith & Nephew AHP, Inc., 629 N.W.2d 727, 745 (Wis. 2001)); see also Morden v. Cont’l AG, 611 N.W.2d 659, 673 (Wis. 2001); Greiten v. LaDow, 235 N.W.2d 677, 683–86 (Wis. 1975) (controlling opinion of Heffernan, J.). Despite these differences, the two claims have at least one thing in common: “Both causes of action require a plaintiff to prove that the product causing injury was ‘defective.’” Godoy, 768 N.W.2d at 681 n.7 (citing Wis. JI—Civil 3200); accord Mor- den, 611 N.W.2d at 673 (“The coexistence of the two theories [i.e., negligence and strict liability] has sparked confusion and Nos. 20-1774 et al. 35

criticism because both rely on an underlying product de- fect.”). As for what constitutes a product defect, Godoy ex- plains that “Wisconsin cases have discussed three categories of product defects—manufacturing defects, design defects, and defects based on a failure to adequately warn.” 768 N.W.2d at 683. Thus, a negligence claim must be predicated on one of those three categories of product defects. See id. at 681 n.7, 683. We have acknowledged, as Morden did, that distinguish- ing between strict liability claims and negligence claims in the products liability context can be confusing. Insolia v. Philip Morris Inc., 216 F.3d 596, 603 (7th Cir. 2000); Flaminio v. Honda Motor Co., Ltd., 733 F.2d 463, 467 (7th Cir. 1984). Indeed, Wis- consin’s continued distinction between the two claims has generated significant criticism. Morden, 611 N.W.2d at 673; Sharp ex rel. Gordon v. Case Corp., 595 N.W.2d 380, 388 (Wis. 1999) (summarizing criticism but declining to overrule prece- dent); Restatement (Third) of Torts: Products Liability § 2 cmt. n (discussing the “mischief caused by dual instructions on both negligence and strict liability”). We need not wade into this debate. We do note, however, that the claims’ shared re- quirement of a product defect has been at the heart of the crit- icism. Morden, 611 N.W.2d at 673; see Sharp, 595 N.W.2d at 388; Restatement (Third) of Torts: Products Liability § 2 cmt. n. Requiring a product defect for negligence claims makes sense because otherwise a defendant might be found negli- gent merely for making and selling a potentially dangerous product. “It is boilerplate law that, merely because a product or an operation is not as safe as possible, because there are better methods of manufacture or performing an operation does not lead to the conclusion that the method employed was 36 Nos. 20-1774 et al.

undertaken with a lack of ordinary care or the product was defective.” Greiten, 235 N.W.2d at 685 (controlling opinion of Heffernan, J.); see also Collins v. Ridge Tool Co., 520 F.2d 591, 594 (7th Cir. 1975) (“[A] manufacturer is under no duty to pro- duce accident or fool-proof products. Neither is the manufac- turer an insurer that its product is incapable of producing in- jury.”) (citation omitted). Allowing a claim of negligence without a product defect, as the district court did here, allows a jury to find the defendant negligent in the absence of any actual negligence, whether in the design, manufacture, or marketing of a product. There is a name for this type of liabil- ity—it is called strict liability, not negligence. See Restatement (Second) of Torts § 402A(2)(a) (allowing liability even if “the seller has exercised all possible care in the preparation and sale of his product”). The district court’s theory was that the defendants could be found negligent because they continued to sell white lead carbonate pigment for residential uses while knowing the dangers that it posed to homeowners and their children. It found support for this broad theory of liability in Judge Learned Hand’s famous B

underpinning Wisconsin negligence law. But the court should not have relied on Carroll Towing to create a pathway to liabil- ity that Wisconsin courts had foreclosed. In any event, Carroll Towing does not support the court’s theory of liability. Carroll Towing was not a products liability case, and it certainly did not hold that a company could be found negligent merely for marketing a dangerous product. Critically, Carroll Towing in- volved a negligent act: the defendant there was liable because it left a barge unattended. Here there is no negligent act to speak of, apart from Sherwin-Williams’s sale of a potentially dangerous product. Given these distinctions, the court should not have transposed the B

only whether risk contribution could apply. Moreover, Thomas’s claim did rely on a product defect—failure to warn. See Thomas, 701 N.W.2d at 528 n. 2 (declining to reach defend- ants’ arguments about failure to warn); id. at 569 n.2 (Wilcox, J., dissenting) (“The claims before this court are predicated on the defendants’ failure to warn of the dangers of their prod- uct.”); Godoy, 768 N.W.2d at 683 (“Thomas was based on fail- ure to warn claims.”). Even if Thomas were susceptible to the plaintiffs’ reading of it, Godoy forecloses that reading by providing that negligence claims in the risk-contribution con- text require a product defect. The absence of a product defect forecloses recovery on the plaintiffs’ negligence claims. The plaintiffs concede, as they must, that their negligence claims do not rest on a product de- fect. The district court held at summary judgment that the plaintiffs could not proceed on a negligent failure-to-warn theory. And the plaintiffs have never asserted a design defect or manufacturing defect. As Godoy makes clear, those are the only types of product defects that Wisconsin courts have rec- ognized. With no product defect, there can be no negligence liability. This means that Sherwin-Williams is entitled to judg- ment as a matter of law on the plaintiffs’ negligence claims. 2. Duty & Causation We reject Sherwin-Williams’s other attacks on the negli- gence verdict. Apart from the lack of a product defect, Sher- win-Williams objects to the district court’s duty analysis and its jury instruction on causation. Duty and causation are es- sential elements of a negligence claim under the risk-contri- bution theory (as they are in tort law generally). Thomas, 701 N.W.2d at 564. Nos. 20-1774 et al. 39

The court’s duty analysis was sound. Under Thomas, a plaintiff must prove that “the Pigment Manufacturers’ con- duct in producing or marketing the white lead carbonate con- stituted a breach of a legally recognized duty to [the plain- tiff].” Id. Setting aside the district court’s failure to require a product defect, which impacts “breach,” the court did not err in concluding that Sherwin-Williams had a “legally recog- nized duty” to the plaintiffs. Thomas altered the causation standard, but it did not modify the duty of ordinary care. See id. at 563 (noting that Thomas’s burden was relaxed “only with respect to establishing … the specific type of white lead carbonate [he] ingested”). In Wisconsin, “one has a duty to exercise ordinary care under the circumstances.” Hoida, Inc. v. M & I Midstate Bank, 717 N.W.2d 17, 28 (Wis. 2006). “Ordinary care involves the concept of foreseeability, in that a reasonable person exercising ordinary care would have foreseen injury as a consequence of his act.” Id. at 29. “Whether a duty exists under the circumstances, and the scope of any such duty, are questions of law ….” Brenner v. Amerisure Mut. Ins. Co., 893 N.W.2d 193, 198 (Wis. 2017). Here, the court ruled that “the defendants owed a duty to exercise ordinary care in manufacturing and marketing” white lead carbonate for residential application because the risks it posed to children were foreseeable. Contrary to Sher- win-Williams’s arguments, the court did not predicate its rul- ing on a general duty to the world at large. Sherwin-Williams is also wrong to suggest that the court’s finding of a duty of ordinary care conflicted with its finding that Sherwin-Wil- liams had no duty to warn the plaintiffs. The duty to warn is just one potential manifestation of the duty of ordinary care. See Wis. JI—Civil 3200; see also Morden, 611 N.W.2d at 673–75. In some circumstances, the duty of ordinary care requires a 40 Nos. 20-1774 et al.

warning. Here, the district court found that Sherwin-Williams had no duty to warn for purposes of the plaintiffs’ negligence claims. Even so, it correctly recognized that Sherwin-Williams might still have a duty to take other actions to prevent fore- seeable harms. (Again, a separate question is whether Sher- win-Williams breached that duty. See Morden, 611 N.W.2d at 675–76.) Nor, if we assume contrary to fact that the negligence claim should have gone to the jury, did the court err in in- structing the jury on causation. “We review jury instructions de novo to determine whether, taken as a whole, they cor- rectly and completely informed the jury of the applicable law.” Huff v. Sheahan, 493 F.3d 893, 899 (7th Cir. 2007). The court instructed the jury to decide whether the white lead car- bonate that the plaintiffs ingested caused their injuries. Sher- win-Williams contends that the proper question was whether its negligence caused the injuries. The court’s jury instruction correctly stated the law. The second element of a negligence claim under the risk-contribu- tion theory requires the plaintiff to prove “[t]hat the white lead carbonate caused his injuries.” Thomas, 701 N.W.2d at 564. That is precisely what the court here told the jury to de- cide. Granted, Thomas’s articulation of the causation element of a negligence claim conflicts with ordinary negligence law, which requires a causal link between a defendant’s negligence and the plaintiff’s injury. Fondell v. Lucky Stores, Inc., 270 N.W.2d 205, 209 (Wis. 1978). But the whole point of Thomas was to modify ordinary causation principles. We cannot ac- cept Sherwin-Williams’s suggestion that Thomas was simply sloppy with its language. Thomas was setting forth the ele- ments of a novel cause of action for white lead carbonate Nos. 20-1774 et al. 41

plaintiffs. We find it hard to believe that the court was sloppy in carrying out that task. Indeed, we know that Thomas’s state- ments about causation were not an accident because one of the dissents criticized the very language about which Sher- win-Williams now complains. Thomas, 701 N.W.2d at 591 (Prosser, J., dissenting). The district court lifted its causation instruction straight from Thomas. That was not error. C. Strict Liability Sherwin-Williams and Armstrong both challenge the strict liability verdict. They say (1) the district court erred in ruling that they could have a duty to warn for purposes of strict lia- bility but not negligence; and (2) that the plaintiffs failed to prove that any failure to warn caused their injuries. The dis- trict court rejected the first argument at summary judgment and again when denying the defendants’ motion for judg- ment as a matter of law. Our review is therefore de novo. Turubchuk, 958 F.3d at 548; see also Chemetall, 320 F.3d at 719. The court rejected the second argument when denying the de- fendants’ motion for a new trial. We review that ruling for abuse of discretion. Turubchuk, 958 F.3d at 548. Thomas set forth the elements of a prima facie case of strict liability in white lead carbonate cases. To prevail at trial, the plaintiffs had to prove: (1) That the white lead carbonate was defective when it left the possession or control of the pigment manu- facturers; (2) That it was unreasonably dangerous to the user or consumer; (3) That the defect was a cause of [the plaintiffs’] inju- ries or damages; 42 Nos. 20-1774 et al.

(4) That the pigment manufacturer engaged in the business of producing or marketing white lead car- bonate or, put negatively, that this is not an isolated or infrequent transaction not related to the principal busi- ness of the pigment manufacturer; and, (5) That the product was one which the company ex- pected to reach the user or consumer without substan- tial change in the condition it was when sold. Thomas, 701 N.W.2d at 564. The defendants’ challenge focuses on the first and third elements: defect and causation. 1. Product Defect We begin with product defect. Unlike their negligence claims, the plaintiffs predicated their strict liability claims on a product defect—failure to warn. Although the district court had ruled at summary judgment that the defendants had no duty to warn the plaintiffs for purposes of the negligence claim, it distinguished the duty to warn in the strict liability context. Whereas negligence required a duty to warn the plaintiffs, strict liability required a duty to warn the “ordinary users and consumers” who purchased or used the defend- ants’ products when the defendants sold them—decades be- fore they reached the plaintiffs. The court found a genuine is- sue of material fact as to whether these “ordinary users or consumers” would have needed a warning about the dangers of lead-based paint, given that the dangers were less widely known in the early to mid-twentieth century. A properly designed and manufactured product may still be defective if “an intended use of the product is dangerous, but the manufacturer did not provide sufficient warning or instruction.” Godoy, 768 N.W.2d at 684. Importantly, warnings Nos. 20-1774 et al. 43

are necessary only when a product’s dangers are hidden. In a negligence action, a manufacturer is not liable unless it “has no reason to believe that those for whose use the chattel is supplied will realize its dangerous condition.” Strasser v. Transtech Mobile Fleet Serv., Inc., 613 N.W.2d 142, 155 (Wis. 2000) (quoting Restatement (Second) of Torts § 388 (1965)). Similarly, in a strict liability action, “manufacturers have an obligation to warn consumers about the hidden dangers of their products.” Godoy, 768 N.W.2d at 685 n.15 (emphasis added) (citing Restatement (Second) of Torts § 402A cmts. h, i). The Wisconsin Supreme Court has not addressed the dif- ference, if any, between the duty to warn for negligence and strict liability claims, but we and other courts applying Wis- consin law have treated them as materially identical. See, e.g., Flaminio, 733 F.2d at 467 (holding that the difference between negligent and strict liability duty-to-warn claims, “if any,” was “so small” that the district court’s failure to give separate instructions on them was harmless); Gracyalny v. Westinghouse Elec. Corp., 723 F.2d 1311, 1317 n.11 (7th Cir. 1983) (“[I]n duty to warn cases, emphasis upon the nature and scope of the warning has led to a convergence of the functional identities of strict products liability and negligence.”); Mohr v. St. Paul Fire & Marine Ins. Co., 674 N.W.2d 576, 590 n.10 (Wis. Ct. App. 2004) (“[T]he proof requirements for an inadequate warning on a strict product liability claim are the same as for breach of the duty to warn on a negligence claim.”); Tanner v. Shoupe, 596 N.W.2d 805, 811 n.3 (Wis. Ct. App. 1999); Lemmermann v. Blue Cross Blue Shield of Wis., 713 F. Supp. 2d 791, 810 (E.D. Wis. 2010); Nigh v. Dow Chem. Co., 634 F. Supp. 1513, 1517 (W.D. Wis. 1986). 44 Nos. 20-1774 et al.

It makes sense that the duty to warn would be the same for both types of claims because failure to warn is a type of product defect and defect is a shared requirement for negli- gence and strict liability claims. Godoy, 768 N.W.2d at 681 n.7; Morden, 611 N.W.2d at 673. If there is any substantive differ- ence between negligent and strict liability failure-to-warn claims, it likely stems from the second requirement of a strict liability claim: that the defect render the product unreasona- bly dangerous. See Morden, 611 N.W.2d at 673 (“In a negli- gence action, by contrast, it is not necessary to show that the condition of the product reached the level of unreasonable dangerousness.”); Sharp, 595 N.W.2d at 388. The district court rooted its divergent approach in Thomas, which requires, as the first element of a strict liability claim, that the “white lead carbonate was defective when it left the possession or control of the pigment manufacturers.” 701 N.W.2d at 564 (emphasis added). From this language, the court in- ferred that the duty to warn in strict liability hinges on whether the dangers of white lead carbonate were hidden to the consumers of residential paint products who existed when the defendants released their products into the market. We grant that, under Thomas, the absence of a warning must have existed when the white lead carbonate left the pos- session or control of the pigment manufacturer. Still, the ne- cessity of a warning should depend on what the ultimate con- sumer (i.e., the plaintiffs or their caregivers) knew, rather than what consumers in general knew at the time the manufacturer released the product into the market. See Restatement (Sec- ond) of Torts § 402A cmt. g (“The rule stated in this Section applies only where the product is, at the time it leaves the seller’s hands, in a condition not contemplated by the ultimate Nos. 20-1774 et al. 45

consumer, which will be unreasonably dangerous to him.”) (emphasis added); Morden, 611 N.W.2d at 673 (“It is sufficient [in strict liability] for the plaintiff to show that the product reached him in a dangerously defective condition.” (emphasis added) (quoting Greiten, 235 N.W.2d at 684) (controlling opin- ion of Heffernan, J.))). The purpose behind the requirement that a product be defective when it leaves the hands of a man- ufacturer is to protect the manufacturer from liability for later-arising defects: “The seller is not liable when he delivers the product in a safe condition, and subsequent mishandling or other causes make it harmful by the time it is consumed.” Restatement (Second) of Torts § 402A cmt. g. It is not to un- tether the duty to warn from the ultimate consumer. The focus must always remain on the ultimate consumer. See id. In shifting the focus away from the plaintiffs and toward “ordinary consumers” living decades earlier, the district court seems to have relied on the “consumer-contemplation” test. See Restatement (Second) of Torts § 402A cmt. i (“The article sold must be dangerous to an extent beyond that which would be contemplated by the ordinary consumer who pur- chases it, with the ordinary knowledge common to the com- munity as to its characteristics.”); Green, 629 N.W.2d at 739. Even if that test was relevant to duty to warn, it focuses on ordinary consumers in the plaintiffs’ situation. See Insolia, 216 F.3d at 600 (“[W]hat the ordinary consumer contemplated about the dangers of smoking should be evaluated at the time the plaintiffs began smoking.”). If any “ordinary consumers” were relevant, it was those who consumed residential paint when the plaintiffs (or their caregivers) did. It was not the broad and amorphous category of individuals who consumed residential paint at any time when the defendants were 46 Nos. 20-1774 et al.

manufacturing white lead carbonate—i.e., during large chunks of the twentieth century. For these reasons, the court legally erred in finding that the defendants had a duty to warn for purposes of strict lia- bility after ruling at summary judgment that they had no duty to warn the plaintiffs on their negligence claims. The plaintiffs have not appealed the court’s ruling that the defendants had no duty to warn for purposes of the negligence claims. This ruling compels judgment as a matter of law for Sherwin-Wil- liams and Armstrong on the strict liability claims. 2. Causation We turn next to causation. Before trial, the court ruled that each plaintiff had to prove “that the presence of inadequately- warned-about [white lead carbonate] on the market during the existence of the residence where the plaintiff ingested [white lead carbonate] was a substantial factor in causing the plaintiff’s injury.” It acknowledged that proving such a “ten- uous connection” would be “a tall order.” But it rejected the defendants’ argument that the plaintiffs had to present expert testimony to prove causation; in the court’s view, inferring causation did “not require specialized technical, scientific or medical knowledge,” but could “be made on the basis of com- mon understandings of human behavior.” Consistent with this ruling, the court instructed the jury to decide whether the plaintiffs had proven that “the allegedly defective warnings were a cause of the plaintiffs’ injuries.” To answer “yes,” the jury had to find that “the sale of white lead carbonate with inadequate warnings was a substantial factor in causing the plaintiffs’ injury.” For example, the jury could “consider whether the persons responsible for selecting and Nos. 20-1774 et al. 47

applying paint in the plaintiffs’ homes throughout the build- ings’ existence would have heeded warnings about the risk of childhood lead exposure if such warnings had been issued.” It turns out that this causation standard was indeed a “tall order” for the plaintiffs. It is essentially undisputed that the plaintiffs presented no evidence at trial that the supposed de- fect—failure to warn—caused their injuries. In closing argu- ments, the plaintiffs appealed to the jurors’ “common sense,” arguing that “moms and dads back in the ‘20s and ‘30s and ‘40s and ‘50s” would never have purchased paint containing white lead carbonate if they had known the risks it posed to their children. But the only evidence they relied on was testi- mony that consumers in the first half of the twentieth century were unaware of the risks of white lead carbonate. They in- troduced no evidence of whether adequate warnings would have altered consumer behavior—much less prevented their injuries. The defendants brought this failure of proof to the district court’s attention in their motions for a new trial. The plaintiffs did not respond to these arguments for a new trial. Even so, the court denied the defendants’ motions, and in doing so it departed from its earlier theory of causation. Implicitly ac- knowledging the plaintiff’s total failure of proof on this ele- ment, the court explained that the plaintiffs did not have to show that “appropriate warnings would have changed con- sumer behavior;” rather, they only had to show “that the dan- gers posed by [white lead carbonate] were outside the con- templation of the ordinary consumer,” at which point it be- came the defendants’ burden to prove that their warnings “were sufficient to render safe [the] otherwise dangerously 48 Nos. 20-1774 et al.

defective product.” On appeal, the plaintiffs have again failed to respond to the defendants’ causation arguments. The court was correct the first time: the plaintiffs had to prove that the product defect—i.e., failure to warn—caused their injuries. See Thomas, 701 N.W.2d at 564 (requiring proof “[t]hat the defect was a cause of Thomas’s injuries or dam- ages”); accord Godoy, 768 N.W.2d at 681 n.7. The court’s post- trial ruling improperly collapsed the second and third ele- ments of a strict liability claim and shifted the burden to the defendants. In Green, the Wisconsin Supreme Court advised that a defendant could show that its product was not “unrea- sonably dangerous” by proving “that the product include[d] a warning or directions that effectively alert[ed] the ordinary consumer” to the product’s dangers. 629 N.W.2d at 754. This inquiry, which goes to the second element of a strict liability claim (unreasonable danger), is separate from the third ele- ment (causation). See Thomas, 701 N.W.2d at 564; Godoy, 768 N.W.2d at 681 n.7. A plaintiff must always prove causation. See Thomas, 701 N.W.2d at 564. Because the plaintiffs do not dispute that they failed to present evidence of causation at trial, we need not address what type of evidence might suffice to prove causation. More- over, because the defendants do not directly challenge the court’s pretrial ruling on causation or its jury instructions, we need not address whether the court correctly stated the rele- vant legal principles (before departing from them post-trial). Normally, the plaintiffs’ failure to prove causation would require at least a new trial. In this case, however, any such relief would be redundant because Sherwin-Williams and Armstrong are already entitled to judgment as a matter of law Nos. 20-1774 et al. 49

on the strict liability claims, given the absence of a product defect. *** We reject the defendants’ other attacks on the strict liabil- ity verdict. The defendants argue first that the “ordinary con- sumers” of white lead carbonate were master painters and paint manufacturers, who required no warning as to the dan- gers of white lead carbonate. The defendants’ argument ap- pears to rest on one of two mistaken premises: (1) that prod- ucts liability claims require “privity of contract,” or (2) that component parts suppliers cannot be liable to the ultimate consumer. The Wisconsin Supreme Court has flatly rejected both arguments. Godoy, 768 N.W.2d at 681 (noting that the Second Restatement discarded the privity-of-contract re- quirement); id. at 688 (“When component manufacturers in- troduce defective components into the stream of commerce, they may be held liable for resulting injuries under the partic- ular circumstances of the case.”). Moving to the fifth element of the strict liability claim, the defendants argue that their white lead carbonate could not have reached the plaintiffs without undergoing a “substantial change in condition”—i.e., integration into paint, application of paint to the plaintiffs’ homes, and a lack of maintenance allowing the paint to deteriorate and become ingestible. Thomas, 701 N.W.2d at 564. “[T]o succeed under the substan- tial change defense, the change must be both substantial and material.” Godoy, 768 N.W.2d at 687. “The purpose of this re- quirement is to protect a manufacturer from liability when the dangerously defective aspect of the product was altered or in- troduced after the product left the manufacturer’s control.” Id. This is a “fact-intensive inquiry.” Id. 50 Nos. 20-1774 et al.

We agree with the district court that any changes to the defendants’ white lead carbonate were “not material, since the [white lead carbonate] was already toxic and in powder form before it underwent these changes.” White lead car- bonate is the singular source of lead poisoning in paint that contains it. As Thomas said, “the inherent dangerousness of the white lead carbonate pigment existed the moment the Pig- ment Manufacturers created it.” 701 N.W.2d at 563. If any- thing, mixing it into paint “diluted the white lead carbonate’s toxicity.” Id. The deterioration of the paint may have helped facilitate the plaintiffs’ ingestion of the white lead carbonate, but it did not introduce or alter its toxicity. The jury was on firm ground in finding no substantial change in condition. D. Other Issues Sherwin-Williams and Armstrong raise several other chal- lenges. Ordinarily we would avoid reaching these issues if possible. We are mindful, however, that there are a slew of similar cases pending at the district court and moving toward trial. As such, we will address some of these other issues in hopes of providing clarity for future trials. 1. Expert Testimony Sherwin-Williams and Armstrong claim that the plaintiffs failed to prove the existence, cause, and extent of their injuries through admissible evidence. They challenge the district court’s admission of Dr. Trope’s and Dr. Besunder’s testi- mony on these issues. “We review de novo whether a district court properly followed the framework for determining the admissibility of expert testimony.” Schultz v. Akzo Nobel Paints, LLC, 721 F.3d 426, 430–31 (7th Cir. 2013); see also Fed. R. Evid. 702; Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579 Nos. 20-1774 et al. 51

(1993). If it did, “we review its decision to admit or exclude expert testimony only for an abuse of discretion.” Schultz, 721 F.3d at 431. We review de novo the court’s denial of judgment as a matter of law on this ground. Turubchuk, 958 F.3d at 548. Federal Rule of Evidence 702 permits a qualified expert witness to offer an opinion if: (a) the expert’s scientific, technical, or other specialized knowledge will help the trier of fact to understand the evidence or to determine a fact in issue; (b) the testimony is based on sufficient facts or data; (c) the testimony is the product of reliable principles and methods; and (d) the expert has reliably applied the principles and methods to the facts of the case. Courts can examine the reliability of an expert’s principles and methods by looking at factors such as “(1) whether the scientific theory or technique can be (and has been) tested; (2) whether the theory or technique has been subjected to peer review and publication; (3) whether a particular technique has a known potential rate of error; and (4) whether the theory or technique is generally accepted in the relevant scientific community.” Schultz, 721 F.3d at 431 (citing Daubert, 509 U.S. at 593–94). This list is not exhaustive. Daubert, 509 U.S. at 593. Although Rule 702 “places the judge in the role of gate- keeper for expert testimony, the key to the gate is not the ul- timate correctness of the expert’s conclusions” but rather “the soundness and care with which the expert arrived at her opin- ion.” Schultz, 721 F.3d at 431. “So long as the principles and methodology reflect reliable scientific practice, ‘[v]igorous 52 Nos. 20-1774 et al.

cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible ev- idence.’” Id. (quoting Daubert, 509 U.S. at 596). We begin with Dr. Trope. The defendants claim that her method for determining that each of the plaintiffs had brain damage—administering a neuropsychological evaluation— was unreliable, primarily because it failed to exclude alterna- tive causes. As Dr. Trope testified, however, she did not need to exclude alternative causes to testify to the fact of brain dam- age. And the defendants’ other attacks on the reliability of neuropsychological evaluations are unpersuasive. Dr. Trope and Dr. Besunder both testified that neuropsychological eval- uations are the standard, well-accepted method for ascertain- ing whether an individual has brain damage. (The defend- ants’ own expert relied on a similar methodology to testify that the plaintiffs did not have brain damage.) Many of the defendants’ arguments wrongly equate neuropsychological evaluations with basic aptitude tests. As Dr. Trope explained, a neuropsychological evaluation tests how different parts of the brain function compared to one other—not how they function on the whole. Large discrepancies indicate brain damage, according to both Dr. Trope and Dr. Besunder. We see no abuse of discretion in the court’s decision to allow Dr. Trope to testify that the plaintiffs had brain damage. The de- fendants were free to cross-examine Dr. Trope about any weaknesses in her methodology. Schultz, 721 F.3d at 431. Dr. Trope’s testimony that lead poisoning caused the plaintiffs’ brain damage presents a more difficult question be- cause the plaintiffs did not offer her to prove causation; she testified to it only on cross-examination, over the plaintiffs’ Nos. 20-1774 et al. 53

objections. Moreover, the defendants suggest that Dr. Trope, who is not a medical doctor, was unqualified to diagnose the plaintiffs’ brain injuries. We need not resolve this issue. Any error in the admission of Dr. Trope’s causation testimony was harmless because the court was within its discretion to admit Dr. Besunder’s causa- tion testimony. Dr. Besunder testified that childhood lead poi- soning was the source of the plaintiffs’ injuries. He reached that conclusion by comparing their brain functioning—as shown in their neuropsychological evaluations—to docu- mented patterns of brain functioning in other individuals who had suffered childhood lead poisoning. He testified that this type of comparison is a well-accepted method of ascer- taining the cause of brain damage, and that he uses it in his own practice. He also testified that he reviewed the plaintiffs’ medical records to determine whether there were potential al- ternative causes of their brain damage. See Schultz, 721 F.3d at 434 (noting that “a reliable expert should consider alternative causes” but need not “rule out every alternative cause”); My- ers v. Illinois Cent. R.R. Co., 629 F.3d 639, 644–45 (7th Cir. 2010); see also Thomas, 701 N.W.2d at 563 (reasoning that the poten- tial for alternative causes of brain damage was an issue for the jury). To the extent that Dr. Besunder could not rule out every alternative cause, the defendants could—and did—try to bring that out through cross-examination. Moreover, the plaintiffs had to prove that lead poisoning was a substantial factor leading to their lead poisoning—not the sole cause. Schultz, 721 F.3d at 433 (applying Wisconsin law). Dr. Besunder’s testimony about the extent of the plaintiffs’ injuries is a different story. Dr. Besunder’s IQ-loss testimony was a critical component of the plaintiffs’ cases because it 54 Nos. 20-1774 et al.

attempted to measure the extent of their injuries—and, thus, their damages. Yet the district court performed essentially no analysis of whether Dr. Besunder’s methodology—using gen- eral epidemiological studies to quantify the plaintiffs’ indi- vidual IQ losses—was reliable. The defendants argued, with- out apparent contradiction, that Dr. Besunder’s methodology was unprecedented. They also argued that it was unreliable because Dr. Besunder did not personally evaluate the plain- tiffs, nor did he have any evidence of their baseline IQs—even though the plaintiffs’ other experts testified that they could not quantify the plaintiffs’ IQ losses without knowing the plaintiffs’ parents’ IQ scores. The court dismissed these objec- tions as issues for the jury. But trial judges are the “gate- keeper[s] for expert testimony.” Schultz, 721 F.3d at 431. Be- fore allowing an expert to testify before the jury, the judge must ensure that the expert’s “principles and methodology reflect reliable scientific practice.” Id. The district court failed to perform its gatekeeping function and ensure that Dr. Be- sunder’s IQ-loss testimony rested on a reliable methodology, so the court abused its discretion in admitting it. The district court’s admission of Dr. Besunder’s IQ-loss testimony requires a new trial because there is “a significant chance” that the jury’s identical $2 million damages awards rested on Dr. Besunder’s testimony about the extent of the plaintiffs’ injuries. Smith v. Hunt, 707 F.3d 803, 808 (7th Cir. 2013). (After all, the court remitted Burton’s damages award to $800,000 post-trial because Dr. Besunder admitted that, of the 10 IQ points that Burton lost due to lead poisoning, he lost six of them before he moved to the home that he focused on at trial.) As a practical matter, this relief only impacts Arm- strong’s liability on the negligence claim. For reasons dis- cussed above, Sherwin-Williams is entitled to judgment as a Nos. 20-1774 et al. 55

matter of law on both claims and Armstrong is entitled to judgment as a matter of law on the strict liability claim. 2. Bifurcation Next, Sherwin-Williams and Armstrong contend that the district court erred in bifurcating the trial into a liability phase and an apportionment phase and then excluding evidence about National Lead—the dominant producer of white lead carbonate pigment in the Milwaukee market—from the liabil- ity phase. They claim that the district court’s rulings ham- pered their defenses and gave the jury the mistaken impres- sion that at least one of the defendants on the verdict form must have been the responsible party. We review the court’s decision to bifurcate the trial for abuse of discretion. Volkman v. Ryker, 736 F.3d 1084, 1089 (7th Cir. 2013). We use the same standard to review whether the court erred in excluding evidence of National Lead or deny- ing a new trial on this basis. Turubchuk, 958 F.3d at 548–49. We see no abuse of discretion. Absent some clear preju- dice, we will not fault the court for attempting to inject some semblance of order into this complex case—especially given district courts’ “inherent authority to manage the course of trials.” Luce v. United States, 469 U.S. 38, 41 n.4 (1984). The court’s structure of the trial harmonized with Collins’s design to (1) create “a pool of defendants which can reasonably be assumed ‘could have caused the plaintiff’s injuries’” and then (2) apportion liability among those “defendants who cannot exculpate themselves” through comparative negligence. Thomas, 701 N.W.2d at 565 (quoting Collins, 342 N.W.2d at 52). In this respect we find it relevant that Collins expressly de- clined to adopt a “market-share” theory of liability, instead 56 Nos. 20-1774 et al.

holding that market share was “a relevant factor in apportion- ing liability among defendants.” Collins, 342 N.W.2d at 48–49. Under Collins, the defendants had every right to point the fin- ger at National Lead during the apportionment phase of trial. They chose instead to settle and assign National Lead 12.5% of the damages. Their informed decision to settle prevents them from now complaining about not having been able to introduce evidence of National Lead’s market domination. To the extent that evidence of National Lead had any rel- evance in the liability phase—which is far from clear—the dis- trict court did not abuse its discretion in concluding that the likelihood of confusion and prejudice substantially out- weighed any such probative value. The defendants strain to connect evidence of National Lead to their liability defenses. As the district court pointed out, it is hard to see how evidence of National Lead’s market domination had any bearing on whether another manufacturers’ white lead carbonate could have reached the plaintiffs, which is the liability inquiry un- der Collins. Id. at 52. Meanwhile, it would have been all too easy for the defendants to point the finger at National Lead and rely on the unstated implication that National Lead was probably the responsible party. Such a maneuver would have been inconsistent with the risk-contribution theory that Col- lins and Thomas adopted. Finally, there is no basis for the de- fendants’ assertion that the jury was led to believe that one of the defendants who went to trial must have produced the white lead carbonate that injured the plaintiffs. In fact, the court dismissed American Cyanamid at the close of evidence and then instructed the jury not to consider any claims against American Cyanamid and not to speculate as to why it was no longer a defendant. Nos. 20-1774 et al. 57

3. Public Policy Forging along, Sherwin-Williams and Armstrong contend that Wisconsin’s judicial public policy factors require over- turning the jury verdicts against them. They claim that Wis- consin public policy does not tolerate holding them liable for injuries that they likely did not cause, which occurred decades after they stopped making white lead carbonate. They add that their liability is vastly out of proportion to the culpability of producing or marketing white lead carbonate at a time when it was widely used in residential paints. We review de novo the court’s denial of judgment not- withstanding the verdict. Turubchuk, 958 F.3d at 548; see also Fandrey, 680 N.W.2d at 350 (whether the public policy factors preclude liability is a question of law). Wisconsin’s public policy factors are judicial policy con- siderations that function like a proximate cause analysis. See Fandrey, 680 N.W.2d at 350–51, 351 n.7. “[W]hen a court pre- cludes liability based on public policy factors, it is essentially concluding that despite the existence of cause-in-fact, the cause of the plaintiff’s injuries is not legally sufficient to allow recovery.” Id. at 353. Overriding a jury verdict based on public policy considerations is “infrequent” and requires “unusual and extreme considerations.” Roehl Transp., Inc. v. Liberty Mut. Ins. Co., 784 N.W.2d 542, 568 (Wis. 2010) (internal quotation marks and citations omitted). The six public policy factors are: (1) is the injury too remote from the defendant’s conduct? (2) is the plaintiff’s injury disproportionate to the defendant’s culpability? (3) is it too extraordinary, in hindsight, that liabil- ity would attach here? (4) is the potential liability too burden- some for the defendant? (5) will imposing liability open the way to fraudulent claims? and (6) is there a sensible or just 58 Nos. 20-1774 et al.

stopping point to the theory of liability? See Fandrey, 680 N.W.2d at 348 n.1. If the defendants’ public policy arguments look familiar, it is because they closely resemble the dissents in Thomas. At their core, the defendants’ public policy arguments are argu- ments against Thomas itself. One of the dissents in Thomas sim- ilarly argued that the public policy factors precluded liability. 701 N.W.2d at 596–97 (Prosser, J., dissenting). The majority did not directly engage with that argument, id. at 565 n.54, but its reasoning for extending the risk-contribution theory was anchored in countervailing policy considerations, and many of its statements were directly responsive to the dissent’s pub- lic policy arguments. See, e.g., id. at 562 (“[T]he Pigment Man- ufacturers’ argument must be put into perspective: they are essentially arguing that their negligent conduct should be ex- cused because they got away with it for too long.”); see also Collins, 342 N.W.2d at 52 (accepting the possibility that inno- cent defendants could be found liable as “the price the de- fendants, and perhaps ultimately society, must pay to provide the plaintiff an adequate remedy under the law”). We cannot accept that the Wisconsin Supreme Court would create a cause of action based on judicial policy considerations only to later hold that the same cause of action would violate judicial policy considerations. The public policy factors operate as a check on liability in rare cases. We will not hold, in essence, that they categorically preclude liability under Thomas. Nor does Wis. Stat. § 895.046 affect our conclusion. Wis- consin’s public policy factors are judicial policy considera- tions whose role stands in “stark contrast” to the legislature’s declarations of public policy. Fandrey, 680 N.W.2d at 354. That Nos. 20-1774 et al. 59

is why the public policy factors may preclude liability even where a statute authorizes it. Id. at 350. 4. Successor Liability Armstrong also challenges the district court’s ruling at summary judgment that Armstrong was MacGregor’s succes- sor-in-interest. Armstrong contends that the relevant asset purchase agreement is ambiguous as to whether Armstrong’s admitted predecessor-in-interest assumed MacGregor’s fu- ture legal liabilities, such as damages arising from the plain- tiffs’ lawsuits. And because the contract is ambiguous, Arm- strong says its meaning was an issue for the jury. Our review is de novo. Turubchuk, 958 F.3d at 548; see also Chemetall, 320 F.3d at 719. Neither party raised choice of law issues below, so we ap- ply the law of the forum—Wisconsin. Camp v. TNT Logistics Corp., 553 F.3d 502, 505 (7th Cir. 2009) (citing Wood v. Mid-Val- ley Inc., 942 F.2d 425, 426 (7th Cir. 1991)). Under Wisconsin law, “[w]hen the terms of a contract are clear and unambigu- ous, we construe the contract’s language according to its lit- eral meaning.” Ash Park, LLC v. Alexander & Bishop, Ltd., 866 N.W.2d 679, 685 (Wis. 2015). “When the terms of a contract are ambiguous, however, evidence extrinsic to the contract it- self may be used to determine the parties’ intent, and any re- maining ambiguities will be construed against the drafter.” Id. “A contract provision is ambiguous if it is fairly susceptible of more than one construction.” Id. (quoting Mgmt. Computer Servs., Inc. v. Hawkins, Ash, Baptie & Co., 557 N.W.2d 67, 75 (Wis. 1996)). In relevant part, the asset purchase agreement provides that Armstrong’s predecessor will assume “the liabilities and 60 Nos. 20-1774 et al.

obligations” of the sellers, including “[o]bligations with re- spect to litigation and claims against Sellers other than litiga- tion and claims involving or arising out of liabilities and obli- gations expressly not assumed.” The agreement also includes a catchall providing that, “[e]xcept for the obligations and li- abilities” excluded elsewhere, Armstrong’s predecessor will “assume all other obligations of sellers, whether accrued, con- tingent or future, based on, related to or arising out of events or occurrences prior to the closing, whether known or un- known and regardless of when arising or asserted.” Armstrong maintains that the terms “obligations” and “li- abilities” are ambiguous, such that its predecessor’s assump- tion of “obligations with respect to litigation and claims” does not necessarily include all future legal “liabilities,” including damages arising from a lawsuit based on unforeseen events. We disagree. To be sure, the contract is not a model of clear drafting, and the terms “obligations” and “liabilities” are not used consistently throughout. But, at least in the provisions quoted above, “obligations” must encompass “liabilities” be- cause “obligations” is used as an umbrella term from which other “liabilities and obligations” are excluded. Moreover, the plain terms of the exceptionally broad catchall easily encom- pass the damages liability in this case (and Armstrong cannot point to any other provision that specifically excludes it). Even if the contract were ambiguous, moreover, Arm- strong has not demonstrated a genuine issue of material fact as to its meaning. When courts are stumped about the mean- ing of contracts, they do not simply hand the ambiguous con- tracts to juries. Rather, if a court decides that a contract is am- biguous, the parties may introduce extrinsic evidence of in- tent, and the jury resolves the fact question of what the parties Nos. 20-1774 et al. 61

intended, using on the extrinsic evidence. See Town Bank v. City Real Estate Dev., LLC, 793 N.W.2d 476, 483 (Wis. 2010). Armstrong does not represent that it has any extrinsic evi- dence of the parties’ intent, so it has not demonstrated a jury issue. 5. First Amendment Finally, Sherwin-Williams argues that the district court vi- olated its First Amendment rights by allowing the plaintiffs to introduce evidence of its product advertisements and associ- ations with industry groups. The district court deemed this evidence relevant to certain aspects of the plaintiffs’ claims, including Sherwin-Williams’s knowledge of the hazards of lead-based paint and its presence in the Milwaukee market. And because Sherwin-Williams’s liability rested on its pro- duction and sale of white lead carbonate—rather than its pro- tected activities—the court saw no constitutional problem with admitting the evidence. We review the court’s First Amendment analysis de novo, while reviewing its decisions to admit the evidence and deny a new trial for abuse of dis- cretion. See Turubchuk, 958 F.3d at 548. Sherwin-Williams is profoundly mistaken about the role of the First Amendment in evidentiary rulings. The First Amendment does not bar the admission of any and all evi- dence that falls within its protection. Just last month, we held that such a view of the First Amendment “utterly misunder- stands the burdens of production and persuasion in litiga- tion.” Gonzales v. Madigan, 990 F.3d 561, 564 (7th Cir. 2021). The First Amendment steps in only when the protected activ- ity itself is the basis for liability. See NAACP v. Claiborne Hard- ware Co., 458 U.S. 886, 918–19 (1982) (“The First Amendment … restricts the ability of the State to impose liability on an 62 Nos. 20-1774 et al.

individual solely because of his association with another.”) (emphasis added). Here, the plaintiffs introduced evidence of Sherwin-Williams’s advertisements and associations to prove elements of their claims. They did not ask the jury to find Sherwin-Williams liable for engaging in these protected activ- ities. Cf. Snyder v. Phelps, 562 U.S. 443, 461 (2011) (overturning a jury verdict finding the defendants liable for picketing on matters of public concern). Thus, the district court did not abuse its discretion in admitting this evidence. *** We have considered the defendants’ remaining arguments and determined that none merits discussion. III. Conclusion For these reasons, we REVERSE the judgments and REMAND for further proceedings consistent with our opin- ion and the following instructions. Sherwin-Williams is enti- tled to judgment as a matter of law on both claims that went to trial. Armstrong is entitled to judgment as a matter of law on the strict liability claims and a new trial on the negligence claims. DuPont is entitled to a new trial on both claims. We close by commending Judge Adelman for his thought- ful dedication to these complex cases. Circuit Rule 36 supplies a default rule of reassignment when cases are remanded for new trials. Given Judge Adelman’s enormous investments of time and effort in these cases, we believe that the interests of judicial efficiency favor retaining Judge Adelman as the trial judge in these cases. We therefore direct that Rule 36 shall not apply on remand. See Cir. R. 36 (providing for reassignment “unless the remand order directs … that the same judge retry the case”).