This case is another in the series of personal injury, diversity cases involving asbestos exposure of naval shipyard employees. The case is confined to the question of manufacturer liability for plaintiffs exposure during and just after World War II. It is a case of first impression in the Tenth Circuit and requires our review of a jury trial in the federal district of Kansas, tried by stipulation under Nebraska law.
The appellant asbestos manufacturers— The Celotex Corporation (“Celotex”), Eagle-Picher Industries, Inc. (“Eagle-Picher”), and Fibreboard Corporation (“Fibre-board”) — challenge the denial of a judgment notwithstanding the verdict (“JNOV”), 641 F.Supp. 1429, contending that the evidence of their strict liability or their allegedly negligent failure to test, warn, or otherwise control for asbestos hazards was insufficient to support a jury verdict against them. Alternatively, they allege a number of prejudicial errors in the jury instructions, especially with respect to their state-of-the-art defense and the burden of proving proximate cause under Nebraska law.
After careful consideration of the complexities of this case, we remand for a new trial. The jury instructions erroneously suggest at least two different interpretations of the causation standards, rendering the jury’s verdict against the appellants essentially unreviewable and unreliable. If the jury used traditional Nebraska standards suggested by one part of the instructions, then the evidence is insufficient to support the jury’s finding that certain of the defendants proximately caused the plaintiff’s asbestos-related disease. If the jury’s finding was based on the special *1456burden-shifting rule suggested by other parts of the instructions, we might have been able to uphold it but for the fact that we are unable to determine that the jury applied the correct threshold proof of causal relationship before shifting the burden. We also conclude that the trial court misin-structed the jury with respect to Nebraska’s state-of-the-art defense.
BACKGROUND
Asbestos is a naturally occurring mineral fiber with fireproof and corrosion-resistant properties that have made it valuable as an insulator. Unfortunately, it is also hazardous to one’s health. The fibers are breathable and can cause asbestosis (a scarring of the lungs), mesothelioma (a cancer of various linings of body organs, notably of the lungs or abdomen), and lung cancer. Asbestosis is a type of pulmonary fibrosis produced by infiltration of microscopic asbestos fibers into lung tissue. It is a progressive disease for which there is no cure; the asbestos fibers are essentially indestructible and their scarring effect is cumulative over time. The latency period can be anywhere from five to thirty years, R.Vol. 5 at 217, depending on the intensity and duration of the exposure. Mesothelioma is an incurable cancer that can be caused by as little as an intense two to three month episode of breathing asbestos dust. R.Vol. 6 at 714. The latency period is typically 30 years or more, R.Vol. 5 at 367, but can range from less than twenty to more than fifty years, R.Vol. 6 at 751. Of the three basic kinds of asbestos fibers — amosite, crocidolite, and chrysotile — the straight, solid amosite and crocidolite fibers are less likely to break up in the lungs and more likely to cause mesothelioma than are the curly, hollow chrysotile fibers. Chrysotile is used in textile and friction products, while amosite and crocidolite are used in pipe-covering. R.Vol. 6 at 742.
The plaintiff Don Menne (“Menne”), now deceased,1 had a forty year history of on-the-job exposure to asbestos fibers. As a teenager in the 1930s he worked part time for his father installing and insulating heating and plumbing pipes. For approximately four and a half years from June 1941 through late 1945 or early 1946 he worked at the Puget Sound Naval Shipyard (“Shipyard”) in Bremerton, Washington, first as a pipe-installer trainee for six months, then as a lay-out man for about two years, and then as a pipefitter. As a pipefitter he worked in particularly close proximity to insulators who were installing asbestos products and generating asbestos dust on board ships docked in the Shipyard.
After World War II Menne returned to his native Nebraska and briefly attempted to farm. In late 1946 he returned to the Shipyard and worked there in an undisclosed capacity for approximately a year. For the next thirty years or so he worked in plumbing and heating jobs primarily in private industry, where he had various exposures to the asbestos products of multiple manufacturers. He retired in 1980 and then worked part time until 1985 as a school janitor.
In 1983, at the age of approximately sixty-six, Menne sought medical treatment for increasing shortness of breath, fatigue, and leg aches. Discovery of a large tumor attached to the pleura (lining) of the lung prompted surgical removal of his entire left lung along with portions of the diaphragm and the pericardium (the sac surrounding the heart). Upon examination of the removed tissue, the pathologist diagnosed the tumor as malignant mesothelioma and reported evidence of asbestosis secondary to the mesothelioma. Prior to surgery, however, pulmonary function tests on Menne had not revealed clinical evidence of asbestosis. R.Vol. 5 at 327.
Menne claimed that his mesothelioma2 had been proximately caused by the *1457defendants’ negligent failure to test adequately, warn, or otherwise protect him from the known dangers of asbestos exposure during the period from 1942 to 1948 when his exposure to asbestos dust was its most intense. Alternatively, he claimed that defendants’ failure to issue warnings made their products unreasonably defective under a theory of strict liability.3
The original defendants included the following manufacturers of asbestos products: Celotex, Eagle-Picher, Fibreboard, Keene Corporation, Owens-Illinois, Inc., and Raymark Industries, Inc. Asbestos products of these manufacturers or their predecessor companies had been identified as among those installed by pipe insulators on the ships at the Shipyard during the decade from 1940 to 1950. Not before the court were Johns-Manville, Mundet, Unar-co, and Keasbey & Mattison, whose products also had been identified as used on the ships during the time period in question.4 At trial the jury found Keene and Owens-Illinois not liable.5 The jury returned a two and a half million dollar verdict against the remaining four defendants. After trial and entry of judgment, Raymark settled in the amount of $150,000, thus leaving Celo-tex, Eagle-Picher, and Fibreboard as the appellants in this case.
For the limited purposes of this trial, Celotex and Fibreboard accepted responsibility for the acts of their predecessor companies, Philip-Carey and Plant respectively. Those predecessors (designated as Cel-otex and Fibreboard throughout this opinion) manufactured and supplied the Shipyard with pre-formed asbestos pipe covering and with asbestos “blocks” (large sheets) for insulation of large equipment such as boilers and turbines. R.Vol. 6 at 607, 609. Celotex, along with Eagle-Picher, also supplied the Shipyard with cement products containing asbestos. R.Vol. 6 at 604, 609. No actual evidence (but instead only assumptions) as to the physical properties of the Celotex asbestos block and pipe covering is in the record, but its cement products consisted of both asbestos cements and magnesia cements (15% asbestos). Id. at 609, 629. The Fibreboard and Eagle-Picher products contained only a small percentage of asbestos, id. at 630, in contrast to the products of some of the manufacturers such as Johns-Manville and Unarco, who were not before the court.
At trial Menne presented evidence that he was exposed to asbestos dust as both a lay-out man and a pipe and valve installer (pipefitter) at the Shipyard. His primary exposure came during the years he was pipe fitting. He estimated that 75-80% of his time as a pipefitter was spent on board ships which were either under construction or repair, and that about 80% of his time on the ships was spent working side by side with insulators in stagnant and hazy air that was full of suspended asbestos dust. R.Vol. 7 at 905-06. In other words, he *1458alleges that from three-fifths to nearly two-thirds of his working time as a pipefit-ter was spent breathing air that contained visible asbestos dust. The significance of the dust’s visibility, according to the record, is that when the dust particles are visible they have exceeded what was considered in the 1940s to be the safe threshold level of five million particles per cubic foot. R.Vol. 5 at 234-36.
Menne could not identify any particular asbestos products to which he was exposed during his time at the Shipyard. His key witness, Maurice Lane, however, was able to identify ten different manufacturers whose products were used at the Shipyard during the years in question. Lane was an asbestos worker — a material man and insulator — at the Shipyard from 1940 to 1950. As a material man in the early 1940s he recalled routinely unloading and transferring appellants’ products from several of the six major materials shacks to whatever ships were being serviced from the shacks. As an insulator, he also visited each of the shacks on many occasions. He stated that on any given day several different brands would be in each shack, and he believed that over time all the various asbestos products were used by insulators on all the ships on which he worked. R.Vol. 6 at 590-91, 613-14. Although there was no evidence that Menne and Lane worked on the same ships, and although Lane could not say if or how much of a particular product was used on any given ship at any given time, he did say that he used appellants’ products on every ship on which he worked. Id. at 605-10.
Lane also testified that appellants’ products — asbestos blocks, pipe-covering, and cement — were among those producing asbestos dust when used by the insulators; dust was generated when the blocks and pipe covering were sawed or cut to fit the pipes and equipment, when the wires holding the insulation in place were embedded in the insulation by beating them with mallets, and when asbestos-containing cements to cover the insulation were mixed in the confined areas in which the insulators and pipefitters worked side by side. R.Vol. 6 at 595-96, 605, 612. He noted, however, that generally cement was mixed in the shacks and then carried to the ships. Id. at 559-60, 569. Dust was also created when asbestos cloth was ripped to fit over the thicker pre-formed asbestos covering and when old insulation was torn off damaged ships. Among the defendants, however, only Raymark was identified as a manufacturer of asbestos cloth, and it is not an appellant here. Appellants also could not be identified as the manufacturers of the asbestos products on ships under repair, although Lane said that the old insulation in many cases appeared to be the same as the products that were being installed on the new ships. About twenty-five percent of Menne’s time as a pipefitter was spent working on ships under repair. R.Vol. 7 at 933.
At trial the manufacturers introduced expert opinion testimony, vigorously disputed by plaintiff’s experts, that Menne’s tumor was a hyperplasia (a large benign tumor mass) rather than mesothelioma. Defendants also asserted that: (1) the state-of-the-art evidence for the 1940s showed no epide-miologically significant evidence that asbestos dust caused mesothelioma and therefore no evidence of negligence or strict liability and (2) even if asbestos dust had been known to cause mesothelioma in the 1940s, Menne had not been able to establish that the asbestos products of each defendant caused or substantially contributed to his alleged mesothelioma. Defendants sought a directed verdict based on the insufficiency of the evidence to establish proximate cause, among other things. After denial of the directed verdict, and over defendants’ objections, the trial court submitted jury instructions and special interrogatories that, according to the defendants, either misconstrued Nebraska law or confused the jury with respect to causation and the state-of-the-art defense. In particular, the defendants objected to instructions which allegedly adopted a theory of alternative liability — a theory requiring defendants to prove the absence of causation upon plaintiff’s proof of a threshold connection to the injury. After the jury verdict, the remaining defendants moved for a *1459JNOV or alternatively for a new trial, which motion was also denied.
The appellants contest the denial of their motion, focusing on various errors of law in the jury instructions.6 They primarily challenge the court’s attempted invocation of alternative liability theory or a variant thereof, arguing that a jury instructed to apply traditional Nebraska tort law necessarily would have found insufficient evidence to return a verdict for Menne. In response, Menne insists that the burden of proof on causation never shifted and that the evidence was sufficient under traditional Nebraska standards.
The appellants also argue that the state-of-the-art instruction, under which the jury concluded that negligence or strict liability for failure to warn had been established, did not accurately reflect Nebraska law. Additionally, they assert that the judge mi-sinstructed the jury by announcing both the maximum amount of damages sought by Menne and the public policy justification for strict liability in tort. Finally, they contend that under Nebraska law Menne should not have been allowed to include a loss of consortium claim on behalf of his wife. We discuss each argument in turn.
I. Proximate Cause Under Nebraska Tort Law
A. Traditional Nebraska Tort Standards
At the heart of this case is a need to resolve Nebraska’s proximate cause standards in situations where there are multiple and concurrent causes of harm from asbestos dust. Jury instruction # 137 correctly reflected that Nebraska law incorporates both a “but-for” and a “substantial factor” test of causation.8 Instruction # 14 then informed the jury as to Nebraska’s law of concurrent cause, quoting Nebraska Pattern Jury Instruction (“NJI”) # 3.42 (1969), as follows:
*1460“Where the independent acts or omissions to act of two or more persons combine to proximately cause an injury indivisible in its nature, each such act or omission to act is itself a proximate or concurrent cause, and any one or more of such persons may be held responsible for the entire injury. This is true even though one may have been more negligent than the other. It is no defense that all those who might have been responsible are not made parties to this action.”
R.Yol. 1 at tab 37 (emphasis added). A preliminary question emerging from this instruction is whether Nebraska requires individual causation by each defendant in cases of concurrent cause or whether it provides for a test of group causation to replace individual causation.
Appellants insist that Nebraska law requires each of the multiple defendants to be, first, a but-for and, secondarily, a substantial cause of the single harm. They implicitly interpret the above-italicized phrase — “combine to proximately cause”— to mean simply that there can be more than one actor contributing to the same indivisible harm and that liability attaches to those multiple contributors on the basis of ordinary but-for and substantial cause standards and not mere exposure to defendants’ asbestos products as a group. In other words, it is not the causation that is single and indivisible; it is the resultant harm.
In contrast, Menne asserts that Nebraska’s concurrent cause doctrine currently includes not only a concept of individual causation but of group causation as well. He implicitly interprets the phrase “combine to proximately cause” to mean that individual actors can become one entity for purposes of a plaintiff’s prima facie case of causation, thereby allowing each negligent defendant to be liable even though the collective action of all negligent actors was the only but-for and substantial cause of the injury. In effect, any member of the group can be jointly and severally liable for the harm caused by the whole group, regardless of whether its own contribution to the harm was slight and regardless of whether the harm would have occurred without it.
After reviewing Nebraska case law, we conclude that Menne’s position is an incorrect reading of present Nebraska law. Individual causation has carried over into all the Nebraska concurrent cause cases of which we are aware. And all have involved situations where but for the negligence, in turn, of multiple actors, the injury would not have occurred. In other words, each of the negligent actors was a but-for cause; all were needed to produce the injury. See Lindgren v. City of Gering, 206 Neb. 360, 292 N.W.2d 921 (1980) (negligence of both defendants was a prerequisite to the backup of water and sewage into plaintiff’s basement; if either of the negligent acts had not been committed, the damage would not have occurred). See also Libbey-Owens Ford Glass Co. v. L & M Paper Co., 189 Neb. 792, 205 N.W.2d 523 (1973) (negligence of three defendants concurred to produce fire damage; negligence of any one [or two] of them alone would not have caused the result); Umberger v. Sankey, 151 Neb. 488, 38 N.W.2d 21 (1949) (if two pilots were concurrently negligent, plaintiff could recover from one of them, “though one of them alone might not have caused the injury” (quoting Bergendahl v. Rabeler, 133 Neb. 699, 276 N.W. 673, 674 (1937))); Koehn v. City of Hastings, 114 Neb. 106, 206 N.W. 19, 21 (1925) (“If one suffers damage as the proximate result of the negligence of two others, and the damage would not have occurred but for the negligence of each of such parties, both are liable to the person so injured.”); Schweppe v. Uhl, 97 Neb. 328, 149 N.W. 789, 790 (1914) (injury to driver of horse team was caused by combined actions of several autos, each of which, in procession, noisily passed the team at a dangerous spot, and no one of which alone would have caused the damage). In all these cases and in Nebraska’s automobile injury cases finding concurrent cause, the resultant single injury would not have occurred but for the contribution of each negligent actor. In addition, each has been a substantial contributing factor in the resultant injury. *1461See, e.g., London v. Stewart, 221 Neb. 265, 376 N.W.2d 553 (1985). In short, under the Nebraska concurrent cause cases, neither group causation nor reliance on a solitary substantial factor test has replaced the need for each liable defendant to be a but-for and substantial contributor to the indivisible injury.9
B. Sufficiency of the Evidence Under Traditional Standards
The case-in-chief was tried as if the test for causation in Nebraska were a substantial factor test alone, and Menne attempted to prove that the defendants, either individually or collectively, substantially contributed to his mesothelioma.10 After hearing the case-in-chief, the trial court announced to the parties its decision to deviate from Nebraska’s traditional tort standards. Believing the situation to be inherently unfair to an innocent plaintiff exposed to asbestos dust by multiple manufacturers, the court predicted that Nebraska would adopt a special rule to relax the burden of proof and redress the difficulties of proving causation. Menne argues on appeal that the court’s prediction was never communicated clearly to the jury and that he overcame the difficulties and proved causation under traditional burdens of proof. Before examining the trial court’s prediction, we pause to show that Menne did not and could not prove causation under either a pure substantial factor test or under Nebraska’s but-for and substantial factor standard, in effect illuminating why the court felt compelled to depart from them and attempt a burden shift.
As a threshold matter, the trial court instructed the jury that Menne had to prove his exposure to defendants’ products; without proof of actual exposure, there could be no proximate cause.11 The fact that Menne merely worked at a job site where defendants’ asbestos products were used was insufficient to establish actual *1462exposure. See Roehling v. Nat’l Gypsum Co. Gold Bond Bldg. Products, 786 F.2d 1225, 1228 (4th Cir.1986) (evidence must establish that plaintiff “was in the same vicinity as witnesses who can identify the products causing the asbestos dust that all people in that area ... inhaled”); see also Blackston v. Shook & Fletcher Insulation Co., 764 F.2d 1480, 1485 (11th Cir.1985) (asbestos victim must “prove exposure to a particular defendant’s product at some time and place”); Gideon v. Johns-Manville Sales Corp., 761 F.2d 1129, 1144-45 (5th Cir.1985) (no evidence that asbestos fibers from Raymark products were a substantial causative factor because plaintiff had only slight contact with Raymark products, and no testimony showed that any dust was created when plaintiff handled the products; same conclusion with respect to exposure to products of Standard Insulation, Inc., after 1969); Migues v. Fibreboard Corp., 662 F.2d 1182, 1185 (5th Cir. Unit A 1981) (three co-workers established decedent’s exposure to defendant’s asbestos products).12
Menne used Maurice Lane’s testimony to prove his (Menne’s) actual exposure. Lane’s testimony indicated that, typically, on any given ship, he worked with the asbestos products of all the identified manufacturers, allowing the inference that the products of each appellant were among those used by other insulators on other ships, including those on which Lane did not work. Since Menne worked alongside insulators at least during 1944-45 (whether the actual length of time is longer than that is indeterminable from the record), a jury could infer the probability that Menne was exposed at some point to some asbestos dust from products of each appellant. The jury apparently found this indirect, circumstantial evidence sufficient to establish Menne’s exposure to appellants’ products.
More significant under traditional causation tests than the question of mere exposure to appellants’ products is whether the exposure was sufficiently sustained (or frequent) and intense to constitute a proximate cause of Menne’s mesothelioma.13 *1463Dr. Louis W. Burgher, Menne’s pulmonologist, testified that slight or negligible exposure does not create any significant risk. R.Vol. 5 at 347-48. Based on his experience with other shipyard workers, Dr. Burgher opined that a two-year exposure to asbestos at significant levels, as he considered Navy World War II shipyard exposure to be, was enough to be the producing cause of mesothelioma. Id. He was not in a position to address whether Menne’s exposure to defendants’ products contributed to the attainment of significant levels.
Menne’s chief expert witness on causation, Dr. Elliott McCaughey, testified that the duration and intensity of exposure are the key factors in determining the risk of developing mesothelioma from inhalation of asbestos fibers. Based on the assumptions that (1) the products of each defendant were on all the ships on which Menne worked and (2) those products were exposing Menne to asbestos dust at visible levels, i.e., above what was then considered the safety threshold level, Dr. McCaughey stated that in his opinion the products of each defendant were more likely than not to have significantly contributed to the development of Menne’s mesothelioma. R.Vol. 6 at 693, 702-05, 708-12, 755. In response to a question regarding exposure to dust from Raybestos cloth, Dr. McCau-ghey clarified that his medical opinion assumed exposures of more than just very short, limited time periods. Id. at 755. He also acknowledged, however, that exposures of just several months’ duration had been known to cause mesothelioma, presumably where they were sufficiently intense. Id. at 714.
Even if the three above-italicized assumptions, which formed the basis for Dr. McCaughey’s medical opinion, were proved by a preponderance of the evidence as required by Nebraska law, and even if the jury then accepted the doctor’s opinion, his opinion would establish only that the products of specific defendants created a significant risk of harm. His opinion cannot and does not establish that the products of a given defendant were a but-for and substantial cause of Menne’s mesothelioma. But-for causation does not follow naturally and logically even from extensive exposure to asbestos dust from defendants’ products if the harm would have occurred without it, i.e., from other sources such as the products of nondefendants. The evidence was insufficient to establish whether the dust from the products of each appellant might have pushed the dust levels over the safety threshold enough times during Menne’s working day or added enough amounts to already dusty air so that without them Menne would not have contracted mesothe-lioma. The expert testimony on causation never addressed but-for causation; the thrust of Dr. McCaughey’s opinion was that the products of each appellant were a significant contributor if they created visible asbestos dust for more than short, limited periods of time on all the ships on which Menne worked. Evidence as to but-for causation is entirely lacking, and the jury would have had to base its decision on pure conjecture.
Under a federal test of the sufficiency of circumstantial evidence, an inference must not be based on conjecture, speculation, or mere possibility. See Lohrmann v. Pittsburgh Corning Corp., 782 F.2d 1156, 1163 (4th Cir.1986) (“permissible inferences must ... be within the range of reasonable probability”) (quoting Ford Motor Co. v. McDavid, 259 F.2d 261, 266 (4th Cir.), cert. denied, 358 U.S. 908, 79 S.Ct. 234, 3 L.Ed.2d 229 (1958)); Lisa-Jet, Inc. v. Duncan Aviation, Inc., 569 F.2d 1044 (8th Cir.), cert. denied, 439 U.S. 828, 99 S.Ct. 100, 58 *1464L.Ed.2d 121 (1978) (quoting Ford Motor Co. v. Mondragon, 271 F.2d 342, 345 (8th Cir.1959), that “[t]he essential inference cannot be left to conjecture and speculation”). The same is true of Nebraska law.14
Had the trial court applied Nebraska’s law of but-for and substantial causation, it necessarily would have concluded that Menne could not and did not meet his burden of proof under traditional Nebraska proximate cause standards. Apparently recognizing that Menne could not meet the stringent causation requirements, the trial court predicted that under the circumstances of this case, the Nebraska Supreme Court would modify those requirements and allow part of the burden of proof on this issue to shift to the defendants.
The trial court apparently also concluded that even if Nebraska eliminated but-for causation in a case such as this, a jury could not reasonably find that each appellant was a substantial contributing cause of Menne’s mesothelioma. Otherwise it would have felt no need to invoke a burden shift. We reach the same conclusion. The testimony was insufficient to establish the likelihood of frequent or sustained exposure to asbestos dust from the products of each appellant, and a jury finding of some Shipyard exposure from each’s products is not enough. From the circumstantial evidence, there is no way of ascertaining the regularity or frequency of Menne’s exposure to visible asbestos dust from a given defendant’s products. Such exposure might have been frequent or sustained, or it might have been sporadic and short. On this matter the jury could only speculate. In short, even under a pure substantial factor test, Menne could not have proved causation. We, therefore, turn to the burden shift attempted by the trial court.
C. Burden Shifting Under a Theory of Alternative Liability
Recognizing the peculiar problems of causation that frequently arise in asbestos litigation, the trial court twice instructed the jury (in instructions # 8 and # 25) that, under the factual circumstances of this case, Nebraska law would allow a shift in the burden of proof with respect to causation. Instruction # 26 infers some kind of burden shift as well. The court’s published opinion explaining its jury instructions makes clear that the court was attempting to invoke a theory of alternative liability whereby the plaintiff’s burden of proving causation is relaxed and defendants are charged with proving the absence of causation. Menne v. Celotex, 641 F.Supp. 1429, 1438 (D.Kan.1986). The opinion indicates that the court meant to be making an Erie prediction that Nebraska law would apply a theory of alternative liability in asbestos tort litigation, so that injured, innocent plaintiffs could overcome the frequently impossible burden of proving proximate cause under traditional tort standards.15
*1465Appellants challenge the burden shift found in the jury instructions, asserting that there is no basis for concluding that Nebraska would adopt a theory of alternative liability in this case. In contrast and as already noted, Menne asserts that he made out a prima facie case of causation under the existing Nebraska concurrent cause standards and that the case was not decided on grounds of alternative liability. We rejected this view of Nebraska law (and the evidence) in Sections IA and IB of this opinion, but because of this view Menne treated appellants’ arguments on alternative liability in summary fashion, relying instead on the trial court’s justifications in the Menne opinion.
The most explicit jury instruction on the burden shift appears in Instruction #25, which states in pertinent part:
“In the course of your deliberations, you will be required to make certain findings, including the following:
Did one or more of the defendants supply asbestos-containing products at plaintiff’s workplace?
Was the plaintiff exposed to quantities of asbestos dust from defendants’ products at his workplace?
Did plaintiff sustain mesothelioma and/or asbestosis caused by his exposure to asbestos dust from defendants’ products at the workplace?
Should the jury make these findings from a preponderance of the evidence, the burden shifts to each defendant to prove its products did not cause or significantly contribute to plaintiff’s disease and injuries. If a defendant proves to your satisfaction that, more likely than not, its products were not contributing factors, you may find for that defendant and against plaintiff. ...
Special interrogatories are provided to assist you in these deliberations.”
R.Yol. I at tab 37 (emphasis added).16 Although, in our view, this language does not clearly invoke alternative liability or any variant thereof, we will assume for the moment that the jury so understood it and will proceed to examine the validity of an alternative liability theory under the circumstances of this case.
Alternative liability is a theory that first attracted widespread attention in the California Supreme Court case of Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1 (1948). In Summers, two quail hunters both negligently fired shotguns toward the plaintiff, and birdshot from one of the guns injured his eye. Because the two hunters fired more or less simultaneously, the plaintiff could not identify whose shot was responsible for his eye injury. The court found that both defendants had acted negligently toward an innocent plaintiff. Under traditional proximate cause standards, however, the plaintiff would have been unable to show which defendant was likely to have caused the eye injury. To deal with this *1466problem, the lower court shifted to each defendant the burden of proving that he was not the cause of the injury, and the California Supreme Court affirmed.
Under alternative liability theory, a plaintiff does not prove individual causation as part of a prima facie case but rather just a threshold level of factual connection (e.g., gunfire from one of the defendants caused the eye injury). The problem is that, although any one of the defendants could have caused the injury, only one actually did so and the plaintiff cannot identify which one. The court, therefore, presumes that each defendant caused the whole injury unless it can prove otherwise. The Summers burden shift seems intuitively fair where all those who could have caused the injury are before the court and the odds are equal and significant that each is liable and to the same degree. If none can prove nonliability, then all can fairly be held jointly and severally liable. Even if, ultimately, one is required to pay for the entire injury, judgment against one defendant alone may not be fundamentally unfair since the odds are equally high that he caused the entire injury.
The theory of alternative liability was generated where only one of the defendants was thought to have caused the specific harm to a specific party. Under such conditions, the requirement that all the potential defendants be before the court assures that the cause is before the court and will be among those found liable.17 Thus, the threshold test for invocation of alternative liability is proof that the one who caused the plaintiffs injury is among the defendants. The theory is being applied here, however, in a concurrent cause situation where more than one defendant is thought to have caused the harm, i.e., substantially contributed to it. Here, in our view, the threshold test for a burden shift should be modified to require that more than one of the defendants contributed at least some harm, even if an unquantifiable amount. Thus, all those who are subsequently found liable will have been found at least to have caused some harm. Asbestos cases citing the requirement that all possible defendants must be before the court ignore the difference in the two circumstances, incorrectly in our view. See Marshall v. Celotex Corp., 651 F.Supp. 389, 392 (E.D.Mich.1987); Vigiolto v. Johns-Manville Corp., 643 F.Supp. 1454, 1457 (W.D.Pa.1986), aff'd, 826 F.2d 1058 (3d Cir.1987); Starling v. Seaboard Coast Line R.R., 533 F.Supp. 183, 188 (S.D.Ga.1982).
We are convinced that in cases of concurrent causation such as this, if all or substantially all of the available and identifiable, implicated manufacturers are before the court, and if some of these defendants can be shown to have each contributed some harm at a possibly substantial level, then all potential defendants need not be before the court. Where such a defendant can be shown actually to have exposed a plaintiff to visible, and hence unsafe, levels of asbestos dust, some harm has occurred, even if it is small and unmeasurable.18 And where frequent or extended exposure cannot be ruled out, the harm may have been substantial. Where such is the case, it seems no more unfair to require the defendant to prove that its dust did not produce substantial harm than to require a wrongdoer with no demonstrated causal connection to prove that it was not a cause at all.19
*1467Some jurisdictions, in effect, have interpreted their concurrent cause standards in a similar fashion, allowing exposure to constitute proof of injury. See, e.g., Lockwood v. AC & S, Inc., 109 Wash.2d 235, 744 P.2d 605, 612-13 (1987) (jury instruction on concurrent causes required that each be a substantial factor; since exposure has cumulative effect in contributing to asbestosis, evidence of exposure to a defendant’s asbestos dust established prima facie case of causation, i.e., in effect, exposure even at unknown levels is a substantial factor); Roehling v. Nat’l Gypsum Co. Gold Bond Bldg. Products, 786 F.2d 1225, 1228 (4th Cir.1986). Cf. Restatement (Second) of Torts § 433B(2) (1965).20 The fact that Nebraska has not yet had occasion to extend its own common law in a parallel manner does not indicate that it would not do so under the circumstances presented here. Where not all potential defendants are available to be sued, and yet where at least some of the available defendants can be shown each to have produced some harmful exposure, we think Nebraska would conclude that a burden shift is appropriately invoked, although we prefer the term concurrent liability to the term alternative liability as a means to describe the nature of the burden shift.21 Shifting the burden seems at least as fair where some, if not all, defendants are shown to have contributed some of the harm as where only one of them is thought to have caused all the harm: in the former situation a liable defendant will be shown at least to have actually caused some harm; in the latter a defendant who is entirely innocent of causing any of the single injury can be found liable.
In trying to anticipate the law of Nebraska the trial court placed heavy reliance on the Nebraska case of Osteen v. A.C. and S., Inc., 209 Neb. 282, 307 N.W.2d 514 (1981) to support its view that Nebraska would adopt a form of alternative liability.22 In Osteen a widow sought worker’s compensation from forty employers for whom her deceased husband had been an asbestos worker. The widow was unable to pinpoint which exposure co asbestos had caused her husband’s mesothelioma. The Nebraska Supreme Court eliminated the guesswork by charging the entire amount of the awarded compensation against the last employer contributing to the de*1468ceased’s exposure in a causal way. In the absence of a statute determining liability of successive employers, the court adopted the last injurious exposure rule, emphasizing that the purpose of the workers’ compensation legislation was to serve the interests of employees “rather than attempt an adjustment of their rights in the light of equities that may exist between [successive employers].” Osteen, 307 N.W.2d at 519 (quoting Wilson v. Van Buren County, 198 Tenn. 179, 278 S.W.2d 685, 688 (1955)). In other words, the statutory preference given to injured employees by the worker’s compensation act was a factor in the outcome.
The Osteen court’s choice of the last injurious exposure rule is closer to abandonment of but-for causation and substitution of a substantial factor test23 than to a special burden-shifting rule. Under the last injurious exposure rule, “some contributing exposure” is enough, thus obviating the need for the plaintiff to prove the unprovable and litigate the unlitigable. Osteen, 307 N.W.2d at 520. Since the rule is widely employed in workers’ compensation cases, the hardship that falls to a given employer in one situation falls to others in other situations, thus spreading the risk. While the successive employer context of a workers’ compensation case, where the statutory purpose of favoring the employee is clear, goes only part way toward establishing that the Nebraska Supreme Court would invoke a burden shift on causation in Menne’s situation, it does suggest a willingness to depart from but-for causation requirements in a special context and demonstrates the Court’s acceptance of pragmatic solutions to difficult tort problems.24
Giving credence to Osteen’s sympathies, and siding with what we think to be the fairest approach to the dilemma of proving causation from harmful exposures of unknown frequency, we affirm the trial court’s prediction that a burden shift would be invoked by the Nebraska Supreme Court in this case. We hold that, under Nebraska law, proof of actual exposure in confined spaces to visible asbestos dust from a defendant's products, within a time period relevant to the acquisition of the injury, and under circumstances where the exposure could have been extensive enough to produce substantial harm, establishes a prima facie case of substantial causation against that defendant. Under these conditions, where Menne can demonstrate the likelihood of exposure to visible dust from a defendant’s product(s), Nebraska law would then require that defendant to prove the exposure was unlikely to have been frequent or long enough to be a substantial factor in causing Menne’s mesothelioma.25 In other words, where the requirement of but-for causation would defeat a plaintiff’s claim in a concurrent cause case, we believe that Nebraska would adopt the more lenient substantial factor test of causation. And where the plaintiff can show that a defendant’s asbestos dust caused some *1469harm and possibly substantial harm but cannot show the relative contribution of that defendant to the overall harm, we believe that Nebraska would allow a burden shift to the defendant. Where a defendant lacks evidence as to the frequency or duration of exposure, the burden shift may well result in a finding that the defendant is a cause of the harm. As a policy matter, this seems fairer than a finding that none of the defendants was one of the causes even though each had been shown to have caused some harm to the plaintiff which could be linked to his injury.
The trial court appreciated that its alternatives to traditional tort liability were limited by Nebraska law. It recognized the difficulties with fashioning any judicial apportionment on the basis of market share and deferred to the Osteen case in declining to impose apportionment.26 Similarly, it properly deferred to Nebraska law in not trying the case under a doctrine of comparative fault of defendants.27 What the trial court did not consider is the possibility that where there is no basis for apportionment or comparative fault among concurrent causes (i.e., several liability), Nebraska would impose joint (pro-rated) liability instead. We believe that such a prediction would have completed the conceptualization employed by the trial court. Where a plaintiff does not prove causation under traditional tort standards, and where all potential defendants cannot be brought before the court, yet where the plaintiff has sued all or substantially all available, identifiable, and implicated manufacturers and where liable defendants are each found to have contributed some causally connected harm, we believe Nebraska would view the concomitant liability as joint rather than joint and several.28 Cf. Pressler & Schieffer, Joint and Several Liability: A Case for Reform, 64 Denver U.L.Rev. 651, 674 (1988) (no joint and several liability should attach for being one among multiple concurrent causes rather than the cause of injury); Comment, DES and a Proposed Theory of Enterprise Liability, 46 Fordham L.Rev. 963, 988 (1978) (joint liability usually found with concurrent cause, although theories vary). Since Nebraska has not addressed the burden-shift question, it has not addressed the liability that would complete the equation. Nonetheless, we predict that, out of a sense of fairness to both plaintiffs and defendants, it would determine that pro-rated liability balances the equities better than does joint and several liability.29
If judgment can be executed against all liable defendants, joint liability will provide a full recovery with the risk spread across the available defendants. If only partial recovery is possible, the recovery still may be fuller than under some comparative fault or apportionment standards where the amount recoverable from defendants can be limited to their share of the total damages assessed against all wrongdoers *1470(including “phantom” defendants, i.e., those not before the court).30 Should a plaintiff not be able to recover a portion of the total damage from some of the defendants, a partial recovery is better than no recovery at all and is not fundamentally unfair to a plaintiff who cannot show that a substantial contributor is among the defendant wrongdoers. In our judgment, Nebraska would conclude that the complexities of some asbestos cases require something of a compromise solution rather than a winner-take-all mentality.
D. Effect of the Jury Instructions
Although we affirm the trial court’s prediction that Nebraska would adopt a burden shift with respect to causation under the circumstances here, the relevant instructions, particularly # 25, # 26 and # 8, and the special interrogatories do not make clear what the prima facie elements of causation are, nor clearly invoke a burden shift.31 First of all, the second question posed both in instruction # 2532 and the interrogatories does not make explicit what threshold level of exposure is required to invoke a burden shift. By asking the jury to decide whether Menne was exposed to “quantities” of asbestos dust, the trial court did not specify what threshold standard of exposure the jury was to use, and we have no way of knowing whether it equated the term to exposure to visible asbestos dust under conditions allowing the possibility of substantial harm. The term is imprecise enough that the jury even may have thought that it was to establish its own definition of quantities. Furthermore, we cannot tell if it interpreted the term as applying, in context, to dust from the products of each individual defendant or to dust from the collective products of all defendants.
Even more crucially, the third question in instruction # 25 and the interrogatories did not clearly invoke a burden shift on causation. To do so, the question should have asked simply whether Menne had mesothe-lioma (as opposed to a benign hyperplasia). If so, and if the jury had construed the second question under the proper threshold test of exposure, then the instructions and interrogatories would have allowed the jury to presume the mesothelioma was substantially caused by that defendant, unless it could prove otherwise. Instead, the third question suggested that Menne retained the burden of proving, under Nebraska’s combined “but-for” and substantial-factor standard, that his mesothelioma was caused by exposure to defendants’ products, either individually or collectively.
In reviewing claims of error in an aspect of the jury instructions, the instructions are to be considered as a whole to see if together they convey the correct statement of applicable law. See Hipsher v. Lund, 827 F.2d 337, 339 (8th Cir.1987); Furr v. AT & T Technologies, Inc., 824 F.2d 1537, 1549 (10th Cir.1987). “The test is ... *1471whether the jury was misled in any way and whether it had understanding of the issues and its duty to determine those issues.” Borel, 493 F.2d at 1100. Using these standards, we conclude that the jury instructions and special interrogatories concerning causation were ambiguous enough to constitute significant and prejudicial error.
We conclude that the jury had an inadequate basis for determining the threshold test of exposure. We also conclude that the jury could not ascertain from the instructions which side bore the burden of proof with respect to causation. On the one hand, reference to a burden shift suggests a departure from traditional Nebraska concurrent cause standards. On the other hand, failure to delete language indicating that the plaintiff must actually prove that his mesothelioma was “substantially caused” by exposure to defendants’ products suggests that the burden shift was not a genuine one. Instead it may have been one that referred simply to the defendants’ burden to rebut the plaintiff’s prima facie case with evidence that Menne was not actually exposed to any quantities of asbestos dust from their products. We cannot determine with reasonable assurance whether the jury found the evidence sufficient under individual but concurrent causation standards, a form of alternative liability possibly based on an incorrect view of the required threshold test of causal relationship, or even conceivably under group causation' standards.
If the jury viewed the burden of proving causation as Menne’s, then the evidence was insufficient to support a jury verdict under but-for and substantial causation standards. As described in sections IA and IB of this opinion, Menne could not win under traditional Nebraska causation standards. He also could not win under a group causation standard, even if Nebraska had adopted one.33 If, however, the jury viewed the plaintiff’s threshold burden as proof of some exposure to visible asbestos dust from an individual defendant’s products that reasonably could have caused substantia] harm originating within a relevant time frame, then perhaps Menne would have been entitled to his jury verdict on the issue of causation under the evidence developed in the record. As noted, however, we cannot ascertain whether the jury used such a threshold level of proof, and we decline to review the evidence for sufficiency under a standard that we cannot determine was used by the jury.
II. Strict Liability and State of the Art
Appellants assert other errors in the jury instructions. In particular, they contend that even if Menne made out a prima facie case of causation under traditional tort law or a properly invoked special rule of alternative liability, the judge instructed the jury improperly with respect to the defendants’ affirmative defense of the state of the art in the asbestos industry during the 1940s. First, appellants argue that instruction #11 did not conform to Nebraska statutory law on the state-of-the-art defense. Second, they argue that subsequent jury instructions were inconsistent with the state-of-the-art instruction and confused the jury. Finally, they argue that the special interrogatories should have included a question on the state-of-the-art defense.
As part of his prima facie case, Menne was required to establish that the defendants knew or should have known of the unreasonable dangers attaching to a failure to warn or otherwise control for the hazards attached to asbestos dust exposure. Under a theory of either strict liability or of negligent failure to take the proper precautions, a state-of-the-art defense was available to the defendants. Nebraska’s statutory provision allowing a state-of-the-*1472art defense in product liability cases reads as follows:
“In any product liability action based upon negligent or defective design, testing, or labeling, proof establishing that such design, testing, or labeling was in conformity with the generally recognized and prevailing state of the art in the industry at the time the specific product involved in the action was first sold to any person not engaged in the business of selling such product shall be a defense. State of the art as used in this section shall be defined as the best technology reasonably available at the time (emphasis added).”
Neb.Rev.Stat. § 25-21,182 (Reissue 1985). Initially, we note that we need not decide whether the statutory concept of “defective labeling” includes a complete failure to warn against the health hazards arising from exposure to asbestos dust; the record and appellate briefs reveal that all parties assumed that it was and do not raise this issue. Cf. Rahmig v. Mosley Machinery Co., 226 Neb. 423, 412 N.W.2d 56, 72 (1987).
Instruction #11 on the state-of-the-art defense reads as follows:
“With regard to your deliberation, principally as to defendants’ duties as manufacturers or suppliers of asbestos products, you are reminded that we deal here with events as they allegedly transpired in the year 1948 and prior years. You will test the respective defendants’ liabilities, if any, within the framework of the generally recognized and prevailing state of the art in the asbestos industry at that time.
State of the art is defined as the best technology reasonably available at a given time.
Your verdict must be for the defendants if you find that prior to 1948 the defendants acted in conformity with the generally recognized and prevailing state of the art in the industry for the design, testing and labeling of insulation.
But where scientific or medical evidence exists tending to show that a certain danger is associated with the use of a product, the manufacturer may not ignore or discount that information in drafting its warning solely because it finds it to be unconvincing.”
R. Vol. I at tab 37 (emphasis added).
Appellants, as would be expected, have no objection to that portion of the jury instruction which merely states that proof of their compliance with the generally recognized and prevailing state of the art will be a complete defense to their liability. What they object to is the final paragraph quoted above. Appellants contend that this paragraph misled the jury and imposed a stricter duty on appellants than that imposed by statute. We agree. The additional language suggests that manufacturers should have supplied warnings reporting any scientific or medical evidence tending to show an association between inhalation of asbestos dust and mesothelioma regardless of whether that evidence represented the generally recognized and prevailing best technology reasonably available to the industry during the 1940s. Under a state-of-the-art defense, warnings of scientific or medical concerns should not be required if the concerns are not generally recognized as reflecting the best data reasonably available at the time.34
The source of the disputed statement used in the jury instruction is Seley v. G.D. Searle & Co., 67 Ohio St.2d 192, 423 N.E.2d 831, 837 (1981) (citing Mahr v. G.D. Searle & Co., 72 Ill.App.3d 540, 28 Ill.Dec. 624, 641, 390 N.E.2d 1214, 1231 (1979)).35 *1473See also Hamilton v. Hardy, 37 Colo.App. 375, 549 P.2d 1099, 1108 (1976). Although the Seley court holds that under Ohio law the strict liability standard for a failure to warn may be stricter than that for mere negligent failure to warn, the Seley language with respect to the duty to warn has been echoed in cases being tried under negligent failure to warn theories. See Wooderson v. Ortho Pharmaceutical Corp., 235 Kan. 387, 681 P.2d 1038, 1054, cert. denied, 469 U.S. 965, 105 S.Ct. 365, 83 L.Ed.2d 301 (1984). Importantly, neither Seley nor Wooderson used the language in the face of a statutory state-of-the-art defense.
A state-of-the-art defense is provided to benefit defendants, not plaintiffs. It should not be construed to have the effect of tightening rather than relaxing Nebraska’s standards governing failure to warn. Although a jury instruction need not limit itself to Nebraska statutory language so long as the instruction as a whole correctly represents Nebraska law, Monahan v. Flannery, 755 F.2d 678, 681 (8th Cir.1985), here the additional language, introduced by the negating term “but,” had the effect of undermining the state-of-the-art defense. Neither before nor after the passage of Nebraska’s state-of-the-art provision is there any indication that Nebraska would adopt the trial court language quoted from the Seley case.
We conclude that instruction # 11 is prejudicially flawed. As a result of our ruling we need not reach the question of whether various instructions were inconsistent with respect to the state-of-the-art defense, nor whether that defense should have been included in the special interrogatories. We observe, however, that on retrial a jury should not have to strain to read separate instructions in a manner that makes them consistent with one another. We also note that, at least in this Circuit, once the court decides to submit special interrogatories on a given element of a plaintiff’s claim, all material factual issues should be included in such interrogatories. Martinez v. Union Pacific Railroad Co., 714 F.2d 1028, 1032 (10th Cir.1983).
III. Miscellaneous Issues
Appellants assert that the court improperly instructed the jury by explaining the supposed public policy justification for strict liability in tort and by stating the maximum amount of damages sought in this case. Finally, they argue that since the loss of consortium claim was tried under Nebraska rather than Kansas law, it should have been filed in Mary Menne’s name rather than that of her husband. Appellants contend that filing the claim in Don Menne’s name, as required under Kansas law, exposes them to duplicate claims for loss of consortium.
As for the public policy justification for strict liability in tort, we make no blanket ruling. We simply caution against statements explaining policy justifications for various theories of liability, since those rationales shift with time and subject matter and may not apply to a given subset of litigation. The public policy justification offered here, i.e., that manufacturers are in a better position than consumers to know whether their products are unreasonably dangerous and are more likely to minimize the risk of injury if they are held responsible for injuries caused by the product, seems fair enough in the context of a failure to warn of the hazards attached to use of the product. Nonetheless, we can conceive of situations where the attempted statement of policy justification may not be applicable to the case at hand. In cases raising complex, multiple policy considerations it seems to us better to err on the side of omitting it than misstating it.36
*1474As for the instruction limiting plaintiffs damages to five million dollars, guidance as to the giving of such an instruction is provided and implicitly controlled by Federal Rule of Civil Procedure 54(c) and is a matter of federal procedure, not state substantive law. Rule 54(c) provides that “Except as to a party against whom a judgment is entered by default, every final judgment shall grant relief to which the party in whose favor it is rendered is entitled, even if the 'party has not demanded such relief in his pleadings (emphasis supplied).” The Eighth Circuit has applied this rule to allow judgment in excess of the amount sought in the complaint. See Troutman v. Modlin, 353 F.2d 382 (8th Cir.1965); Stineman v. Fontbonne College, 664 F.2d 1082, 1088 (8th Cir.1981).
If a plaintiff is not limited to the amount of damages sought in the complaint, then it would seem improper for a judge so to instruct a jury, regardless of whether the effect is to limit the damages awarded or to enhance them as is alleged here. We conclude, therefore, that it was error to tell the jury that the damages were limited to five million dollars. Nonetheless, we view the error as harmless because there is no evidence or argument that the jury would have awarded more than five million dollars if not so instructed. The only argument is the converse, i.e., that telling the jury of a five million dollar limit was somehow responsible for the jury award of two and a half million dollars, since the compensatory damages for lost wages and past medical expenses were estimated at less than $50,000. Although the amount is indeed extremely large, the amount to be awarded for pain and suffering, anticipated medical expenses, and loss of consortium lies in the reasonable discretion of the jury. Mere speculation that the jury instruction unduly influenced the award is no basis for overturning it. Not only did the trial court find no basis for remittitur but the appellants decided not to appeal the trial court’s denial of remittitur. Although we decline to reverse on the inclusion of a damage ceiling in the instructions, we do caution the trial court against its reassertion on retrial.
Finally, as for the Nebraska requirement that a spouse, in order to recover damages for loss of consortium, must be a party to the suit, we simply instruct the parties to amend the pleadings to meet this technical requirement prior to retrial.
IV. Conclusion
One might wish for a system that produced just outcomes with more nicety in complex cases such as this one. Both traditional and innovative judicial remedies for asbestos tort injuries leave serious difficulties in their wake, and as others before us have observed, asbestos tort injuries seem particularly well suited for redress through legislative action.
Our rulings in this case have attempted to provide a framework for retrial that minimizes and balances the injustice to both plaintiff and defendants, based on our understanding of the constraints of Nebraska law. We believe that uncertainty about Nebraska tort law and the correct way to invoke a special burden-shifting rule permeated this trial.37 We affirm the trial court’s predictions with respect to a burden shift on the issue of causation and its effect on Nebraska’s causation standards — with the modification that joint liability would apply in a concurrent cause situation such as the one here. We hold that, in the case of mesothelioma caused by exposure to asbestos dust during the 1940s, the threshold level of proof required prior to shifting the burden to a defendant is proof of exposure to visible asbestos dust from its products in confined spaces *1475within a limited time period and under circumstances creating the reasonable possibility of substantial harm. Because of confusing and ambiguous jury instructions with respect to causation, and incorrect instructions with respect to the state-of-the-art defense, we remand for a new trial. If, on remand, the trial court wishes to certify to the Nebraska Supreme Court any of the questions addressed in our rulings or other uncertainties about Nebraska law, it, of course, is free to do so.38
. After Menne’s death, his wife was substituted as the party plaintiff in her role as representative of the estate. For convenience we refer to the appellant as Menne.
. Menne sought recovery for his mesothelioma, not his asbestosis, and the expert opinion testimony with respect to causation was limited to mesothelioma. Despite this fact, the trial court asked the jury to determine whether Menne’s "mesothelioma and/or asbestosis" was caused *1457by exposure to defendants’ products. Appellants challenged the inclusion of asbestosis in this instruction, but because of our resolution of the other issues in this case, we need not reach this issue.
. In cases such as this it is not necessarily the asbestos products themselves that are unreasonably defective but rather the manufacturer’s failure to warn of their reasonably foreseeable hazards. Failure to adequately warn of such hazards renders the product unreasonably dangerous. Cf. Restatement (Second) of Torts § 402A, comments j and k (1965); see Borel v. Fibreboard Paper Products Corp., 493 F.2d 1076, 1089, 1091 (5th Cir.1973), cert. denied, 419 U.S. 869, 95 S.Ct. 127, 42 L.Ed.2d 107 (1974).
. By the time of the suit, both Johns-Manville and Unarco were in bankruptcy. Keasbey & Mattison had been disbanded and sold in 1962. Mundet, or at least its contract operations, had been acquired by Baldwin-Ehret-Hill, which in turn had been acquired by Keene Corporation, a defendant at trial. Appellee’s Brief at 24; Appellants’ Reply Brief at 9.
.Although the reason for the jury’s finding of nonliability is not available, we note that Maurice Lane, a Shipyard insulator and key witness at trial, had not been sure when the Owens-Illinois asbestos product, Kaylo, had actually arrived at the Shipyard; in either case it was apparently not until 1946 or 1948. Also, Lane’s memory with respect to the Keene (Ehret) products was dim, and the defense pointed out that no certain identification of Keene products at the Shipyard had been made in an earlier deposition by Lane. The plaintiff does not appeal their nonliability, and they are, therefore, dismissed from inclusion among the defendants at any retrial of this case.
. Appellants do not challenge the sufficiency of the evidence to support a jury finding that Men-ne had mesothelioma, and Menne’s mesothelio-ma is not only established for purposes of this appellate review but also for purposes of retrial.
. Instruction # 13 reads:
‘“Proximate cause’ means a moving or effective cause or fault which, in a natural and continuous sequence produces the injury, and without which the injury would not have occurred.
A party’s conduct is a proximate, or legal, cause of harm to another if its conduct was a substantial factor in bringing about that harm.”
R.Vol. 1 at Tab 37 (emphasis added).
. See, e.g., Borland v. Gillespie, 206 Neb. 191, 292 N.W.2d 26, 29 (1980) and Daniels v. Andersen, 195 Neb. 95, 237 N.W.2d 397, 402 (1975) (both invoking but-for causation, i.e., without the fault of the defendant, the harm would not have occurred); see also Petznick v. United States, 575 F.Supp. 698, 709 (D.Neb.1983). Cf. Pendleton Woolen Mills v. Vending Assoc., Inc., 195 Neb. 46, 237 N.W.2d 99, 102-03 (1975) (employing both a but-for and a substantial factor test of causation and noting that both had been cited in the past). A search of Nebraska law reveals that when substantial factor language has been used, it has functioned to clarify when to apply the but-for test of causation; i.e., if one’s negligence is so slight as not to be a substantial factor, then even though it may have been a but-for cause of the harm, it is not significant enough to result in legal responsibility for that harm. See, e.g., Landmesser v. Ahl-berg, 184 Neb. 182, 166 N.W.2d 124, 127 (1969). Even in concurrent cause cases invoking substantial-factor language alone, but-for causation is present. See, e.g., Brown v. Nebraska Public Power Dist., 209 Neb. 61, 306 N.W.2d 167, 171 (1981) (quoting Libbey-Owens Ford Glass Co. v. L &M Paper Co., 189 Neb. 792, 205 N.W.2d 523, 529 (Neb.1973) that negligence can be a concurring proximate cause if it constitutes a "continuing and substantial factor" in producing the harm). In the Nebraska concurrent cause cases, even though the action of any one negligent actor has been insufficient to produce the injury, the harm would not have occurred but for its contribution. We, therefore, reject Men-ne’s assertion that Nebraska has substituted a substantial factor test for a test of but-for causation.
The only abandonment of Nebraska’s but-for test of which we are aware occurred in the context of a workers’ compensation case, where compensation for asbestos-related diseases depended on establishing that they were occupationally related and where the Nebraska Supreme Court acknowledged that the proximate cause burden was relaxed to require just a causal relationship between the worker’s diseases and his employment. See Osteen v. A.C. and S., Inc., 209 Neb. 282, 307 N.W.2d 514, 520 (1981). The trial court used this case as a basis for predicting that in private tort cases involving concurrent cause Nebraska would adopt an alternative liability theory. In so doing, it implicitly predicted that Nebraska would substitute a substantial factor test for but-for causation. See Section IC of this opinion.
.Prosser and Keeton suggest that collective or group but-for causation could be a permissible test of causation in tort suits but only when "application of the but-for rule to the [negligent actors] individually would absolve all of them” and when “each of the defendants bears a like relationship to the event.” Prosser and Keeton On The Law of Torts, § 41, p. 268 (5th ed. 1984). In these circumstances Prosser and Keeton see group but-for causation as preferable to application of an individual substantial factor test, presumably since it would allow recovery by an innocent plaintiff against defendants all of whose individual contributions might have been slight.
In the case before us, if none of the asbestos manufacturers could be shown to be a but-for cause, then all would be absolved, thus meeting the first part of the Prosser and Keeton test. Whether the asbestos manufacturers in this case bear a like relationship to Menne's injury then would become a critical issue under group but-for causation. If all the defendants were shown to have supplied asbestos to the Shipyard and to have exposed Menne to their asbestos products, one might argue that they bear such a relationship. On the other hand, if the physical characteristics, fiber emission levels, and amount of asbestos products supplied by each to the Shipyard are quite different, arguably they do not bear such a relationship to the harm. Moreover, if not all negligent actors are before the court, it may be impossible to ascertain whether a limited number of defendants collectively rise to a but-for and substantial cause of the harm. If the number of defendants is quite small and the extent of exposure to each’s asbestos dust is unknown, it may be impossible to conclude that collectively their unknown dust levels were a substantial factor in producing Menne’s meso-thelioma. Since we conclude that Nebraska has not adopted group causation in concurrent cause cases and would not do so in this one, the argument is moot. See footnote 33. We discuss the trial court’s predictions about Nebraska law in section IC of this opinion.
. The trial court asserted in chambers that neither side understood Nebraska’s causation standards when each agreed to have the case tried under Nebraska law. The court speculated that the choice of Nebraska law was dictated by the plaintiffs wish to avoid Kansas’ comparative fault doctrine and the defendants’ wish to escape Kansas’ punitive damages doctrine. The appellants’ brief suggests instead that under Kansas’ choice of law rules, Nebraska law applied; i.e., although Menne’s alleged exposure to hazardous levels of asbestos dust occurred in the state of Washington, his alleged injury occurred in Nebraska, thereby invoking Nebraska law. Appellants’ Brief at 25.
. It is undisputed that exposure to, i.e., inhalation of asbestos fibers can cause mesothelioma. Although only about ten percent of those exposed to heavy amounts of asbestos dust develop mesothelioma, R.Vol. 5 at 361, mesothelioma is causally related to inhalation of asbestos fibers in approximately 80% of all cases of the disease. R.Vol. 7 at 1006.
. In some cases, what was judged to be sufficient evidence of actual exposure has been extremely attenuated. See, e.g., Lockwood v. AC & S, Inc., 109 Wash.2d 235, 744 P.2d 605, 610-13 (1987) (inferred presence of defendant’s products on large ship during period that plaintiff worked on it created prima facie case of exposure where evidence was presented that asbestos dust "drifted”); Rocco v. Johns-Manville Corp., 754 F.2d 110, 113 (3d Cir.1985) (evidence of general use of Pittsburgh Coming’s Unibestos at job site was sufficient to allow jury to find Pittsburgh Corning to be a factor in causing the plaintiffs injury).
. In some jurisdictions, mere exposure to unknown quantities of asbestos dust apparently is sufficient to raise a supportable inference of causation. Where such is the case, a relaxed burden of proof appears to be sanctioned, whether or not alternative liability theory is mentioned. See, e.g., Roehling v. Natl Gypsum Co. Gold Bond Bldg. Products, 786 F.2d 1225, 1228 (4th Cir.1986) (evidence of exposure allows a jury to infer injury from defendants’ products); the dicta in Starling v. Seaboard Coast Line R.R., 533 F.Supp. 183, 191 (S.D.Ga.1982) (prima facie case made out if plaintiff can prove exposure to defendant’s products at least on one occasion (perhaps contradicting its earlier rejection of alternative liability theory and arguably misreading the nature of the cumulative injury from asbestosis discussed in Borel v. Fibreboard Paper Products Corp., 493 F.2d at 1094)); Migues v. Fibreboard Corp., 662 F.2d 1182, 1185 (5th Cir. Unit A 1981) (evidence of mere exposure to Nicolet asbestos fibers was sufficient to constitute a producing cause of plaintiffs mesothelio-ma). See also the cases cited in footnote 12.
The testimony in the Menne trial attempts to meet a higher evidentiary standard and is consonant with Maryland’s "substantial causation" approach applied by the Fourth Circuit in Lohrmann v. Pittsburgh Corning Co., 782 F.2d 1156, 1162 (4th Cir.1986) (holding that evidence that a defendant was a substantial factor in producing an asbestos-related injury requires that a plaintiff must "prove more than a casual or minimum contact with the product”). Under Maryland law, evidence of exposure will be sufficient to get to the jury only if it shows the “‘frequency of the use of the product and the regularity or extent of the plaintiffs employment in proximity thereto’ ” (quoting the district court decision in the case). Id. Cf. the use of quantifiable data in Jackson v. Johns-Manville Sales Corp., 727 F.2d 506, 523 (5th Cir.1984), reh’g en banc 750 F.2d 1314 (5th Cir.1985) (relevant portions reinstated, portions superceded); question certified en banc, 757 F.2d 614 (5th Cir.1985), certification declined, 469 So.2d 99 (Miss.1985), reh’g en banc, 781 F.2d 394 (5th Cir.1986), cert. denied, 478 U.S. 1022, 106 S.Ct. 3339, 92 L.Ed.2d 743 (1986). In Jackson, the Fifth Circuit applied Mississippi’s substantial *1463factor test of proximate cause and found that the jury had sufficient evidence to find that Jackson’s injuries were proximately caused by Raybestos-Manhattan and Johns-Manville. The circumstantial evidence of exposure that was judged sufficient was that Raybestos-Man-hattan products were ordered for use on at least seven to ten of the sixty-nine hulls on which the plaintiff worked and that Johns-Manville products were ordered for use on thirty or more of the hulls, allowing estimates of the odds of frequent contact. Co-workers recalled using those products under extremely dusty conditions, which speaks to the intensity of the exposure. See also Flatt v. Johns Manville Sales Corp., 488 F.Supp. 836, 838 (E.D.Tex.1980) (exposure must be sufficient to be a producing cause of mesothelioma).
. In the Nebraska case law there are several lines of cases with seemingly different tests of the sufficiency of circumstantial evidence. The oldest line suggests that necessary inferences from circumstantial evidence must be the only ones that can be drawn from the evidence. A second line of cases requires that the necessary inferences must be able to be drawn with “reasonable certainty.” A newer line suggests that circumstantial evidence is equally competent with direct evidence and that the necessary inferences must reasonably arise from the evidence (a reasonable probability test). See Fenner, Circumstantial Evidence in Nebraska, 19 Creighton L.R. 236-77 (1986). If the evidence here does not meet the "reasonable probability” standard, it certainly would not meet the "reasonable certainty” or "only inference” tests. Fenner reconciles all the tests by concluding that "[w]hat all of these cases mean to say is that when the evidence on an issue is evenly balanced so that there is no way to decide between conflicting interpretations of the evidence except through speculation, guess, conjecture, or chance then no jury question is raised regarding that issue, the court should resolve that issue, as a matter of law, against the party with the burden of proof.” Id. at 274.
. Erie R.R. Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938) requires that a federal court sitting in diversity apply state substantive law. As noted in the Menne decision, 641 F.Supp. at 1434, in the absence of specific guidance from a state's supreme court as to its substantive law, a federal court’s prediction of state law looks to (1) lower state court decisions and state Supreme Court dicta; (2) the lower court ruling in the case, (3) the general rule on the issue, (4) the rule in other states looked to by Nebraska courts when they formulate their own substantive law, and (5) other available *1465legal sources, such as treatises and law review commentaries (paraphrasing Jackson v. Johns-Manville, 781 F.2d 394, 397 (5th Cir.1986)).
. Instructions # 26 and # 8 are more oblique. In pertinent part, #26 states:
"Each defendant is entitled to a fair and separate consideration of its case, ... [b]ut where the wrongful acts of two or more manufacturers combine producing a single, indivisible injury, both are liable for that injury even though there was no common duty, common design or concerted action. Thus, if you find plaintiff was exposed to defendants’ asbestos products and that exposure proximately or legally caused his injuries, you may find for plaintiff and against any one or more of the defendants who fail to establish their products were not a contributing cause of plaintiffs injuries.”
R.Vol. I at tab 37. Instruction #8 refers to a burden shift only if the plaintiff makes out a prima facie case of causation. It reads:
“The burden of proof is on the plaintiff to prove every element of his claim by a preponderance of the evidence. Additionally, and dependent upon certain findings to be made by you, this burden may shift to each defendant to prove that it has not caused or contributed to the plaintiffs alleged injuries or damages.”
Id. As we note in the body of this opinion, taken together with the special interrogatories, these instructions can be interpreted in multiple ways, i.e., as invoking a prima facie case of traditional tort liability, a burden shift to defendants on the element of causation, or perhaps even group causation.
. In a comparable fashion, the requirement that substantially all of the potential defendants be before the court under the alternative liability variant of market-share apportionment satisfies the need to demonstrate the probability that the cause is before the court.
. Since asbestos fibers accumulate in the lungs until they reach a critical mass to trigger the onset of mesothelioma in susceptible persons, any exposures above the accepted safety threshold are harmful in the legal sense (and especially so if other exposures follow).
.In concurrent cause cases involving just two or three wrongdoers, a plaintiff frequently can demonstrate the substantiality of each defendant’s contribution even though the exact proportion of each’s contribution to the single harm may not be ascertainable. As the number of wrongdoers mounts, however, it becomes increasingly difficult to demonstrate each’s substantial contribution to the whole. It is under such circumstances that a burden shift with respect to causation can be usefully employed.
. Section 433B(2) reads:
“Where the tortious conduct of two or more actors has combined to bring about harm to the plaintiff, and one or more of the actors seeks to limit his liability on the ground that the harm is capable of apportionment among them, the burden of proof as to the apportionment is upon each such actor.”
The language of 433B(2) does not make clear whether the referred-to burden shift is with respect to causation or rather requires a plaintiff to demonstrate substantial causation by each defendant but then allows a defendant to escape joint and several liability by proving how the damages should be apportioned. Some jurisdictions have interpreted 433B(2) as invoking an alternative liability burden shift on causation in cases of concurrent cause, especially in multiple automobile collision cases, and the trial court in Menne so interpreted it. Menne, 641 F.Supp. at 1433.
. The term concurrent liability may better reveal the variation of Summers-type alternative liability being invoked. Summers-type alternative liability requires as threshold proof of causation that the cause be among the defendants, whereas concurrent liability, as the term is used here, requires at the threshold simply that more than one defendant each caused some harm.
.The court’s secondary reliance on two important cases from other circuits is misplaced. We conclude that, first, it misinterpreted Borel v. Fibreboard Paper Products Corp., 493 F.2d 1076 (5th Cir.1973), cert. denied, 419 U.S. 869, 95 S.Ct. 127, 42 L.Ed.2d 107 (1974), by inferring that Borel had invoked alternative liability. Menne, 641 F.Supp. at 1434. The evidence in Borel of exposure to the products of “all” defendants on “many” occasions allowed the plaintiff to meet his ordinary burden of proof on the element of causation, using a substantial factor test. Id. at 1094. The only burden shift discussed in Borel was with respect to apportionment of damages, not causation; if defendants could show how damages could be apportioned, they would be relieved of joint and several liability. Id. Second, the court quoted extensively from and analogized to a case imposing a burden shift coupled with market share apportionment rather than joint and several liability. In re “Agent Orange” Product Liability Litigation, 597 F.Supp. 740 (E.D.N.Y.1984 as Modified), aff'd, 818 F.2d 145 (2d Cir.1987). The justification for the burden shift in the class action Agent Orange case cannot be separated from the invocation of market share apportionment, which is not available here.
. Although Osteen suggests that Nebraska might adopt a substantial factor test and eliminate but-for causation in Menne’s case, the Men-ne court instructed the jury that but-for causation continued to be part of Nebraska’s requirements for proximate cause. See footnote 7. When it attempted to invoke a form of alternative liability, it in effect predicted that but-for causation would no longer be required, creating an inconsistency in the jury instructions.
. In a related set of cases, a federal magistrate declined to predict that Nebraska would adopt a market share theory of liability such as that adopted in Sindell v. Abbott Laboratories, et al., 26 Cal.3d 588, 163 Cal.Rptr. 132, 607 P.2d 924 (Cal.1980), cert. denied, E.R. Squibb & Sons, Inc. v. Sindell, 449 U.S. 912, 101 S.Ct. 285, 66 L.Ed.2d 140 (1980). Osteen v. Combustion Engineering, CV79-0-47 (D.Neb. August 25, 1981) and its companion case, Hohl v. Combustion Engineering, CV80-0-40 (D.Neb. August 25, 1981). For the reasons discussed therein and other reasons, we too would decline to predict market share or apportioned liability. We do not, however, find the ruling to be directly relevant to the question of a burden shift under a theory of concurrent liability.
.We note that, in this case, only asbestos dust was thought to have caused Menne’s mesothelio-ma. Although Menne was a heavy smoker, evidence established that cigarette smoking does not contribute to the onset of mesothelioma, unlike the situation with asbestosis or lung cancer. R.Vol. 5 at 359-60; R.Vol. 7 at 1006-07. We express no opinion as to whether a burden shift should be adopted where both asbestos and nonasbestos factors are alleged to have contributed to a single injury.
. Apportionment of damages among multiple employers was expressly repudiated as a judicial remedy by the Nebraska Supreme Court in Osteen v. A.C. and S., Inc., 209 Neb. 282, 307 N.W.2d 514. The Osteen court suggested that the factors to be used as a basis for apportionment must be adopted by the Nebraska legislature and not by the courts, at least in situations where the onset of an occupational disease and its successive aggravation from subsequent exposures cannot be determined. Id. 307 N.W.2d at 519.
. Nebraska has only a limited comparative negligence statute, Neb.Rev.Stat. § 25-21, 185 (Reissue of 1985), which requires a comparison of levels of negligence between plaintiff and defendant. Where the defendant is liable, the statute provides that the plaintiffs recovery may be reduced in proportion to the negligence attributable to the plaintiff. There is no statutory provision for assessment of comparative fault or apportionment of damages among negligent defendants.
. The joint and several liability attaching in traditional concurrent cause cases is arguably fair because the fact that not all wrongdoers are before the court does not negate the proven "but-for" and substantial contribution of those who are. Such is not the case here.
. Arguably, such a conclusion moves one step beyond § 433B(2) of the Restatement. Rather than allowing a liable defendant to escape joint and several liability by proving how damages should be apportioned, it would also remove the possibility of several (total) liability where not only does the plaintiff lack evidence of substantial causation but also the defendants lack sufficient data with which to apportion liability.
. Of course, it might balance the equities among wrongdoers better here if damages were to be pro-rated among (or comparative liability assessed against) all known Shipyard suppliers, including those not before the court. Neither Nebraska common law nor statutory law, however, provides a basis for extending any predictions to such a degree. See, e.g., footnote 27. If a similar case should arise in Nebraska, or if causation questions should be certified on retrial, it may be that Nebraska would decide to take steps to balance the equities in a manner that we are unable realistically to predict.
. The special interrogatories to the jury asked:
“1. Do you find one or more of the defendants supplied asbestos-containing products at plaintiffs workplace?
2.Do you find that plaintiff was exposed to quantities of asbestos dust from any of defendants’ products at his workplace?
3. Did plaintiff sustain mesothelioma and/or asbestosis, which were substantially caused or contributed to by his exposure to asbestos dust from any of the defendants’ products at his workplace?
4. Do you find any of the defendants to be at fault and liable for plaintiffs injuries under either strict liability or negligence?
5. If you answered "yes” to all of the previous four interrogatories, identify which of the defendants listed below are at fault:”
[each defendant listed separately with spaces next to each defendant to check yes or no]
R.Vol. 1 at tab 38 (emphasis added).
.See the text of this opinion at 1465.
. If the instructions are read as allowing for group causation, Menne’s evidence did not show that collectively the exposures reached a level of significance. He presented insufficient evidence as to frequency of exposure to the products of each defendant, and adding unknown levels of frequency from three or four defendants does not sum to known levels. For group causation to have worked here, all or virtually all possible defendants would need to have been before the court. Then their unknown but presumably insignificant individual contributions might have been ascertained collectively to reach substantiality.
. Best "technology” reasonably available cannot mean simply the technological feasibility of giving a warning independent of the need for the warning, since warnings are virtually always technologically feasible. The question is not whether a warning is feasible but whether the manufacturer knew or should have known that the generally recognized and prevailing best scientific and medical evidence reasonably available at the time was finding an association between mesothelioma and inhalation of asbestos dust. If so, then users deserved to know about it.
. It appears in an expanded context, following on the heels of a comment that “[a] jury may find that a warning is inadequate and unreasonable even where the existence of a ‘risk,’ i.e., a causal relationship between use of the product *1473and resulting injury, has not been definitely established.” Seley, 423 N.E.2d at 837.
In the light of conflicting Ohio cases, the 6th Circuit has concluded that Seley’s invocation of a strict liability duty to warn is limited to certain drugs or other products that are "incapable of being made safe for their intended and ordinary use.” Overbee v. Van Waters & Rogers, 706 F.2d 768, 770 (6th Cir.1983).
. Contrary to Menne’s assertion, Symons v. Mueller Co., 493 F.2d 972 (10th Cir.1974) did not approve the same strict liability policy justification statement. We simply noted in Symons that the judge had instructed the jury regarding *1474the policy behind strict liability. Id. at 974. Our reading of Symons leads us to conclude that the propriety of that portion of the instruction was not at issue.
. We imply no criticism of the district court. The parties themselves were uncertain about Nebraska tort standards, even though they stipulated shortly before trial that the case would be tried under Nebraska law. On short notice and without benefit of briefing by the parties, the district performed admirably in attempting to determine what standards would govern.
. The parties’ stipulation that Nebraska law applies governs on retrial. And it should go without saying that the law of this case, as established by our rulings, would be superceded by any contrary rulings on Nebraska law by the Nebraska Supreme Court.