OPINION OF THE COURT
SCIRICA, Circuit Judge.
In this diversity case involving asbestosis, the district court declined to charge the jury on the apportionment of decedent’s damages between cigarette smoking and asbestos exposure. Because we predict that the Pennsylvania Supreme Court would not find a reasonable basis for apportionment in this case, we will affirm.
I.
Richard Borman worked as an insulator at various locations and with various employers from 1956 to 1987, when he became unable to work. During the first thirteen years, he came in daily contact with asbestos products manufactured by the appellant, Celotex. He also smoked over a pack of cigarettes a day for thirty-five years (1950-1985). In 1985, he was diagnosed as having asbestosis, and in 1988, lung cancer. In that same year, he died of lung cancer.
On February 11, 1987, he and his wife filed suit in district court against several manufacturers of asbestos products, including Celotex. The complaint alleges that occupational exposure to these products caused Mr. Borman’s disability1 and [329]*329that defendants should be strictly liable for failure to warn of the products’ dangerous propensities.2 Before trial, several defendants, later joined by Celotex, asked the trial judge to “instruct the jury to allow apportionment of the harm alleged by the plaintiff between that caused by cigarette smoking and that caused by exposure to asbestos.” The motion was taken under advisement.
On August 15, 1989, trial commenced on a reverse bifurcated basis against the non-settling defendants, including Celotex. At the conclusion of testimony, the trial judge found that there was no reasonable basis for apportionment of damages,3 and denied the request for charge.4 Subsequently, the jury returned a verdict on damages, awarding $532,719 to Mrs. Borman and the estate of her husband. Later, the jury returned a verdict on liability against Celotex but exonerated the remaining defendant.
Celotex moved for a directed verdict and/or j.n.o.v., and a new trial, asserting inter alia that the evidence warranted an apportionment of damages charge and did not support damages for lost wages. The trial court denied this motion in an order without opinion. Celotex appealed, challenging the court’s failure to charge on apportionment of damages and to direct a verdict on the issue of wage loss.5
We have jurisdiction under 28 U.S.C. § 1291 (1988).6 We will affirm the district court’s denial of j.n.o.v. unless the record is “ ‘critically deficient of that minimum quantum of evidence from which the jury might reasonably afford relief.’ ” Honeywell, Inc. v. American Standards Testing Bureau, Inc., 851 F.2d 652, 654 (3d Cir.1988) (citations omitted), cert. denied, 488 U.S. 1010, 109 S.Ct. 795, 102 L.Ed.2d 787 (1989). We will review the denial of a motion for new trial only for abuse of discretion unless, as in this case, the motion was based on the application of a legal precept, in which case our review is plenary. Id. As a federal court sitting in diversity, we look to state law. Our review of the district court’s interpretation of state law is plenary, Compagnie des Bauxites de Guiñee v. Insurance Co. of North America, 724 F.2d 369, 371 (3d Cir.1983).
II.
Celotex’s primary argument concerns apportionment of damages. According to Celotex, the evidence was sufficient to require the trial court to charge the jury on apportionment of damages between asbes[330]*330tos exposure and tobacco consumption. Our first step, therefore, is to review the expert testimony presented at trial.
Mrs. Borman called Dr. Daniel DuPont, a specialist in pulmonary medicine. On direct examination, Dr. DuPont testified that Mr. Borman’s “exposure to asbestos dust was a substantial contributing factor to” both his asbestosis and his lung cancer. On cross-examination, Dr. DuPont testified that “Mr. Borman’s cigarette smoking history [was] a substantial contributing factor to the development of his lung cancer.”
Much of the cross-examination of Dr. DuPont, however, focused on the increased risk of lung cancer caused by tobacco smoking and asbestos inhalation:
Q And what is [the risk of developing lung cancer for persons with a significant smoking history] as compared with the non-smokers and ... the non-asbestos exposed folks in the general population?
A Up to 12 — up to 12 to 15 times the group you compared, which is known as the background group.
Q The background group is — that’s the so-called clean livers, the people who aren’t exposed to asbestos and who didn’t smoke; right?
A Correct.
* * * * * *
Q Doctor, can you tell the jury what the risk of an occupationally exposed worker who was not a cigarette smoker, what his risk is of getting lung cancer as compared to the background population of non-smokers, non-asbestos exposed people?
A Yes.
Q What is it?
A Up to five to six times the background incidence.
Q So would you agree with me, therefore, Doctor, that taken alone, cigarette smoking is two or three times more likely to cause a lung cancer than asbestos is, taken alone?
A In the interpreted statistically, and that is, in looking at a large number of people or a body of people, which the rule of epidemiology, that that’s what the statistics would indicate, yes.
Not in the particular case, however. As I previously testified, under any one individual, which is different than the information that you have reviewed, in one individual that any one can give us an assignment of causation.
Q Well, Doctor, along those lines, since you’ve testified that — on the one hand you’ve given us these numbers, on the one hand you said that you agree with me that smoking alone is two or three times as likely to cause lung cancer as asbestos alone — why should this jury not assume that in this case, cigarette smoking was two or three times as responsible, under your analysis, for the lung cancer as asbestos was?
A Among other reasons, you haven’t gotten into the statistics as to the risk of lung cancer in the population that Mr. Borman would fit. And that would be the population of people who have combined asbestos exposure and tobacco consumption. And those factors are very impressive.
Those factors, just to complete the statement, refer to synergistic effect. And that is, if you take the 12 to 15 up to — I want to make that clear for the record — up to 12 to 15 fold increase of cigarette smoking alone. And the up to five to six fold increase of asbestos exposure alone and you put the two together and you take the statistics in that area, which is really the only group that applies to Mr. Borman, that risk factor is multiplied and therefore, the risk factor is up to 60 to 90 times the background incidence.
Despite the extensive evidence on likelihood of causation, Dr. DuPont emphasized that he could not apportion the damages:
Q ....
Now Doctor, can you tell us which of these two factors was more responsible for the development of his lung cancer? A Not in Mr. Borman’s case, I can’t.
Q You can’t do that at all?
A Well, you’re taking data applying to 100,000 people and asking me to refine [331]*331that to one individual. As I previously testified ... I don’t know that you can, in one individual, give an assignment as to what the risk factor was or proportion which of the two significant risk factors was more or what percentage there was. BY MR. MCDAVID:
Q All you can say is that asbestos and cigarette smoking were each greater than zero percent and less than 100 percent responsible; is that right?
A Each factor — this patient[’]s lung cancer, which was the cause of his death, was caused to a substantial degree by asbestos exposure and tobacco consumption.
Q Well, would you agree with me that each factor must have been more than zero and less than 100 percent?
A Yeah, I guess I would agree with that.
Q Can you tell us anything other than that about the percentage?
A I don’t believe that, speaking on one individual, and narrowing those statistics down to one person, that I could.
The defense put on its own expert, Dr. William Weiss, also a specialist in pulmonary medicine.7 Dr. Weiss testified that asbestos exposure “did not cause [Mr. Bor-man’s] lung cancer.” On the contrary, he said only smoking “caused the lung cancer in Mr. Borman’s case.” Dr. Weiss was not asked to apportion damages.
A.
The parties agree that Pennsylvania law applies on the apportionment of damages. Before looking at specific Pennsylvania law, however, some background is appropriate. Certain types of harm “are normally incapable of any logical, reasonable, or practical division.” Restatement (Second) of Torts § 433A comment i (1965). Such harms include death, a broken limb and any single wound. Id. “By far the greater number of personal injuries ... are ... normally single and indivisible.” Id. In this case, although Mr. Borman suffered numerous diseases, his disability is the sole harm for which recovery is sought. Martin v. Johns-Manville Corp., 349 Pa.Super. 46, 56, 502 A.2d 1264, 1269 (1985) (Martin I) (“While it is apparent from the testimony at trial that appellant suffered separate harms — chronic bronchitis, emphysema and asbestosis — the ‘harm’ for which he sued and as to which the jury found him entitled to recover was his disability”), rev’d on other grounds sub nom. Martin v. Owens-Corning Fiberglas Corp., 515 Pa. 377, 528 A.2d 947 (1987) (Martin II).
While a single harm may resist apportionment, apportionment can nonetheless be invoked when the particular contribution of multiple causes can be determined on a factual basis. Restatement (Second) of Torts § 433A. The procedure remains equally viable when the plaintiff’s conduct, even if innocent, was a substantial cause of his harm. Id. comment a. In both cases, apportionment promotes fairness by “limiting] a defendant’s liability to that part of the harm of which that defendant’s conduct has been a cause in fact.” See Prosser and Keeton on Torts § 52, at 345 (5th ed. 1984). Apportionment can be harsh to the plaintiff, however, because it shifts to him “the risk of financial irresponsibility of each wrongdoer.” Id. at 351.
B.
A federal court exercising diversity jurisdiction must apply state law as declared by the highest state court. See Erie R.R. v. Tompkins, 304 U.S. 64, 71-80, 58 S.Ct. 817, 818-23, 82 L.Ed. 1188 (1938); Gruber v. Owens-Illinois Inc., 899 F.2d 1366, 1369 (3d Cir.1990). When the state’s highest court has not addressed the issue, the federal court must predict its holding. See Commissioner of Internal Revenue v. Estate of Bosch, 387 U.S. 456, 465, 87 S.Ct. 1776, 1782, 18 L.Ed.2d 886 (1967); Gruber, 899 F.2d at 1369-70. As we shall see, resolution of this matter is somewhat complicated because the Pennsylvania Supreme [332]*332Court has addressed the issue but only in its plurality opinion in Martin II.
“The rules in [Pennsylvania] governing apportionment of damages are consistent with those expressed in the Restatement (Second) of Torts." Martin II, 515 Pa. at 381, 528 A.2d at 949. At the close of evidence, “[t]he trial court must determine, as a matter of law, whether the harm is capable of apportionment.” Id. at 382, 528 A.2d at 949 (citing inter alia Restatement (Second) of Torts § 434(l)(b)). The following provision controls this determination:
(1) Damages for harm are to be apportioned among two or more causes where
(a) there are distinct harms, or
(b) there is a reasonable basis for determining the contribution of each cause to a single harm.
(2) Damages for any other harm cannot be apportioned among two or more causes.
Restatement (Second) of Torts § 433A (1965), quoted in Martin II, 515 Pa. at 381, 528 A.2d at 949. “[T]he burden of proving apportionment rests on the party seeking it.” Martin II, 515 Pa. at 382, 528 A.2d at 949 (citing Restatement (Second) of Torts § 433B(2)). “Once it is determined that the harm is capable of being apportioned, the actual apportionment of the damages among the various causes is a question of fact, which is to be determined by the jury....” Restatement (Second) of Torts § 434 comment d.
This case implicates § 433A(l)(b)— whether “there is a reasonable basis for determining the contribution of each cause to a single harm.” Because we apply Pennsylvania law, resolution of this issue hinges on one case, Martin II. Martin, an insulator for thirty-nine years and a cigarette smoker for approximately thirty-seven years (two packs a day over the last eighteen years), sought damages for pulmonary disability that he claimed was caused by working with asbestos-containing products manufactured by the defendants. With respect to apportionment, the following evidence was adduced:
1. Plaintiff’s experts “testified that his disability was due to both asbestosis, caused by asbestos exposure, and emphysema, caused by [his] long-term cigarette smoking and aggravated by asbestos exposure and the resulting asbestosis.” Martin II, 515 Pa. at 380, 528 A.2d at 948.
2. Plaintiffs experts could not assign a percentage of contribution to either cause. Id. at 383-84, 528 A.2d at 950.
3. Defendants’ experts testified that “[plaintiff’s] disability was solely the result of emphysema, caused by cigarette smoking.” Id. at 380, 528 A.2d at 948.
In addition, the evidence demonstrated bullous changes associated with cigarette smoking in the upper portions of the lungs and pleural plaques consistent with asbestos inhalation, in the middle and lower zones of the lungs. Id. at 388-91, 528 A.2d at 952-53. On this evidence, the trial judge instructed the jury that it could apportion damages for plaintiff’s disability between asbestos inhalation and cigarette smoking. See id. at 380, 528 A.2d at 948.
On appeal, the Pennsylvania Superior Court affirmed, holding the evidence “sufficient to enable the jury to reach a reasonable approximation of the harm due to each cause.” Martin I, 349 Pa.Super. at 51, 502 A.2d at 1267. Writing for the court, Judge Spaeth reasoned that the injuries to the upper and lower lobes and the different periods of asbestos and tobacco inhalation showed that “the causes of the harm were ... distinct and capable of rough approximation.” Id. at 59, 502 A.2d at 1271. He relied particularly on the principle that “damages may be apportioned if the plaintiff’s conduct, even though innocent, has contributed to the harm. Thus, the question whether appellant was negligent in smoking cigarettes is irrelevant to deciding whether his damages may be reduced because of the part that smoking played in his disability.” Id. at 56-57, 502 A.2d at 1270 (paraphrasing Restatement (Second) of Torts § 433A comment a).
[333]*333In a plurality opinion,8 the Pennsylvania Supreme Court reversed and remanded for a new trial, holding the evidence on apportionment of damages too speculative to support submission of the issue to the jury. Martin II, 515 Pa. at 384-85, 528 A.2d at 950. As Justice Larsen explained,
[T]he jury cannot be expected to draw conclusions which medical experts, relying on the same evidence, could not draw. The causes of disability in this case do not lend themselves to separation by lay-persons on any reasonable basis. Thus, common sense and common experience possessed by a jury do not serve as substitutes for expert guidance, and it follows that any apportionment by the jury in this case was a result of speculation and conjecture and hence, improper. “Rough approximation” is no substitute for justice.
Id., 528 A.2d at 950 (footnotes omitted).
In his concurring opinion, however, Justice McDermott made clear that Martin II would not preclude apportionment in the appropriate asbestos case:
The Majority Opinion stands for a single proposition, i.e., under the facts and circumstances of this case there was not enough evidence to submit the issue of apportionment to the jury. With this I can agree.
However, this case should not be construed as standing for the proposition that evidence of contributory negligence is inadmissible in an asbestosis case. Furthermore, it should not be interpreted as precluding the defendants in this case from introducing new evidence of decedent’s negligence, nor precluding a jury from returning a lesser verdict if the evidence would support such.
Id. at 386, 528 A.2d at 951.
Because Justice Larsen’s opinion does not represent a majority view, it is not considered controlling precedent under Pennsylvania law. Decatur Contracting v. Belin, Belin & Naddeo, 898 F.2d 339, 344 (3d Cir.1990) (quoting Vargus v. Pitman Mfg. Co., 675 F.2d 73, 75 (3d Cir.1982)); McGowan v. University of Scranton, 759 F.2d 287, 293 (3d Cir.1985). Therefore, “without an authoritative announcement, we still must predict how the Pennsylvania Supreme Court would rule on this issue.” Decatur Contracting, 898 F.2d at 344 (citation omitted).9
C.
To determine whether the Pennsylvania Supreme Court would find the evidence in this case too speculative to support submission of the apportionment issue to the jury, we must compare the evidence presented in this case with that adduced in Martin. The evidence is remarkably alike. First, the decedents have similar histories of asbestos and tobacco consumption. Martin was an asbestos worker for thirty-nine years; he smoked cigarettes for about thirty-seven years, two packs a day during the last eighteen years. Mr. Borman worked with asbestos insulation for thirteen years; he smoked cigarettes for thirty-five years, over a pack a day. Second, the experts espouse the same diagnostic framework. Pulmonary disease may be restrictive (the lung cannot fully expand) or obstructive (the airway passages are blocked). Asbestosis is a restrictive disease, although the plaintiffs’ experts testified that it also has an obstructive component. Emphysema and chronic bronchitis are obstructive diseases. Furthermore, asbestosis is caused by asbestos exposure, and emphysema and bronchitis are caused by cigarette smoking.10 Third, the decedents exhibited similar symptoms. Both developed pleural plaques, which are [334]*334caused by asbestos inhalation but not by cigarette smoking. In addition, both had chronic obstructive pulmonary disease, revealed in Martin as bullous changes on x-rays and in Mr. Borman as diminished exhalation speed in spirograms. Fourth, the experts drew the same conclusions regarding decedents’ disabilities. While the plaintiffs’ experts concluded that the decedents had asbestosis, all experts agreed that the plaintiffs also had emphysema and bronchitis.11 Thus, the plaintiffs’ experts concluded that asbestos exposure and tobacco inhalation were both substantial causes of the harm while the defendants’ experts asserted that cigarette smoking was the sole cause of the harm. The plaintiffs’ experts also testified that, in the individual case, they could not apportion cause between the two factors; the defendants’ experts, however, did not testify on the issue.
Celotex argues with considerable plausibility that despite the inability of the experts to assign a percentage of contribution to each cause, the evidence provided the jury with a reasonable basis for apportioning damages between the two causes in this case. The jury heard testimony from the plaintiff’s expert regarding the risk of developing lung cancer from cigarette smoking, asbestos exposure or both, and the extent to which the distinctive effects of both asbestos exposure and cigarette smoking were evident in the lungs of Mr. Borman. The jury was aware of the length of time during which Mr. Borman had worked with asbestos and smoked cigarettes. Therefore, Celotex contends that despite the fact that the experts could not calculate the exact proportion that each factor contributed to Mr. Borman’s disability, the causes of the harm were distinct and capable of rough approximation. Thus, Celotex argues that where a factual basis exists, it is preferable in the interest of fairness to permit some rough apportionment of damages, rather than to hold the defendant entirely liable for a harm that was inflicted by separate causes. See Prosser and Keeton, supra, at 345.12
However, we reiterate that we are not free to treat this issue as if it were a matter of first impression in our court. We recognize that Martin II is not controlling. Because the facts of this case mirror those of Martin II, however, we find the limited holding expressed by Justice McDermott — “under the facts and circumstances of this case there was not enough evidence to submit the issue of apportionment to the jury” — critical in predicting how the Pennsylvania Supreme Court would decide this issue. On the facts, we cannot justifiably distinguish this case from Martin. Therefore, based on Martin II, we predict that the Pennsylvania Supreme Court would hold this evidence insufficient to support a charge on apportionment of damages.13
Nor does the statistical evidence presented by Borman’s expert in this case concerning the increased risk of lung cancer associated with cigarette smoking, asbestos inhalation or both, change our conclusion. Although the opinions do not reveal whether such evidence was presented in Martin, Celotex contends that it was not. The question is whether this additional statistical evidence distinguishes this case from Martin and gives the jury a reasonable basis for apportionment.
[335]*335Although the Restatement (Second) requires only a “reasonable basis for determining the contribution of each cause to a single harm” and Justice McDermott’s concurrence contemplates evidence sufficient to support apportionment of damages, we do not believe the Pennsylvania Supreme Court would hold the increased risk evidence sufficient to create a reasonable basis for apportionment. This evidence speaks to the likelihood that either cigarette smoking, asbestos exposure or both will cause cancer, rather than to the apportionment of damages between two causes. Thus, based on this evidence, the plaintiffs expert concluded that cigarette smoking and asbestos exposure were each substantial factors in causing Mr. Borman’s lung cancer. However, the expert was unwilling to rely on this evidence in the individual case as a basis for apportioning cause. Moreover, even assuming that this detailed statistical evidence was not presented in Martin, the jury was informed that “ ‘many people think [cigarette smoking] enhances the genesis of asbestosis. That is it enhances the fibrosis of asbestosis and certainly enhances the associated complication of asbestosis.’ ” Martin I, 349 Pa.Super. at 54, 55, 502 A.2d at 1269. Nonetheless, the Pennsylvania Supreme Court held that there was insufficient evidence to support an apportionment. We find nothing in the plurality or concurring opinions to indicate that the Pennsylvania Supreme Court would reach a different result on the apportionment issue if faced with this additional evidence.
In our research, we examined four cases that have cited Martin II and found no conflict with Martin IPs conclusions. Of these four cases, the most analogous is Taylor v. Celotex Corp., 393 Pa.Super. 566, 574 A.2d 1084 (1990). In that case, as here, the plaintiff was a heavy smoker who had been exposed to asbestos. He was diagnosed as suffering from asbestosis, hypertension, arteriosclerotic heart disease, and emphysema. Defense experts testified that Taylor did not have asbestosis, and neither plaintiffs’ experts nor defendants’ experts attempted to apportion the cause of Taylor’s illness among cigarette smoking, asbestos exposure, and heart disease. Id. at 592-93, 574 A.2d at 1097-98. The Superior Court, following Martin II, affirmed the trial court’s refusal to instruct the jury on apportionment. Id. at 593, 574 A.2d at 1098.14
III.
We now turn to Celotex’s second argument, concerning damages for lost wages. According to Celotex, the only evidence adduced on the cost of Mr. Borman’s personal maintenance is the following testimony of his wife:
Q Before he retired, did — and while he was working full-time and including the overtime, did he take any money for personal maintenance purposes on a weekly basis?
A Yeah.
Q Can you give the ladies and gentlemen of the jury an approximate amount of what he—
[336]*336A He — approximately 20, $25 a week for whatever he wanted.
Mrs. Borman does not cite additional evidence.
Asserting that this evidence — by which the wage loss claim would be reduced— invited speculation as to the cost of Mr. Borman’s personal maintenance, the non-settling defendants moved “that the wage loss claim be precluded.” 15 The district judge denied this motion. After trial, Celo-tex moved unsuccessfully for a directed verdict on the issue. Celotex now challenges the post-trial order.
Under Pennsylvania law, “[l]oss of future earnings is a distinct item of damages, which if properly proved, may result in recovery for the plaintiff.” Kaczkowski v. Bolubasz, 491 Pa. 561, 566, 421 A.2d 1027, 1029-30 (1980) (footnote omitted). If an injured plaintiff dies, “the proper measure of damages includes a deduction based upon decedent’s cost of personal maintenance.” Id. at 566 n. 7, 421 A.2d at 1030 n. 7. The cost of personal maintenance does not encompass “only those expenditures essential to the barest survival.” McClinton v. White, 497 Pa. 610, 617, 444 A.2d 85, 89 (1982). Rather, it includes the “ ‘sum which a decedent would be expected to spend, based on his station in life, for food, clothing, shelter, medical attention and some recreation.’ ” Id., 444 A.2d at 89 (citation omitted).
Celotex does not argue that the amount of maintenance costs should be increased, but only that inadequacy of the proof of maintenance defeats the entire future earnings claim. However, it cites no case in which a Pennsylvania court refused to allow the jury to consider a lost earnings claim for this reason. In any event, the evidence presented by plaintiff in this case is more concrete than that presented in McClinton. Although the evidence on the cost of personal maintenance is spare, it is sufficient. Because damages need not be proven with mathematical exactness, we will affirm.
IV.
For the foregoing reasons, we will affirm the order of the district court.
Each side to bear its own costs.