Orchard View Farms, Inc. v. Martin Marietta Aluminum, Inc.

OPINION BURNS, Chief Judge: HISTORY OF THIS CASE This diversity case is before the court on remand from the Ninth Circuit Court of Appeals for a retrial on the issue of puni *986 tive damages. Orchard View Farms, Inc. v. Martin Marietta Aluminum, Inc., No. 73 3080 (9th Cir. June 23, 1975). On March 31, 1971, Orchard View Farms, Inc. (Orchard View) filed this trespass action, seeking compensatory and punitive damages for injuries to its orchards between March 31, 1965 and the filing date. These injuries were alleged to have been caused by fluoride emitted from the aluminum reduction plant operated by Martin Marietta Aluminum, Inc. (the company or Martin Marietta). In April and May, 1973, the case was tried to a jury, which awarded Orchard View $103,655 compensatory damages and $250,000 punitive damages. The company appealed this judgment on numerous grounds. The Ninth Circuit affirmed the award of compensatory damages but reversed and remanded the punitive damages award because in various rulings at the trial I erroneously admitted evidence of certain events that had occurred before the 1965-71 claim period, events which had been insufficiently linked by the evidence to the company’s conduct and policies during the claim period. Upon retrial, evidence was presented in various forms. Much testimony was presented through written witness statements. Additional testimony, both on direct and cross-examination, was offered live. Most of the evidence was submitted in October, 1977, but a defense witness and the plaintiff’s responding rebuttal witness testified in 1978, and final arguments were heard in 1979. Both sides submitted pretrial and post-trial memoranda. Both parties and their counsel are to be complimented for their cooperation with each other and with the court, their observance of the local procedural rules, and the quality of their work. FACTUAL BACKGROUND Martin Marietta Aluminum, Inc., is a California corporation that owns and operates aluminum reduction plants, including plants located in The Dalles, Oregon, and Golden-dale, Washington. Harvey Aluminum, Inc. (Harvey) constructed the plant located at The Dalles, and owned and operated it when production began in 1958. In 1968 Martin Marietta Corporation purchased a controlling share of Harvey common stock. It voted its representatives into a majority of the Harvey directorship in 1969. In 1972, the name of Harvey Aluminum, Inc., was changed to Martin Marietta Aluminum, Inc. In 1974 Martin Marietta Aluminum, Inc., became a wholly-owned subsidiary of Martin Marietta Corporation. Orchard View Farms, Inc., is an Oregon corporation. It operates three orchards with a combined total acreage of approximately five hundred acres. The orchards are located between 2.5 and 5 miles from the aluminum plant. Donald Bailey is the president and treasurer of Orchard View; he, his wife, and five of their children are the sole stockholders. Orchard View owns part of the land; the Baileys own the rest and lease it to Orchard View. The Bailey family operates the orchards. This case is one of an ever-increasing number filed against Harvey, and later Martin Marietta, by orchardists who charged that fluorides emitted from the plant have damaged their crops. The first such suit, Renken v. Harvey Aluminum, Inc., 226 F.Supp. 169 (D.Or.1963), was filed in May, 1961. It was finally closed in 1966 when the court approved a consent decree providing for arbitration of the growers’ claims and dismissal of the related actions filed in state court during the interim. Since February, 1977, thirteen actions have been filed in the United States District Court. These suits seek compensatory and punitive damages for injury allegedly inflicted by emissions from the plant during the years 1971 through 1977. OPINION I. OREGON LAW OF PUNITIVE DAMAGES. II. SOCIETAL OBLIGATIONS OF BUSINESS ENTERPRISES. III. EVALUATION OF THE DEFENDANT'S CONDUCT IN LIGHT OF ITS SOCIETAL OBLIGATIONS. A. Ascertaining the Harm from Plant Emissions. 1. The Company's Efforts. a. Cherries. b. Peaches. c. Pines. d. Other Vegetation. *987 2. Efforts of Others. a. Cherries. 1) Observation of Trees, Leaves auu Fruit. 2) Scientific Inquiry. b. Peaches. c. Pines. d. Other Vegetation. 3. The Company's Response to Evidence of Harm from the Plant's Emissions. a. Cherries. b. Peaches. c. Pines. d. Other Vegetation. 4. Conclusion. B. Efficiently Controlling the Harmful Emissions. 1. Plant Siting. 2. Monitoring. a. Emissions. 1) Frequency of Monitoring. 2) Accuracy of Monitoring. a) Selection of Monitoring Location. b) Duration of Tests. c) Maintenance of Normal Operating Conditions. d) Sampling Technique. 3) Overall Evaluation. b. Ambient Concentrations. 3. Controlling Emissions. a. Fluoride Evolution at the Cell. 1) Operating Parameters. 2) Selection of Ore. b. Primary Emission Control System. 1) Collection Efficiency. a) Cell Hooding. b) Operating Procedures. 2) Treatment Efficiency. c. Secondary Emission Control System. 1) Collection Efficiency. 2) Treatment Efficiency. d. Overall Performance. 4. Mitigation Measures. a. Tall Stacks. b. Application of Lime Spray. 5. Conclusion. C. Arranging to Compensate for the Remaining Harm. IV. AWARD OF PUNITIVE DAMAGES. I. OREGON LAW OF PUNITIVE DAMAGES. The Oregon Supreme Court has provided specific guidance on punitive damage liability in the context of industrial air pollution. In McElwain v. Georgia-Pacific Corporation, 245 Or. 247, 421 P.2d 957 (1966), the court stated: Although this court has on occasion indulged in the dictum that punitive damages are not “favored in the law,” it has, nevertheless, uniformly sanctioned the recovery of punitive damages whenever there was evidence of a wrongful act done intentionally, with knowledge that it would cause harm to a particular person or persons.... The intentional disregard of the interest of another is the equivalent of legal malice, and justifies punitive damages for trespass. Where there is proof of an intentional, unjustifiable infliction of harm with deliberate disregard of the social consequences, the question of award of punitive damages is for the jury. 245 Or. at 249, 421 P.2d at 958. The court reversed the trial judge’s withdrawal of the issue of punitive damage liability from the jury because the “defendant knew when it decided to construct its kraft.mill in Toledo, that there was danger, if not a probability, that the mill would cause damage to adjoining property,” 245 Or. at 250, 421 P.2d at 958, and because of “the substantial evidence from which the jury could have found that during the period involved in this action the defendant had not done everything reasonably possible to eliminate or minimize the damage to adjoining properties by its mill.” 245 Or. at 252, 421 P.2d at 959. In Davis v. Georgia-Pacific Corporation, 251 Or. 239, 445 P.2d 481 (1968), the court remanded a jury’s award of punitive damages because the trial judge had refused to admit evidence pertaining to “the utility of defendant’s operations and its efforts, as compared with others similarly engaged, to prevent damage to surrounding properties.” 251 Or. at 245, 445 P.2d at 484. In a more recent review of punitive damages liability in Oregon, the court in Harrell v. Travelers Indemnity Company, 279 Or. 199, 567 P.2d 1013 (1977), noted: One whose business involves the operation of a plant which emits smoke, fumes or “particulates” may also have ... liability for punitive damages, even in the absence of any “wanton” or “fraudulent” conduct, upon the ground that he has “intentionally” permitted fumes, smoke or particles to be released and blown by the wind upon another’s property, for the reason that “[t]he intentional disregard *988 of the interest of another is the legal equivalent of legal malice and justifies punitive damages for trespass.” 279 Or. at 210-11, 567 P.2d at 1018, quoting McElwain, 245 Or. at 249, 421 P.2d at 958 (and adding an additional “legal” to the quotation). Additional guidance, though of less precedential value, is provided by Reynolds Metals Company v. Lampert, 316 F.2d 272 (9th Cir. 1963). The District of Oregon trial judge had withdrawn the issue of punitive damages liability from the jury. The Ninth Circuit reversed and remanded. Where there is evidence that the injury was done maliciously or wilfully and wantonly or committed with bad motive or recklessly so as to imply a disregard of social obligations, punitive damages are justified. Fisher v. Carlin, 219 Or. 159, 346 P.2d 641. Here the record discloses that appellants had known for several years that fluorides from their plant were settling on appellees’ land, with resultant damage to appellees’ crops. It thus could have been found that their trespass was done knowingly and wilfully, that it was intentional and in wanton disregard of appellants’ social obligations. To justify an award of punitive damages, it is not necessary that the act have been done maliciously or with bad motive. Where it has become apparent, as it has here, that compensatory damages alone, while they might compensate the injured party, will not deter the actor from committing similar trespasses in the future, there is ample justification for an award of punitive damages.. .. Accordingly, the issue of punitive damages should have been submitted to the jury. 316 F.2d at 275. Upon remand a second jury denied any award of punitive damages, and the plaintiff appealed. In Lampert v. Reynolds Metals Company, 372 F.2d 245 (9th Cir. 1967), the court noted that its earlier view of punitive damages had been confirmed in the interim by the Oregon Supreme Court’s McElwain opinion. During the retrial, the District of Oregon judge hearing the case instructed the jury that, “with regard to punitive damages, it should weigh the apparent value to society of plaintiffs’ farming activities.” 372 F.2d at 247. The Ninth Circuit rejected this view, stating: Without doubt, the operation of the Reynolds Metals Company at Troutdale has social value in that community. But in legal contemplation, the company has no obligation to provide that social value, and certainly no right to do so in disregard of its legal obligation not to cause trespass injuries upon the property of plaintiffs. We find no Oregon decision nor, indeed, any decision from any jurisdiction, which supports the weighing process sanctioned by the trial court. 372 F.2d at 248. The decision was reversed and the cause remanded for a new trial. This guidance provided by the Oregon Supreme Court and the Ninth Circuit Court of Appeals, though specific to the context of industrial air pollution, does not define with precision the circumstances justifying the imposition of punitive damage liability. A broad synthesis of these opinions provides the conclusion that punitive damage awards may be imposed for business activities, harmful to others, carried out in disregard of the corporation’s societal obligations. In brief, the issue is whether the defendant has damaged the property of plaintiff by conduct evidencing an “I don’t give a damn” attitude. For a case as complex as this, however, it is important to describe in some greater detail the societal obligations of business enterprises. II. SOCIETAL OBLIGATIONS OF BUSINESS ENTERPRISES. In essence, any business -is socially obliged to carry on an enterprise that is a net benefit, or at least not a net loss, to society. The company’s management recognized this obligation in a 1960 letter to the Wasco County Fruit and Produce League. In closing I [Lawrence Harvey] would like to reemphasize our desire to foster the prosperity of the entire communi *989 ty. . . . We are doing, and will continue to do so, the best scientific job of control that is possible under the circumstances. These are obligations which we consider part of our community responsibility. Ex. 305a at 2. In a world where all costs of production were borne by the enterprise, determining whether a firm produced a net benefit, or at least not a net detriment, to society would be as simple as examining the company’s balance sheet of income and expenses. In the real world the task is more complex, because enterprises can sometimes shift a portion of their costs of production onto others. In the case of an industrial plant emitting pollution, those harmed by the emissions are, in effect, involuntarily bearing some of the firm’s production costs. Our society has not demanded that such externalized costs of production be completely eliminated. Instead, we tolerate ex-ternalities such as pollution as long as the enterprise remains productive: that is, producing greater value than the total of its internalized and externalized costs of production. A business that does not achieve net productivity is harmful to society, detracting from the standard of living it is designed to enhance. Because firms can sometimes impose a portion of their production costs upon others, the mere fact that a company continues to operate at a profit is not in itself conclusive evidence that it produces a net benefit to society. Our system of law attempts to ensure that businesses are, on balance, socially beneficial by requiring that each enterprise bear its total production costs, as accurately as those costs can be ascertained. A fundamental means to this end is the institution of tort liability, which requires that persons harmed by business or other activity be compensated by the perpetrator of the damage. In the context of pollution, however, the tort system does not always operate smoothly to impose liability for compensatory damages. Among the difficulties encountered are: (1) that the harm may be gradual or otherwise difficult to perceive; (2) that the cause of the harm may be difficult to trace to the pollution and from the pollution to its source; and (3) that the harm may be inflicted in small amounts upon a large number of people, none of whom individually suffer sufficient damage to warrant the time and expense of legal action and whose organization into a plaintiff class is hindered by what has come to be known as the tragedy of the commons. 1 Because of these impediments to smooth operation of the tort system and to ensuring that each enterprise bears its own costs of production, the law imposes upon businesses a societal obligation not to obstruct legal procedures designed to provide compensation to persons harmed by externalized costs of production. Enterprises must cooperate with their neighbors in ascertaining the nature, severity and scope of the harm and in arranging to prevent the damage or to neutralize it through some form of compensation. A breach of societal obligations justifies the imposition of punitive damages to deter uncooperative behavior that impedes the legal system from ensuring that enterprises produce a net benefit to society. III. EVALUATION OF THE DEFENDANT’S CONDUCT IN LIGHT OF ITS SOCIETAL OBLIGATIONS. Although the company did not fail to carry out its societal obligations in every respect, I have concluded that the overall conduct of the business with respect to ascertaining the harm from the plant’s emissions, efficiently controlling the harmful *990 emissions and arranging to compensate for the remaining harm constitutes breach of societal responsibility sufficient to justify the imposition of punitive damages. A. Ascertaining the Harm from Plant Emissions. A business enterprise has a societal obligation to determine whether its emissions will result in harm to others. Because the damage from pollution can be difficult to perceive due to its subtle or incremental nature, and because it can be difficult to trace to its cause, the obligation of the enterprise extends not only to observation of property in the surrounding region but also to initiation and completion of unbiased scientific studies designed to detect the potential adverse effects of the substances emitted. I find that the company failed to fulfill this obligation before or during the 1965-71 claim period by taking less than full cognizance of the damage inflicted upon the orchards and by generally shirking its responsibility to undertake competent scientific inquiry into the adverse effects of its emissions. 1. The Company’s Efforts. The company did not maintain throughout the 1965-71 period a regular program of inspecting the nearby orchards to detect damage that might have been caused by fluoride emissions from the plant. Since late 1961 the company has maintained its own orchard in the vicinity, operated by Frederick Scholes, as a means of monitoring the effects of the plant’s emissions upon fruit crops. Ex. 933 at 4. During the spring and summer of 1965 the neighboring orchards were inspected on several occasions by Scholes, Joseph Byrne (the company’s environmental control officer) and two consultants hired by the company to evaluate orchard conditions, Michael Treshow and Earle Blodgett. Similar inspections were carried out in 1966 by these same persons accompanied by three additional hired consultants: O. C. Taylor, Leonard Weinstein and R. F. Brewer. Treshow and Blodgett continued their inspections during the spring and summer of 1967 and were joined in August 1967 by Neis Benson of Washington State University. Neither company personnel nor hired consultants inspected the orchards during the years 1968-70, after the arbitration committee established by the 1966 Renken consent decree began its scrutiny of the surrounding vegetation. Scholes and two hired consultants, Treshow and David MacLean, visited the orchards in August, 1971, some four months after the close of the plaintiff’s March 31, 1965-71, claim period. Timothy Facteau of the Oregon State University Mid-Columbia Agricultural Experiment Station (MCAES) agreed that during these years the company did call in many recognized experts on the effects of fluoride exposure to fruit crops. TR 747. 1 2 Neither before nor during the 1965-71 period did the company undertake or sponsor any scientific inquiry into the effects of fluoride upon vegetation. Ex. 339 at 178-80; TR 678-84, 873, 886. The company’s sole contribution seems to have been the donation of a $12,000 automatic material sample analyzer to Oregon State University to expedite analysis of the Mid-Columbia Agricultural Experiment Station (MCAES) fluoride fumigation experiments with cherry tree limbs in 1966. Ex. 339 at 163; TR 127, 805. a. Cherries. In 1960 the company hired a consultant, Burton Richards, to inspect the orchards. Ex. 305a. Along with Scholes and Byrne, Richards in 1961 was shown stylar dimpled cherries in the orchards. TR 387-88 (Bailey). During their 1965 inspections the company’s orchard observers noticed damage of various sorts but did not attribute any of it to fluoride exposure. See Exs. 501a-d, *991 502a-c, 503a-e. 3 None of the inspectors in 1966 attributed the damage they observed, including cherry blossom petal browning and necrosis, to fluorides except for Brewer, who commented that a small amount of interveinal leaf chlorosis might have been associated with fluoride toxicity. Ex. 506c; see Exs. 501h, 502d, 503f-k, 505a-d, 507. Taylor reported that “no fluoride type symptoms were observed on foliage of sweet cherry in the 34 orchards inspected,” though he did find some chlorosis, leaf bronzing and rolling in all of the orchards and petal necrosis in several. Ex. 505a, 505d at 2. Treshow and Blodgett in 1967 again reported no fluoride-induced injury, although Treshow mentioned “marginal chlorosis and cupping which reminded somewhat of fluoride symptoms” before concluding that “fluorides were not involved.” Ex. 5031; see Exs. 501e-g, 5031-n. Byrne observed stylar dimple in the Hendricks orchard in 1969 but did not attribute this to fluoride exposure. TR 908-09. The company did not conduct orchard inspections again until August, 1971, when Scholes, Treshow and MacLean did not report any observed fluoride injury to cherries. Exs. 502d, 503o, 504a. Scholes considered the cherry crop from the company’s orchard to be “good” in 1970 and “satisfactory” in 1971. Ex. 933 at 17. b. Peaches. In 1965 the company’s orchard inspection team found no significant fluoride-induced harm to peaches, Exs. 501a-d, 502a-c, 503a-e, although Scholes noted “several light cases” of soft suture in the Curtis orchard. 4 Ex. 502c. In 1966 Taylor reported 3-4% soft suture in the Francois and Curtis orchards, where “the combined injuries from split pit and soft suture may have affected as much as 20% of the crop.” Ex. 505d at 2. Other orchards showed less than 1% soft suture. Ex. 933 at 12 (Scholes). Taylor noted that soft suture was present only where the peach crops had earlier been reduced by the cold weather immediately following the blossoming period. Ex. 505d at 2. From the notes of the other company-hired observers, only Brewer also mentioned this soft suture. 5 Ex. 506d. In 1967 Treshow noted one instance of soft suture at the Myers orchard and a trace of that type of injury in the Hazen and Curtis orchards. Ex. 503n; see Exs. 5031-m. Blodgett reported some soft suture in a sample of peaches provided by the plaintiff and well over 50% soft suture in the Fleck orchard. Benson considered 52 of 64 Fleck peaches to be stricken (81%); Blodgett stated that only about 40-42 had “really convincing symptoms” (63-66%). Ex. 501g at 1; see Ex. 342 at 60; TR 200, 379-80. He considered this high incidence of injury relatively insignificant because the orchard’s crop was too light to justify harvesting anyway. He found a few soft sutured peaches in the Francois orchard and less than 5% soft suture at the Myers orchard. He summarized as follows: ... It would appear that, considering the orchards observed, there was no significant injury to peach fruit in The Dalles area from Fluorine this year except in the Fleck, Company and possibly the Bill Myers orchards. Ex. 501g at 3. Blodgett also noted the presence of soft suture in peaches in the Mosier vicinity (approximately 12 miles northwest of the plant), “out of the contaminated area.” Ex. 501g at 3. *992 The August, 1971, reports of Scholes, Treshow and MacLean did not mention any injury to peaches. Exs. 502d, 503o, 504a. Scholes stated that since 1963 he had seen “only a few instances” of soft suture in The Dalles area. Ex. 933 at 4. c. Pines. There is no evidence that the company undertook systematic observation of pine trees in the area around the plant either before or during the 1965-71 claim period. d. Other Vegetation. Scholes mentioned that he might have observed apricot leaf necrosis “on a small basis” in the company’s orchard during 1964-66. TR 689. Blodgett mentioned some insignificant prune leaf necrosis in 1965 he thought “not typical” of similar symptoms caused by fluoride. Ex. 501e at 2. In September, 1965, Treshow reported “trace amounts of fluoride-type necrosis ... on apricot foliage in the Folehn and Ericksen orchards.” Ex. 503e at 1. In 1966 Taylor noted “very light fluoride-type necrosis at the tips of some [apricot] leaves” in the Ericksen orchard. Ex. 505d at 3. No other damage to vegetation attributed to fluoride was reported by the company-hired inspectors. 2. Efforts of Others. a. Cherries. 1) Observation of Trees, Leaves and Fruit. Walter Mellenthin, O. C. Compton and others at MCAES made surveys of the amount of fluorine in vegetation in The Dalles area within six miles of the plant site in 1953 and twice during each growing season from 1957 through 1967. See Exs. 119-24. The plant began operating on July 26, 1958, after the MCAES July, 1958, survey. These studies reported average fluorine contents of sweet cherry foliage and forage in parts per million as follows (figures in parentheses are average fluorine contents at sampling locations 1-2 miles from the plant): At least part of the dramatic decrease in leaf fluorine content between 1962 and 1963 may be attributed to the initiation in 1963 of the practice of washing the leaves prior to crushing and analysis, see Ex. 342 at 97-98, although the first mention of this procedure in the reports appears in association with the 1965-67 data. Ex. 124 at 2. Each of these surveys found that the fluorine content of cherry leaves generally increased with proximity to the plant. None of these reports mentioned the observation of damage to cherry leaves or fruit from fluoride exposure. Nevertheless, Mellenthin testified that he first noticed scorching of cherry leaves and stylar dimpling of cherries in The Dalles area in 1959 and 1960. Ex. 342 at 20-21. In 1960 and 1961 he photographed scorched cherry leaves and dimpled cherries at the Hendricks, Myers and other orchards but did not know then what had caused these conditions. Ex. 342 at 40-47. U. S. Department of Agriculture inspector Warren Cyrus noticed stylar dimpling around 1960, a condition he had not seen during previous periods of cherry inspection in Medford, MiltonFreewater and Hood River. Ex. 330 at 2007. The cherries from The Dalles area were not dimpled every year but were so afflicted during about four of the six following years. Ex. 330 at 2008, 2022. In August, 1960, the Wasco County Fruit and Produce League complained in a letter to the company that the plant’s emissions were causing “severe damage to the fruit set, fruit, and foliage of our cherry trees.” Ex. 305b at 1. This year the general pattern of production showed that the farther away *993 from Harvey Aluminum plant our cherry orchards were located, the better was our production..... Most of the distant orchards had a good crop while those close in generally had less than half a crop. The district was short of a normal crop by several thousand tons with the resultant loss in value of approximately $1,000,-000. Some of the fruit from orchards near your plant which was picked up for canning, in a fully ripe condition, showed suspicious symptoms. This consists of a blossom end deformity and withering of the fruit. In addition to this, we are receiving a substantial amount of marginal leaf burn and a rolling and drying of the foliage. We feel this will have a definite adverse effect on our cherry crop the next year. Ex. 305b at 2. The league in December, 1960, presented to the Oregon State Sanitary Authority in the presence of the company’s legal counsel an analysis showing that the 1960 cherry crop was reduced to one-third of the 1958-59 amount at orchards experiencing a cherry leaf fluorine content of at least 60 ppm, while the crop remained at 102% of the 1958-59 amount at the orchards 4-6 miles from the plant experiencing a cherry leaf fluorine content less than 60 ppm. Exs. 170g-l, 170g-2; TR 403. In April, 1962, the league adopted a statement linking cherry fruit set with fluoride exposure. Ex. 170b-l. Wasco County agricultural extension agent John Thienes did not notice any fluoride symptoms on cherries in his own orchard (about 4.5 miles from the plant) during the 1961-66 period but he did see cherry leaf burn prior to the “first major improvement” to the plant. TR 226; Ex. 345 at 32. He concluded that he has “over the years seen a definite pattern ... of crop reduction based on distance and direction from the [plant].” Ex. 345 at 33. After conducting fluoride aqueous spray and fumigation experiments, Mellenthin concluded by 1965 that fluoride was the cause of the cherry leaf burn and stylar dimple he had noticed in the orchards. Ex. 342 at 54-55. In a 1966 deposition taken by attorneys for the company he noted that the stylar dimple he had observed in the orchards in 1962 could be associated with the dimpling of cherries resulting from the fluoride fumigation experiments. TR 121. When the previous county extension agent, Elgar Nelson, in 1966 returned from 10 years abroad with the AID, he noticed stylar dimpling and a generally lighter crop of cherries compared to 1956. Ex. 344 at 12-14. The expenses of the arbitration committee established under paragraph 6 of the 1966 Renken consent decree were shared equally between the company and the Renken plaintiffs. This committee visited the orchards in April, 1967, found cherry blossom petal necrosis in several areas, but stated: ... The consistent location of damage to petals of cherry blossoms on the trees and fruiting spurs at each of the three locations is not consistent with expected damage by an atmospheric pollutant but rather with damage mediated through an environmental phenomenon, such as low temperature or rapid temperature changes. Ex. 907. Later that month one of the arbitrators, Taylor, reported: . . . The symptom most commonly attributed to fluoride injury on cherry blossoms was a peripheral burning or necrosis of petals .... This symptom was observed throughout The Dalles cherry growing area, even in the most distant and remote orchard visited, one located on a ridge some six or more miles south of the aluminum plant.. . . Identical petal burn symptoms were prevalent on Royal Ann blossoms at the Oregon State University Station near Mosier [approximately 12 miles northeast of the plant]. Many environmental factors may contribute to the petal burn symptom but there is very strong evidence that cold temperatures, and not necessarily frost or freezing temperatures, are the primary causal agent. There was no evidence that expo *994 sure to gaseous pollutants played any part in producing the petal burn. ... No injury was observed on blossoms or leaves of sweet cherries and peaches in The Dalles area that was considered to be attributable to exposure to air pollutants and not to adverse climatic conditions. Ex. 908. During the spring of 1968, orchardist Walter Ericksen recruited temporary county extension agent Robert Smith to assist him in gathering documentation of fluoride damage to cherry blossoms. They selected cherry tree limbs and photographed them each morning during the blossom period from March 30 through April 9. Ex. 226 at 6. They observed a petal browning effect within a day following the presence of smoke from the plant, an effect different from the petal browning caused by frost they observed at lower elevations. Petal browning from frost shows “a water-soaked effect,” while browning observed after exposure to smoke from the plant shows a “cupping or crinkling on the edge of the petal [with] some browning ... in towards the center of the flower itself.” Ex. 343 at 285-86 (Smith). Smith compiled his observations into a report, which noted that the temperature at the nearest weather station fell below freezing on March 30 and 31, just prior to his observation of abnormal petal browning on the mornings of March 31 and April 1. He did not believe that the blossoms had been injured by this frost, because no water-soaked petal browning was noticed and because the weather station recording the freezing temperatures was located about 400 feet below the trees under observation and had been deliberately placed in a “frost pocket” that was expected to experience the coldest temperatures in the orchards. Ex. 226 at 2-3; Ex. 338 at 514. Smith stated that he observed no other symptom of damage and that the browning of the petals “in itself would probably not be significant.” Ex. 343 at 288. In August 1968 the arbitration committee noticed cherry leaf necrosis in one orchard but attributed it to disease rather than to exposure to air pollutants. Ex. 909 at 3. In April, 1969, the committee visited several orchards and observed cherry blossom necrosis ... throughout the area, and this condition did not appear to be related to direction from the fluoride source or proximity to [the plant]. There was no evidence that this condition was associated with fluorides in the atmosphere. Ex. 508 at 1. Nelson reported, however, that on the morning of April 21, 1969, cherry blossoms in the Hendricks’ orchard began to brown rapidly after the nearby analyzer registered a sharp increase in the atmospheric concentration of gaseous fluoride. Ex. 344 at 30-31, 50-51. He distinguished this effect from the normal browning that overtakes cherry blossom petals within about two weeks of blooming by noting the rapidity of the browning and its association with a cupping of the petals, which ordinarily remain flat as they brown with maturity. Ex. 344 at 61. He also noted a scorching of the pistils and filaments, scorched cherry leaves and “burned matchsticks” (infertile cherry fruit with stems) elsewhere near the plant. Ex. 344 at 61, 75-76, 79-80. Another inspection by the arbitration committee in late June, 1970, again found “no fluoride type symptoms.” Ex. 510. Conversely, Mellenthin again observed and photographed stylar dimpling of cherries in the Cherry Heights area near the plant in 1970 and 1971. Ex. 342 at 72-74. 2) Scientific Inquiry. In 1961 MCAES researchers selected 10 pairs of Royal Ann cherry trees in the Mosier area (about 12 miles northeast of the plant) and sprayed one of each pair with a 500 ppm ammonium fluoride solution seven times between April 7 and July 18. The leaf fluorine content in the sprayed trees increased to 119 ppm after six spray applications-more than 100 times the fluorine content of cherry leaves from the unsprayed trees; about equivalent to the average fluorine content of cherry leaves in The Dalles area tested by MCAES in Au *995 gust, 1959, July, 1960, and October, 1962; and about half of the fluorine content of cherry leaves in The Dalles area tested by MCAES in September, 1960. Ex. 121 at 5, Ex. 119 at 23; Ex. 120 at 11; Ex. 122 at 16. The MCAES researchers found “no visible leaf abnormalities” in the sprayed trees but noted that cherries from the sprayed trees showed “the presence of a slight depression near the tip” and contained 8.3 ppm fluorine in the tip half compared to 4.4 ppm fluorine in the tip half of cherries from the nonsprayed trees. Ex. 121 at 5; see Ex. 342 at 50-51. In April, 1962, the Wasco County Fruit and Produce League issued a statement describing the conclusions of a presentation by Mellenthin of MCAES given at a horticultural meeting: A 50 ppm ammonium fluoride spray reduces cherry pollen viability and, when applied to female flower parts, reduces cherry fruit set by more than 50% and decreases individual cherry fruit weight. In 1965 MCAES conducted fluoride fumigation experiments with cherry trees in Mosier and found that exposure to fluoride-containing gas during 12-15 hours beginning about 5:00 P.M. every day produced stylar dimple in cherry fruit. Ex. 342 at 54-57. Compton had at that time drawn no conclusions about the effect of fluoride exposure upon cherry fruit set. Ex. 342 at 58. In the spring and summer of 1966 MCAES conducted additional fluoride fumigation experiments with cherry trees in Mosier. The trees were exposed to a concentration of 4 micrograms of hydrogen fluoride gas per cubic meter of air (4 mg/m3), which produced dimpled fruit. TR 128 (Mellenthin). The results of this work were presented at a “field day” session attended by representatives of the company, among others. TR 384. In a July, 1966 deposition Mellenthin testified in the presence of attorneys representing the company that the fluoride fumigation reduced cherry fruit set and size. TR 118-19. The possible deleterious effects of the limb cages used in the experiment were accounted for by “caging” several control limbs not exposed to fluoride gas inside the cages. Ex. 342 at 69. The researchers at MCAES could not promptly measure the concentration of fluoride inside the cages due to the lack of an analyzing device. Air samples were sent to Oregon State University for analysis but the results were delayed for months. Ex. 342 at 71-72; TR 128. Mellenthin did conclude, however, that caging could not produce the observed stylar dimpling. Ex. 342 at 80. Dr. Timothy Facteau conducted cherry fruit set studies in 1967, 1969 and 1970. In 1970 he concluded that ambient atmospheric fluoride concentrations in The Dalles area resulting from the plant’s emissions were capable of reducing cherry fruit set, and he so testified in a deposition taken in the presence of attorneys representing the company. TR 736-37. His 1969 fumigation experiment at 20 mg/m3 “effectively reduced [cherry fruit] set to nothing.” TR 738. His 1970 deposition noted that fluoride exposure reduced cherry fruit set but included no conclusions about the ultimate effect upon cherry production. TR 739^40, 780. Mellenthin, however, concluded that a reduced fruit set “would generally reduce the crop.” Ex. 342 at 26. In 1970 Facteau began to examine the effect of fluoride exposure on the cherry pollen tube. TR 740. It wasn’t until 1977 that Facteau, based upon 1976 experimentation, tentatively identified a threshold exposure to hydrogen fluoride below which there is no adverse effect upon cherry fruit set: 5 mg/m3. Ex. 934 at 47. Five weeks later he stated to the contrary that he had not found a threshold level below which no fruit set reduction may be observed. TR 725. ... I’d have to go back to the original idea that any time we can decipher an elevated level of fluoride in our sampling stations during the blooming in The Dalles, I would have to give the opinion I think there is a probability there could be a reduction in set. TR 726. Meanwhile, Curtis Mumford, an agricultural economist at Oregon State University, in February, 1970, issued the results of a *996 study of cherry crop yields in The Dalles from 1951-67 initiated in late -1967 at the behest of the Wasco County Fruit and Produce League and based upon confidential production data supplied by the League. See Ex. 938. The report stated that the average yield of sweet cherries after the plant began operating (1959-67) was 4.16% higher than the average yield before (1951-58). In addition, the data failed “to disclose any definite pattern in sector by sector changes as related to distance from the aluminum plant.” Ex. 695 at 3. Mumford noted, however, that the study had not been controlled to account for changes in the varieties of sweet cherries cultivated, in the number and spacing of trees, and in various horticultural practices, including irrigation, fertilization, pollenization, spraying and management. Ex. 695 at 2. ... It is known that production technology has improved generally, and that yields of all types of crops have increased during the study period. Even though yields of sweet cherries increased slightly during the study period, it is possible that the increase would have been even greater in the absence of hydrogen fluoride in the environment; however, as indicated above, the data do not warrant conclusions concerning yields in the absence of hydrogen fluoride. Ex. 695 at 3. Orchardist Ericksen testified that fruit production in The Dalles had been increased by irrigation, interplanting of additional trees or replanting orchards more densely, and the use of tractors, speed sprayers, fertilizer and aerial spraying for cherry fruit flies. Ex. 338 at 515-20. The combination of irrigation and denser spacing of trees “can increase up to double the production.” Ex. 338 at 518. Considering the potential for increased cherry production offered by these horticultural innovations, the finding of Mumford’s study that cherry crops increased by slightly more than 4% during the 8-year period following commencement of the plant’s operations compared to the prior 8-year period is not strong evidence that the plant’s emissions had no adverse effect upon cherry production in The Dalles. b. Peaches. The 1953-67 MCAES surveys reported average fluorine contents of peach foliage in parts per million as follows (figures in parentheses are average fluorine contents at sampling locations 0 — 1 mile from the plant): As noted earlier, at least part of the dramatic decrease in leaf fluorine content between 1962 and 1963 may be attributed to the practice initiated in 1963 of washing the leaves prior to crushing and analysis. See Ex. 342 at 97-98, Ex. 124 at 2. Each of these surveys found that the fluorine content of peach leaves generally increased with proximity to the plant. The MCAES survey in 1959 found soft suture in peaches “not observed previously in this area.” Ex. 119 at 1, 6. Bailey noted “a serious condition appearing in some of the peaches delivered at the Columbia Fruit Growers dock” in 1959. A representative from the plant observed this damage but could not identify the cause. TR 422. The U. S. Department of Agriculture inspector confirmed that peaches from The Dalles area began to show soft suture beginning in 1960. Ex. 330 at 2009. The Wasco County Fruit and Produce League requested the Department to inspect the peach orchards, and Inspector Cyrus found that the percentage of peaches with soft suture increased with proximity to the plant. Ex. 330 at 2012. Normally the percentage of peaches rejected by USDA inspectors as “culls” averages 6-7% and may range up to a maximum of 20%. After the appearance of soft suture in The Dalles, however, the inspectors began to reject 71-93% of the peaches as unmarketable. Ex. 330 at 2017-18. *997 In August, 1960, the Wasco County Fruit and Produce League complained of damage to peaches in a letter to the company. In 1959 our crops of apricots and peaches showed some signs of damage with which we were not familiar at the time, and we did not become too alarmed about the problem then. In our 1960 crop, we are suffering extreme damage in an area within four to five miles from your local plant, and our growers are greatly concerned. .. . The peach damage consists of a premature ripening and blossom end breakdown of the fruit. The peaches affected are very irregular in shape and color in this blossom end area. This renders from eighty to one hundred per cent cullage in some cases close in to the aluminum plant.... The above described symptoms have been identified by experts in the field as fluoride injury. Ex. 305b at 1. Of 100 Red Haven and Golden Jubilee peaches picked at random from each of six trees at the MCAES orchard at The Dalles, 73-95% showed soft suture. Ex. 120 at 5. The September, 1960 MCAES survey found soft suture in 0.8-33.3% of the Elberta peaches from various orchards inspected at the Columbia Fruit Packing House; Elberta peaches in the Myer orchards showed 77.6-98.6% soft suture. The J. H. Hale peaches inspected at the packing house showed 0 — 40% soft suture; 39.6-55.8% of Myer’s J. H. Hales were afflicted. None of the 79 J. H. Hale peaches from the Troxel orchard in the Mosier area (about 12 miles northwest of the plant) showed soft suture. Ex. 120 at 19-20. MCAES reported that normal Elbertas from The Dalles area had an average fluorine content of 5.3 ppm on the suture side and 4.6 ppm on the dorsal side, while soft sutured Elbertas had 9.0 ppm on the suture side and 5.4 on the dorsal side-about a 70% greater concentration on the suture side and 17% more on the dorsal side than the normal Elbertas. Ex. 120 at 18. In September, 1961 MCAES found 2 ■ 36% soft suture in the Improved Elbertas in The Dalles area compared to 0% in the Mosier area and 2% in the Hood River area; 10-52% soft suture in The Dalles J. H. Hales compared to 0% in Mosier. Ex. 121 at 22. Orchardist Ericksen testified that Scholes visited his orchard that very month and conceded that his early ripening varieties of peaches were a total loss. Ex. 338 at 545 — 46. In 1964 extension agent Thienes testified by means of affidavit in Renken before the Ninth Circuit Court of Appeals that the 1964 peach crop in Wasco County was “near normal” with good quality fruit and “soft suture being a very minor factor this year.” Ex. 749. A U. S. Department of Agriculture inspection of Improved Rochester peaches from the Ellett orchard on August 4, 1965, found that all the peaches showed “serious damage caused by dryness, cracking and/or breakdown on blossom end and premature ripening of lower suture area.” Ex. 154t. An August, 1966 inspection of eight lots of peaches from The Dalles area revealed that 45-60% of the Elbertas suffered this same type of damage, as did 100% of the Improved Elbertas from the Francois orchard. Of the J. H. Hale peaches examined, only 9% showed this injury at the Fleck orchard and 36% at the Myers orchard, but similar damage appeared in 67-70% of the J. H. Hales at the Francois and Ellett orchards. Exs. 1541-s. Scholes and Byrne were present during these inspections. TR 375 (Bailey). In April, 1967, Taylor, on behalf of the arbitration committee, reported the observation of no injury to blossoms or leaves of peach trees attributable to air pollution. Ex. 908 at 2. The arbitration committee visited the orchards in The Dalles area in August, 1968, accompanied by representatives of the company. The committee reported the observation of “split-pit ... but no typical suture-red-spot symptoms” in the Ellett orchards, split-pits and a “very light incidence” of suture-red-spot in the Fleck orchards, no suture-red-spot in the Hendricks orchard (but in half of the 10 peaches from his refrigerator), much split- *998 pit but no suture-red-spot in the Francois orchard, and no evidence of fluoride injury in the Curtis Brothers or Bailey orchards. Ex. 909. In April, 1969, the committee noted severe winter injury to peaches in the Cherry Heights area but not in the Mill Creek area. Ex. 508 at 1. Former extension agent Nelson in August, 1969, observed and photographed “close to 100%” soft suture in the El'lett orchard in the Cherry Heights area. Ex. 344 at 98; see Ex. 344 at 85-102. In June, 1970, Taylor reported on behalf of the arbitration committee that no “fluoride type symptoms” were observed in the Francois peach orchard. Ex. 510. Orchardist Myer in August or early September, 1970, photographed soft-sutured peaches he found lying on the ground in the company’s orchard after some harvesting. Ex. 332a at 18-20. Myer believed that these were worthless peaches that had been thrown on the ground during picking, but Scholes stated that these were ripe peaches that had been detached by a squall wind before he had a chance to gather them up. TR 675. Finally, Mellenthin testified that he observed soft suture in peaches in The Dalles in 1970 and 1971. TR 124. c. Pines. The Stanford Research Institute’s 1955 proposal to the company “for an investigation of potential air pollution conditions in the vicinity of your future aluminum reduction plant at The Dalles, Oregon,” noted: With some vegetation (such as gladioli and pine trees), a relatively low level of accumulation of fluorides causes injury to the plant material, and even death. Ex. 220d at 1. In 1959 agricultural extension agent Thienes noticed damage to spruce that he attributed to fluoride exposure, and he showed this condition to a representative of the company. TR 196-97. Thienes also observed needle burn or “blight” in Ponderosa pine “from high level of fluoride ... locally associated with the areas immediately surrounding The Dalles.” Ex. 345 at 24. Mellenthin observed “so-called die-back of the needles .. . which we hadn’t seen before.” Ex. 342 at 15. He and MCAES researchers in 1961 surveyed one-year-old pine tree needles at 13 sites from one to four miles from the plant and at one site in Hood River, 20 miles west of the plant. At each site 5-10 trees were examined, and from each tree 5-10 branches were cut off and the needles stripped, inspected for needle burn or scorch, chopped and analyzed for fluorine content. At the Hood River site they found needle fluorine contents of 6.5 and 19.3 ppm; at the locations around the plant, an average needle fluorine content of about 65 ppm. None of the Hood River needles suffered the scorch or “blight” previously “associated with abnormal concentrations of fluorine in the needles.” Ex. 159 at 1, 6. In contrast, needles from The Dalles showed scorch averaging from 0.9% the length of the needles at the site 4 miles south of the plant to 67.6% the length of the needles at the site 1.2 miles northwest of the plant. The researchers specifically discounted any connection between this injury and insects or soil conditions. “There was no pathological, entomological, or soil condition that would account for the needle scorch found in the area.” Ex. 159 at 5. Instead, they attributed the injury to exposure to fluorides. ... Generally, the greater amount of scorch was associated with higher fluorine content. The correlation coefficient between scorch and fluorine content was + .50 and was significant at the 5% probability level. Ex. 159 at 4. The researchers also measured the sapwood resin pressure of the trees and found “an abnormal increase in the pressure of the water in the stem” of trees showing the most severe needle scorch. Ex. 159 at 4. They concluded that “[t]he severely affected trees may eventually die of starvation because of the greatly reduced leaf area.” Ex. 159 at 5. During a 1961 tour of the pine groves, Oregon state pollution control administrator Richard Hatchard saw “considerable damage to the needles, which presented a dying tree appearance.” Ex. 346 at 8. In *999 July, 1965, the MCAES researchers found 24.1-104.4 ppm fluorine in one-year-old pine needles near The Dalles and scorch averaging 21.8-47.5% of their length. Needles taken from Hood River contained an average of 7.3 ppm fluorine and showed no scorch. Ex. 124 at 4-5, 23. In August, 1968, the arbitration committee found heavy pine needle tip necrosis on the Hendricks property but did not attribute this affliction to fluoride exposure. Ex. 909 at 3. d. Other Vegetation. Extension agent Thienes in 1958 was alerted by orchardists to the browning of apricot leaves, which he attributed to fluoride exposure. Ex. 345 at 17, 31. In 1959 Mellenthin of MCAES observed and photographed apricot leaf scorch in the Bunn orchard. He did not then know the cause but later concluded it to have been fluoride. Ex. 342 at 38. Also in 1959 MCAES researchers noticed “a small amount of marginal burn characteristic of fluoride scorch on apricot leaves taken from [two stations within 4 miles of the Plant]” in June. In August they found leaf scorch in all apricot orchards at five other sampling stations, as well as prune leaf scorch. The scorched apricot leaves contained about 16% more fluorine than non-scorched leaves; the scorched prune leaves, about 37% more fluorine. Ex. 119 at 6. In 1960 MCAES researchers reported a “marginal burn characteristic of fluoride scorch ... on apricot leaves taken from most of the stations" and on prune leaves at one location. Ex. 120 at 5. Again in 1961 and 1962 they noticed slight marginal necrosis “characteristic of fluoride injury” in some of the orchards. In 1962 injured leaves contained an average 35.4 ppm fluoride; non-in jured leaves, 20.1 ppm fluoride. Ex. 121 at 4; Ex. 122 at 7. 3. The Company’s Response to Evidence of Harm from the Plant’s Emissions. a. Cherries. As noted previously, the Wasco County Fruit and Produce League complained of damage to cherries in a letter to the company in 1960. Since that time, company representatives, particularly Scholes and Byrne, were repeatedly alerted to the possibility of harm by the orchardists, the county extension agents, the federal fruit inspectors and the MCAES researchers. See Ex. 339 at 168; Ex. 343 at 293; TR 153-54, 197-98, 375, 385, 396-98, 419. Though this evidence was sufficient to convince the jury in this case in 1973 to award the plaintiff $88,135 for damage to cherry crops during the 1965-71 period (85% of the total compensatory damage judgment), it was not taken so seriously by the company. Scholes consistently reported to the company that the plant’s emissions were causing no damage to cherry crops. Ex. 933 at 6. Byrne testified that he had never seen damage to cherry trees caused by fluoride emissions, even though Scholes had reported to him his observation of “snubbed-nose cherries” around the United States Steel plant in Utah. Ex. 935 at 2; Ex. 339 at 55; TR 865. Plant manager Taylor Gibson testified that “the most competent scientists available to us ... uniformly report that any unusual symptoms observed in the orchards are not caused by emissions from our plant.” Ex. 931 at 4. After Byrne was hired by the company in 1961 to evaluate the orchardists’ claims that their crops were being damaged, he undertook no survey of the scientific literature on the subject of fluoride effects upon vegetation. Ex. 339 at 51. He discounted the applicability of the research with which he was familiar, concluding that the fluoride fumigation studies performed by Donald Adams and others at Washington State University prior to 1961 found cherry leaf necrosis only at very high concentrations of airborne fluoride. Ex. 339 at 55-58, 155. Instead, he concluded that the marginal chlorosis in cherry leaves in The Dalles was caused by a manganese deficiency. Ex. 339 at 155-56. Byrne was present at the 1963 and 1966 MCAES field days when the researchers discussed the results of their ammonium fluoride spray and hydrogen fluoride fumigation experiments. His evalúa *1000 tion of the experiments was that fluoride damages cherries only “at substantially high levels,” Ex. 339 at 166, “in excess of those levels measured in the ambient air in The Dalles.” Ex. 935 at 6. This appraisal was not entirely accurate. The 1961 MCAES 500 ppm ammonium fluoride spray experiment found stylar dimple corresponding to a cherry leaf fluorine content of 119 ppm-a level frequently exceeded in the area around the plant from its commencement of operations through 1962, if not later. 6 The 1966 MCAES hydrogen fluoride fumigation experiment found a reduction in cherry fruit set at an exposure level of 4 mg/m3 (approximately 4 parts per billion) — a concentration not infrequently experienced in The Dalles during bloom season in 1963. Ex. 122 at 10-11, 34. In contrast, the highest daily average concentration measured by MCAES during bloom season in 1964 was 2.5 mg/m3, Ex. 123 at 5; in 1965 was 3.31 mg/m3, in 1966 was 4.92 mg/m3 and in 1967 was 4.65 mg/m3. Ex. 124 at 5-6, 24-28. These daily average figures, of course, smooth over the peaks and troughs of actual atmospheric hydrogen fluoride concentrations during any particular day, and even Byrne eventually agreed that an ambient concentration of 4-5 mg/m 3 for a period of just 4 hours during bloom season could harm the cherry fruit set. TR 866. In my evaluation, this evidence, combined with the frequent complaints of the orchardists, renders disingenuous Byrne’s claim to a well-founded and sincere belief that the plant’s emissions never damaged cherry crops in The Dalles. Instead of considering seriously the results of the MCAES research, Byrne concluded that the spray and fumigation studies were worthless. TR 416 (Bailey), 817, 874-75. Byrne visited the MCAES facilities only once during the 1967-71 period when Facteau was continuing this research; no other company representatives paid any visits. TR 775. The company’s response to independent research into the effects of fluorides upon vegetation appeared to be somewhat hostile. 7 Spokesmen opposed a proposal before the state legislature for additional funds for MCAES research, instead proposing a “jointly-sponsored research project by the Company and by the growers.” Ex. 339 at 163. Byrne met with the orchardists and MCAES researchers to propose a “comprehensive evaluation of The Dalles cherry growing area.” TR 800. The orchardists rejected the proposal because they felt “there were too many strings attached.” TR 802. ... [Tjhey didn’t want us [the company] in the program. They felt if the company had funds in the program, we would control it, and they didn’t want any part of it. So that ended the cooperative aspect of the investigation. TR 802 (Byrne). The company offered to,fuñid a “reverse fumigation” experiment to compare the viability of fruit crops in The Dalles area with fruit trees shrouded by a material to filter out the fluorides in the surrounding air, but MCAES did not respond. TR 802-04. The company was admittedly capable of undertaking its own experimentation in the company-owned orchard. Byrne stated that he could have conducted a reverse fumigation study for the company but “really didn’t know” why he never did. TR 874. When Facteau of MCAES concluded on the basis of a 1970 fruit set survey that fluoride from *1001 the plant was harming the cherry crop in The Dalles, the company hired a statistician to provide an alternative interpretation of the data. TR 819. Prior to March, 1972, the company began negotiations with the Boyce-Thompson Institute for replication of Facteau’s research, yet Facteau was not invited to participate in this study and was not even aware of it until after its conclusion in 1976. TR 775a-76, 875-76. The preponderance of the evidence shows that, before and during the 1965-71 period, the company was more concerned with denying its responsibility for damage to cherry crops than with accurately detecting any adverse effects from exposure to fluorides from the plant. b. Peaches. Scholes evaluated the damage to peaches from the plant’s emissions as follows: ... I agree that prior to 1963 there was significant damage to peaches caused by emissions from the plant. Since 1963, I have seen only a few instances where peaches from The Dalles suffered soft suture. I believe that a lime spray will prevent soft suture. We ha.ve agreed to supply lime spray to growers who wish to use it. However, the incidence of soft suture has been so insignificant that I no longer feel that lime spray is necessary. Ex. 933 at 4. Byrne agreed that “fluorides were involved to some extent” with soft suture in peaches. Ex. 339 at 8. “[T]o the extent that there is soft suture, it is probably caused by fluoride from the plant.” Ex. 339 at 26. See Ex. 339 at 172. Byrne realized the connection between fluoride emissions and damage to peaches when he first began working for the company in The Dalles in 1961. Ex. 339 at 50. Though he was aware of the orchardists’ continued complaints and the results of the 1965 and 1966 U. S. Department of Agriculture inspections and the 1967 inspections by its hired consultants Benson and Blodgett, Byrne concluded that fluoride injury to peaches “hasn’t been a substantial problem since about 1963.” Ex. 339 at 173; see Ex. 935a at 28. The credibility of this statement is somewhat strained by Byrne’s admission that an atmospheric fluoride concentration of 1 part per billion (roughly 1 mg/m3) during the “particular time period when the peaches are particularly susceptible” would produce soft suture. Ex. 339 at 155. 8 The 1965-67 MCAES measurements, the 1967-71 arbitration committee measurements and the 1963-71 company measurements indicate that the 1 ppb level was frequently exceeded in the orchards around the plant during the 1965-71 claim period. Ex. 701a; Ex. 123. at 5; Ex. 124 at 24-28. Nevertheless, Byrne was willing to admit only, “Yes, there was a single peach in the period 1965 to 1971 that might have had soft suture.... There may have been, yes.” Ex. 935a at 27-28. As with cherries, I find the company’s efforts to ascertain what harm the plant’s emissions might be causing to peach crops in the vicinity to have been less than diligent and the company’s reaction to evidence of harm produced by the efforts of others to have been obstructionist. The company’s attempts to blame what reasonably appeared to be damage from fluoride upon plant diseases, weather conditions (including wind), soil conditions (including nutrient deficiencies), horticultural practices, insects, pesticides and a lack of water are not persuasive. c. Pines. Scholes agreed “that in the early years of the plant’s operations, i. e., 1960-1963, there was damage to pine trees caused by hydrogen fluoride emissions from the plant.” Ex. 933 at 3. Since then, the unhealthy condition of the trees, according to Scholes, is “the result of natural stresses, including lack of irrigation and scale.” Byrne knew about the problem of pine blight around an aluminum reduction plant in Spokane County, Washington, prior to the beginning of *1002 his tenure with the company in 1961. Ex. 339 at 54. He agreed with Scholes that the plant’s emissions have not harmed pine trees in The Dalles since 1963. Ex. 339 at 173. This conclusion is directly contradicted by the findings of the 1965 MCAES pine needle scorch survey described earlier, see Ex. 124 at 4-5, 23, and in my judgment does not constitute a well-founded and sincere evaluation of potential damage from the plant’s emissions. d. Other Vegetation. The plaintiff did not seek compensation for damage to vegetation other than cherry, peach and pine trees, and I have found no mention in the record of any damage to such vegetation from the plant’s emissions since 1966. I therefore do not base my conclusions regarding the social responsibility exhibited by the company upon evidence pertaining to other types of vegetation. 4. Conclusion. In sum, I find that the company’s failure to exercise sufficient diligence in obtaining information pertinent to damage to crops and trees that might have been caused by the plant’s emissions and to evaluate objectively the relevant information compiled by others constitutes evidence that the company was not fulfilling its societal obligations during the 1965-71 claim period. B. Efficiently Controlling the Harmful Emissions. During the claim period the company had a societal obligation to adopt and maintain reasonable pollution control measures, at least those capable of reducing the harm at a cost less than the damage caused by the emissions. Failure to adhere to such a course would result in a net detriment to society. An efficient program of pollution control also requires at least occasional monitoring of emissions as a check on the effectiveness of the control strategy. The plaintiffs contend that the company did not implement reasonably efficient and economical pollution control measures that could have reduced the fluoride emissions from the plant during the 1965-71 claim period. The company’s response is that its control system during that period was among the best in any aluminum plant in the world. Simply being “among the best” does not necessarily discharge the company's societal obligations, because the performance of other plants might also evidence social disregard. Moreover, a company’s obligation to restrict its emissions depends upon the damage they might cause; a greater degree of control is called for in a populated agricultural area than in more sparsely inhabited and cultivated surroundings. Because calculating the precise costs and benefits of various pollution control strategies is impossible, however, descriptions of the feasibility and effectiveness of systems employed in other plants are useful to indicate the degree of emission control the company might have reasonably been expected to achieve during the claim period. Although the company’s social obligation extends to the implementation of only those pollution control measures that efficiently reduce the damage to others, an award of punitive damages does not require the plaintiff to provide detailed analyses of alternative emission control strategies and their costs. Instead, the plaintiff need only show, by a preponderance of the evidence, the existence of pollution reduction measures that could have been adopted and reasonably might have been expected to efficiently decrease the plant’s emissions. It then becomes the burden of the company, with its superior access to scientific, technological, engineering, economic and management expertise, to show, by a preponderance of the evidence, that the measures proposed by the plaintiff were not available before or during the claim period or would not have resulted in efficient emission control. My examination of the evidence leads me to conclude that the company failed to implement reasonably efficient and economical pollution control measures that could have reduced the fluoride emissions from the plant during the 1965-71 claim period. *1003 1. Plant Siting. One means of avoiding harm to neighbors is to locate a plant likely to emit pollutants in a sparsely populated and cultivated area. Choosing a location without consideration of the surrounding land uses may result in the infliction of unnecessary damage upon neighbors during the entire operating lifetime of the plant even if otherwise efficient emission control techniques are implemented. The company apparently made no environmental studies before its decision to locate the plant in The Dalles. Ex. 334 at 653; Ex. 336 at 75. In October, 1955, the Stanford Research Institute submitted to the company a proposal for inspecting the surrounding agricultural areas both before and after commencement of plant operations and for conducting fumigation experiments upon the typical vegetation, but this proposal was not accepted by the company. See Exs. 220d and 220e. Nor did the company contact the county agricultural extension agents. Ex. 344 at 5; Ex. 345 at 46. Rene LeGault, the chief design engineer of the plant, has seen most of the aluminum' reduction installations in the world yet hasn’t visited any other such plant located in a rich fruit-growing region. Asked whether the surrounding land uses were an important siting consideration, he responded: I don’t think that the fume control systems that you are referring to were ever designed in view of fruit growing areas or agriculture areas in that extent. The design of the technology was to take care and recouperate [sic] the fluorine evolution from the cells ... for recycling purposes. Ex. 336 at 56. He had “no knowledge of what would be emitted from that plant.” Ex. 336 at 81. No one ever told him; he never asked. Ex. 336 at 82. In 1971 the Renken arbitrators stated that the plant “was inappropriately located.” Arbitrators’ Findings, Decision and Award 2 (1971). The company’s decision to locate the plant in a productive agricultural region did not in itself violate its societal obligations. One must not be blind, however, as to the choice of location when evaluating the responsibility of the company to society to avoid damage, if possible, to its neighbors. The amount of damage to others may well bear a rough relationship to the size of the expenditure needed to prevent or reduce it. Thus, the company’s original decision to build the plant in The Dalles, while not a ground for recovery here, is nonetheless relevant to evaluation of its conduct during the 1965-71 claim period. Having located the plant near vulnerable or apparently vulnerable crops, the company was required to take damage to those crops into consideration when selecting and carrying out an emission control strategy. 2. Monitoring. a. Emissions. (1) Frequency of Monitoring. The company conducted emissions testing in October and November of 1964 and in early 1965, then suspended any further testing until 1971, when it became required by state authorities. Ex. 339 at 134-35; Ex. 348 at 23; Ex. 935a at 32-33; TR 365-66, 668, 902-03. 9 Nor did any Oregon government agency test the plant’s emissions during the 1965-71 claim period. TR 600-01, 623. Even after the state monitoring requirement went into effect, the company did not adopt any written testing schedule, Ex. 339 at 153, and did not report its emissions during periods when the secondary control system was inoperative. TR 669. (2) Accuracy of Monitoring. (a) Selection of Monitoring Location. Byrne noted that releases from the secondary control system are emitted along almost a mile length of roofs at the plant and that the results of emission testing can *1004 be influenced by the location on the roof from which the sample is taken. Ex. 339 at 132-33. Consulting engineer Aaron Teller doubted the accuracy of 1971 test results submitted by the company to the Oregon Department of Environmental Quality (DEQ) because most of the data came from the secondary control system fans located near the sides of the emission outlet. There is an obvious maldistribution of flow in the buildings, and from thermal lift it is easily inferred that the maximum concentrations in the emissions would be in the center of the building. [The outlets with the highest emissions] are not measured at all by [the company’s] data submitted to [DEQ]. Ex. 303 at 2. After the resumption of testing in 1971, the company routinely sampled only 10% of the primary control system towers and 5% of the secondary system exhaust fans. Ex. 339 at 151. Raymond Rooth, a designer of pollution control systems for aluminum plants for Norsk Viftefabrikk Co., characterized the aluminum industry’s practice of sampling the emissions from just a few scrubbers and then calculating the overall release from the plant as “not good.” TR 570. He questioned the accuracy of short-term emission tests dependent upon a few samples, the results of which can be influenced by selection of the scrubbers to be tested to eliminate those with plugged nozzles or screens. Ex. 932a at 157-58. The most obvious deficiency in the company’s selection of monitoring locations, however, was its failure ever to measure the amount of fluorides escaping through the doors and louvers in the cell buildings. Nor did the tests performed for EPA sample from these outlets for uncontrolled emissions. Ex. 1100 at 43. (b) Duration of Tests. Rooth stated that he did not “believe in any exact emission data representative for a specific smelter without doing a real long-time measurement.” Ex. 932a at 162. As for the plant’s November, 1964 test showing total fluoride emissions of 640 lb./day [approximately 2.5 lb./ton], Rooth found the result to be within the range he would have expected but noted that he had no great confidence in the plant’s short-term measurements as an accurate indication of its long-term average emissions. Ex. 932a at 162-63. He expressed skepticism over the company’s 1974 report to the Oregon DEQ that the plant was emitting only an average of .68 lb./ton. . .. [According to my judgment there exists very few possibilities on a long-term range to judge how much emissions you have out of an aluminum smelter without having installed an amplifying monitoring system along the roof. According to my knowledge, Martin Marietta does not have such a recording system. TR 526. (c) Maintenance of Normal Operating Conditions. Rooth noted that the results of short-term emission tests can be influenced by' avoiding testing when the plant’s releases are enhanced by anode effects or “sick” pots. Ex. 932a at 159. During the tests conducted in 1964 and early 1965 Byrne did not necessarily check to ensure that the cell temperatures and bath ratios were maintained at normal operating parameters. Ex. 339 at 133-34. Consulting chemical engineer Joseph Schulein learned that the company had reported a result of 640 lb./day [about 2.5 lb./ton] for its November 11, 1964 test, about half the 1300 lb./day the company had stipulated to emitting as of 1963. Because he had received no indication that the plant had changed equipment or procedures, Schulein decided to investigate the matter by checking the plant’s use of electric power during the time of the new emission tests and for the months immediately preceding and following. TR 326. When Schulein asked to see the power use records, company comptroller Fred Blatt asserted that no records were kept. *1005 Byrne attributed this recalcitrance to Blatt’s “delusions of grandeur” in his self-appointed role as “superspy 007.” TR 858-59. After later having obtained the records through the company’s legal counsel, Schulein ... found that for the period during which the test was run from somewhere around 10 in the morning until noon, two-hour period, on the 11th of November, as shown on these charts, there were considerable periods when the voltage was so low it was off the charts ... and at other times for a period of a half an hour or so, the. voltage was down considerably from its normal course .... [I]t was a condition that I did not find existing in this magnitude at any time anywhere near the period of time covered by these tests, and I looked backwards and forwards on the charts. I concluded that it would be reasonable to believe that the effect of the reduction of voltage in current on the pot could cool the pots sufficiently to increase the crust on the top of the pot, thereby diverting a higher proportion of the total emissions into the primary system, that it would increase the viscosity of the bath because of the lower temperature, that it would reduce emanations due to the single fact of the current being reduced. I felt that these things combined could reasonably account for the reduction from an average of 1300 pounds to the 640 pounds found in the test. ... [T]he 1964 test ... was the last significant test done by Harvey Aluminum or Martin Marietta until practically the end of the period which I am addressing in my testimony [1965-71].... ... Those [power] records quickly and clearly demonstrated that the company was not operating its plant in a manner which would allow accurate testing or which would permit the company to replicate those results under any normal operating practices and procedures followed by the company. Ex. 348 at 32-35. The plant’s chief operator confirmed that the electric power to potlines 1 and 2 had been shut off at times during November 11, 1964, due to the testing of capacitors at the plant that had been planned a few days in advance. He received no notice that Byrne was going to have the emissions measured on that day. Ex. 337 at 1570-75. Schulein concluded that the emission control system at the plant could not have achieved the 640 lb./day result had the plant been operating at full power. Ex. 348 at 35. In response to Schulein’s accusation that the November 11, 1964, emission test had given artificially low results due to reduced power to the cells, Byrne stated that similar tests conducted by the company during October, 1964, and April, 1965, showed similar results at normal operating power levels. TR 838. The November 11 test result was 68% fluoride removal efficiency for the secondary control system; the October, 1964, result ostensibly was 57% but was flawed by the absence of secondary system input data. The April, 1965, result was also 57%. Ex. 339 at 140. Byrne suggested that the tests conducted by the EPA contractor in late 1971 and in 1972 were an accurate reflection of the plant’s emissions during the 1965-71 claim period, because the plant’s operations and control system did not produce substantially different emissions levels in 1972 compared with 1965. TR 840. Even if this assumption were accurate, however, the tests conducted on behalf of EPA were not performed with standardization of bath ratios or pot temperatures. Ex. 1100 at 41-42. The consulting engineers’ report states: The process during testing from October 2,1972, to October 5,1972, was stable. Very little difference was noted in either stack velocity or stack temperature at the sampling sites. Nothing unusual was done to the reaction cells during testing. All normal stud pulls and crust breaking were carried out. Ex. 778 at 34. *1006 There is no indication in the report that bath ratios, pot temperature or amperages were checked. (d) Sampling Technique. Rooth stated that isokinetic sampling is necessary to ensure that the testing obtains a representative sample of particulates in the emission stream. Ex. 932a at 21. Isokinetic sampling for particulates was also recommended by Teller and Harold Zeh, former chief chemist at the Reynolds-Trout-dale aluminum plant. Ex. 303 at 3; Ex. 1103 at 17-19, 21; ZTR 7. EPA tests were conducted isokinetically on emissions from the primary control system outlets but not isokinetically for the secondary control system outlets. “Isokinetic sampling would substantially reduce the impinger collection efficiency because of low gas velocities at the sampling location.” Ex. 903 at 17. The company’s emission tests were not performed isokinetically. Ex. 1103 at 20; ZTR 9. The sampling probe after the wet electrostatic precipitator pulled in only about a third of the proper volume of air and only a third of the actual amount of escaping particulates. ZTR 11-13. Based upon the evidence presented, however, I am unable to estimate the magnitude of error introduced by the absence of this technique, nor can I conclude that this was a substantial flaw in the company’s emission testing program. Teller suggested that the 1971 results submitted by the company to the Oregon DEQ may have been compromised by the coolness of the filter used to trap fine particulate fluorides for measurement. If the filter ... is not heated to a sufficiently high temperature (above 250 °F) the relative quantities of gaseous and particulate [fluoride] can be significantly distorted, masking the gaseous effluent. This would be true for both the roof and skirt sampling system. Ex. 303 at 3. In a paper presented at the 1973 AIME meeting, Rooth stated: Under the prevailing conditions after a wet scrubber, especially when operating on gas from pots with prebaked anodes, we have experienced errors and discrepancies of the apparent HF-quantity in the absorption bottles ranging up to several hundred percent when using uncontrolled filter temperatures.... The filter temperature must be kept well under control by temperature measurement in the gas stream just after the filter. The increase in temperature should preferably be in the range of 0-30 °C (0-54 °F). Ex. 932a-2 at 332. There is no indication that the company’s tests were conducted with a sampling train that measured the temperature of the air stream just after the filter. The EPA tests used a thermometer after the filter but also after an intervening impinger. Ex. 905 at 47. There is no indication that any of the measurement techniques used at the plant included the regulation of filter temperature. Like the absence of isokinetic sampling, the lack of filter temperature regulation may result in distorted readings of fluoride emissions, but the evidence presented does not justify a finding that this constituted a major source of error in the company’s emission testing program. (3) Overall Evaluation. The company could certainly have improved the frequency with which it monitored its emissions and probably could have improved the accuracy of the results. The 1974 EPA report described a secondary emissions roof monitoring system then in use that drew continuous samples from multiple outlets and then performed isokinetic sampling on the aggregated stream of air, providing results “more representative than those from the single point sample.” Ex. 903 at 18. A computerized continuous emission analyzing system was installed at the Reynolds-Hamburg prebake plant in 1974, the Árdal og Sunndal Verk- Sunndalsdra plant in 1976 and at plants operated by Alcoa and New Zealand Aluminum Smelters, Ltd. Ex. 932 Attachment A; TR 565- *1007 66. Teller in 1972 suggested the use of sonic orifices at each outlet to direct samples to a common heated manifold and then through the EPA sampling train. Ex. 303 at 3. Finally, the company could have sampled the emissions escaping from the cell building doors and louvers or could otherwise have calculated the amount. In sum, the company made no vigorous effort to determine the amount of fluorides it was releasing during the 1965-71 claim period, essentially abandoning this method of checking the effectiveness of its pollution control strategy. While the lack of emission monitoring in itself does not constitute a breach of the company’s societal obligations, it does show the company’s lack of respect for public welfare, a tendency evidenced primarily by the deficient control of emissions from the plant, a subject of subsequent discussion. b. Ambient Concentrations. Monitoring of emissions is a direct method of determining the effectiveness of the plant’s pollution control strategy. Monitoring of ambient pollutant concentrations in the surrounding orchards provides less direct evidence of emission restraint but more directly measures the impact of the plant’s emissions upon its neighbors by accounting for meteorological factors that influence the distribution of pollutants in the atmosphere. Ambient air monitoring is no substitute for at least occasional direct sampling and analysis of plant emissions as a means of evaluating pollution control performance. Ambient monitoring is nevertheless useful as a more accurate indicator of the magnitude of the trespass suffered by the plant’s neighbors. The company monitored the concentration of hydrogen fluoride gas in its own orchard during each April from 1963 through 1971. In the years 1963-67 and 1969-70 most of the measurements found a concentration of 0.1-.9 parts per billion (ppb), roughly equivalent to 0.1-.9 mg/m3. In 1968 and 1971 the readings clustered in the range of less than 0.1 ppb. Ex. 701a, col. 1. MCAES researchers monitored ambient atmospheric hydrogen fluoride concentrations in three orchards in the vicinity of the plant during the spring and summer of 1963 and reported daily concentrations as high as 12.9 mg/m3 (Gilbert orchard, March 28)-substantially higher than the maximum daily concentration of 3.3 ppb measured by the company in its orchard (April 2). MCAES found 15-day average concentrations as high as 5.7 mg/m3 (Gilbert orchard, March 28-April 11). Ex. 122 at 34-35. In 1964 MCAES expanded its air sampling to nine orchards, reporting a high daily concentration of 2.5 mg/m3 (Meyers orchard, May 26) and a high weekly concentration of 0.6 mg/m3 (Kroon orchard, May 22-28). Ex. 123 at 4-5,19. In contrast, the company reported a maximum daily concentration of only 1.2 ppb (Starlight orchard, April 7). Ex. 701a, col. 4. In 1965 MCAES monitoring in 5 orchards showed maximum daily concentrations ranging from 1.16 mg/m3 (Gilbert orchard, April 9) to 3.31 mg/m3 (Hendricks orchard, April 18). Ex. 124 at 24. In contrast, the company reported a maximum of 1.2 ppb in its orchard on April 7. Ex. 701a, col. 4. In 1966 MCAES monitored in 10 orchards and found maximum daily concentrations from 0.77 mg/m 3 (Bailey orchard, April 5), to 4.92 mg/m3 (Kroon orchard, August 21). The Kroon orchard experienced an average concentration of 2.02 mg/m3 during the period from March 29 through April 7; the Meyers orchard experienced an average of 1.75 mg/m 3 during that period. Ex. 124 at 26. In contrast, the company reported no concentrations in its orchard exceeding 1.0 ppb. Ex. 701a, col. 1. In 1967 MCAES monitored 10 orchards and reported maximum daily concentrations ranging from 0.47 mg/m3 (Renken orchard, June) to 4.65 mg/m3 (Fleck orchard, April). The Fleck orchard experienced an average concentration of 2.23 mg/m3 during the *1008 April 1-7 period. Ex. 124 at 27. In contrast, the company reported no concentrations in its orchard exceeding 1.0 ppb. Ex. 701a, col. 1. The air monitoring committee established by paragraph 6 of the Renken consent decree, however, recorded 13 readings in excess of 1.0 ppb at its six sampling stations in the neighboring orchards. Ex. 701a, col. 2. Similarly, in 1968 and 1969 the ¶ 6 committee recorded higher concentrations at its sampling stations than did the company in its orchard. See Ex. 701a. That the company found consistently lower levels of gaseous fluoride in its orchard than others found in other orchards in the vicinity should have alerted the company that the ambient concentrations in its orchard were not necessarily representative of those throughout the immediate area. I conclude that the company’s monitoring program, limited to its own orchard until the ¶ 6 committee assumed responsibility in 1967, was too limited in scope to provide a reasonably accurate picture of the trespass of its fumes into the neighboring orchards. Its failure to undertake a more comprehensive effort has contributed to my conclusion that the company did not fulfill its societal obligations during the 1965-71 claim period. 3. Controlling Emissions. The amount of fluoride the plant released into the atmosphere depended upon the amount of fluoride evolved at each cell and the efficiency of the primary and secondary emission collection and treatment systems. The Martin Marietta plant uses vertical stud Soderberg (VSS) reduction cells. Other aluminum reduction plants elsewhere in the United States and the world employ this type of cell, or horizontal stud Soderberg (HSS) cells, or “prebake” cells, or some combination of these three basic types. A somewhat simplified description of the construction and operation of the VSS-type cell, illustrated by Figure 1, follows. *1009 The cathode is at the bottom of the cell. Above the cathode is the bath, a mixture of molten cryolite and various additives that dissolves the raw ingredients, alumina and fluoride. Centered above the bath is the anode, which is surrounded by a casing. “Studs” extend into the anode from above. The anode covers only part of the bath. A “skirt” of metal attached to the anode easing extends out from the casing and down to cover part of the exposed bath, thus creating a space similar to a tunnel around the outside of the base of the anode, just above the bath. Under stable operat *1010 ing conditions, the part of the bath that is left exposed by the anode and skirt hardens to form a crust. Alumina is piled onto this crust so that it covers the crust from the edge of the reduction cell to the skirt, sealing off from the pot room air the tunnel created by the skirt and the bath. During the aluminum reduction process, electrical current passes through the studs, the anode, the bath and the cathode. The current reacts with the bath, which contains the dissolved alumina and fluoride, causing aluminum to be formed. The aluminum settles to the bottom of the bath and is drawn off. Carbon dioxide, tars and fumes containing, among other things, gaseous fluoride and fluoride particles enter the tunnel created by the skirt and alumina and are drawn off at a low velocity into the primary emission control system. Alumina must be added, or “fed," to the bath periodically. During feeding, the cell operator breaks the crust and sweeps the alumina that was sitting on top of it into the bath through the opening. More alumina is then placed on top of the crust. During the time that the crust is broken, the bath is exposed and fluoride-bearing fumes are released into the pot room. Fumes are also released into the pot room through any fissures in the crust and because of imbalances in cell operation that occur when there is insufficient or overabundant alumina in the bath. When there is too little alumina in the bath, gas builds up at the working face of the anode, creating resistance to the passage of electricity. This gas buildup at the anode is called a “light” or “anode effect.” This resistance and associated increased power consumption and increased temperature cause the cell to overheat, which can result in melting of the crust. Because it is easier for the gases to escape through this thinner, melted crust, a greater amount of fumes escapes the primary collection system and enters the pot room than during stable operating conditions. The cell operator corrects an anode effect by breaking the crust, piercing the buildup with a lance and adding more alumina. More fumes escape into the pot room during this correction process than during normal feeding. During an imbalance that is caused by too much alumina in the bath, the alumina that is not dissolved settles to the bottom of the cell, coating the cathode and creating resistance to the passage of current. The current then seeks to exit the cell through the side rather than the bottom. This overheats the bath, melts the crust and allows an increased amount of fumes to escape into the pot room. a. Fluoride Evolution at the Cell. (1) Operating Parameters. Zeh suggested that the plant could have reduced the evolution of fluoride in its cells by increasing the bath ratio to 1.45-.50 and decreasing the amperage to 91-92,000 amperes. Ex. 1103 at 34. A bath ratio of 1.50 evolves 50% less fluoride than a ratio of 1.25. ZTR 18. The company did not experiment with higher bath ratios or lower cell temperatures that might have reduced fluoride evolution. Ex. 339 at 84-85. Lars Ryssdal, plant superintendent, 1964 ■ 70, and plant manager, 1970 72, defended the company’s use of a bath ratio in the range of 1.25-.30 as permitting operation of the cells at lower temperatures despite the higher amperages (102,000-107,000 amperes) employed at the plant from 1965 through 1971. He stated that adjustment of the bath ratio, temperature and amperage would not have reduced fluoride evolution from the level experienced during that period. Ex. 930 at 8-12. When the plant’s bath ratio temporarily increased to about 1.48 during 1970 due to a shortage of A1F3, the cell temperatures actually increased. Ex. 930 at 10. Nevertheless, Ryssdal did not deny that the cells could have been operated at an acceptable temperature by reducing the amperage while increasing the bath ratio, as suggested by Zeh, and he conceded that an increased bath ratio, from 1.25 to 1.40, along with reduced amperage would have reduced fluoride emissions from the cells and that the cells at the plant could have *1011 been operated at a 1.46 bath ratio at 92,000 amperes. Ex. 930a at 25; TR 460, 463. Ryssdal also indicated, however, that this procedure would have had an adverse effect upon the plant’s aluminum production and that the total amount of fluoride evolution per ton of aluminum produced would not necessarily have been decreased. TR 464, 466. A reduction in the plant’s productive capacity is no less a cost of pollution control than the capital and operating expenses of emission control devices. From the evidence presented I am unable to find that the company could efficiently have reduced its fluoride emissions by adopting the operating parameters suggested by Zeh, except perhaps during the spring fruit tree blossoming period when fluoride-caused damage was particularly acute. The evidence pertaining to the possibility of reducing fluoride evolution and emissions by adjusting the plant’s bath ratio, amperage and cell temperature is insufficient to permit a conclusion that the company disregarded its societal obligations by failing to make such adjustments. (2) Selection of Ore. Byrne explained the increase in the plant’s reported emissions from 640 lb./day in 1964 to 750 lb./day in 1972 by noting that the ore in use at the plant at times during 1971 and 1972 created problems in the operating parameters and resulted in increased emissions. The plaintiff later suggested that the company made no effort to control the selection or use of ore in order to control emissions. Plaintiff’s Opening Post-Trial Memo 17. The evidence is insufficient to support a conclusion that selective use of various ores could have efficiently reduced the plant’s harmful emissions. b. Primary Emission Control System. (1) Collection Efficiency. The primary system collection efficiency is expressed as the percentage of fluorides emanating from the pots in gaseous or particulate form that are captured by the skirt system and directed to the primary system’s emission treatment devices. Because the primary control system can remove fluoride very efficiently from this fluoride-saturated stream of air, the primary system collection efficiency is of prime importance to the overall level of fluoride emissions from the plant. (a) Cell Hooding. The plant was constructed in 1958 with a skirt system to collect fumes from the cells and direct them into the primary control system. Byrne estimated that the skirt system collected 70-85% of the fluoride emissions from the cells, although he had done no testing to confirm this estimate. Ex. 339 at 63-66. See Renken v. Harvey Aluminum, Inc., 226 F.Supp. 169, 171 (D.Or. 1963) (finding of 80% collection efficiency). The company did not attempt modifications to increase the efficiency of its primary emission collection system during the 1964-72 period. Ex. 339 at 125. The plaintiff suggested that the plant could have employed a VSS cell hooding system similar to that installed in 1971 at the Pechiney-St. Jeanne VSS plant, which in 1976 reported 90% collection efficiency. Ex. 932 at 12; TR 473. According to Rooth, other attempts to hood VSS cells have “failed as the hood constructions have disturbed the heat balance of the pot to an unacceptable extent.” Ex. 932 at 12. Although periodic crust-breaking operations require that this hood be lifted from the cell, this arrangement does capture fluoride emissions from crust fissures and from anode effects better than the skirt system at the plant. TR 475-76. LeGault testified in 1963 that VSS cells could be hooded except for the problem of heat and the necessity for working the pots. Ex. 336 at 113-14. In 1973 he stated that hooding a VSS cell was impossible. Ex. 336 at 121-22. In his 1966 deposition, A. J. Rice, former Alcoa chief metallurgist, noted that both prebake cells and HSS cells have been hooded but that hooding of VSS cells was, in his opinion, impractical. Ex. 937 at 2664-72. His first objection was that the hood would *1012 have to be periodically removed in order to break the crust and add alumina to the bath. Ex. 937 at 2664-65. Thus, the hood would not be efficiently collecting cell emissions during this operation. The hood would, however, be in place over the cell at all other times and would collect emissions from crust fissures and anode effects that would otherwise escape into the potroom atmosphere. In addition, the hood would upset the heat balance within the cell. Ex. 937 at 2666-67. The Montacatini plant in Italy attempted to hood its VSS cells in the years following World War II, but the attempt apparently was abandoned. Ex. 937 at 2676-78. Alcoa experimented with hooding nonPechiney type VSS cells in 1954 and again in 1961. Alcoa sought a mechanical method for adding alumina to the cell bath, but such an operation was hampered by the hard crust that developed on the surface of the bath. Alcoa attempted to hood the cells in order to increase the temperature of the crust and thereby keep it thin enough to accept additional alumina from an automatic feeder without necessity for mechanical puncture of the crust. Alcoa did not attempt to design a hooding system for the primary purpose of increasing the collection of fumes from VSS cells. Nor had Rice attempted to adapt a prebake type hooding system for use with VSS cells. Ex. 937 at 2679-2688. Thus, his 1966 conclusion that a VSS cell hooding system would be “impractical” may be viewed with some skepticism, in light of the apparent success of the Pechiney-St. Jeanne system installed in 1971. The plaintiff’s contention that the plant could have more effectively controlled its emissions by operating with primary collection system hoods during the claim period is very difficult for me to evaluate. Although the first VSS plant to implement such a system was the Pechiney-St. Jeanne plant in 1971, at or near the end of the claim period in this case, that does not necessarily mean that the technology was unavailable or could not have been adapted from the prebake or HSS hoods that had functioned for decades. 10 Based upon the evidence presented, however, I can only conclude that the Pechiney VSS hooding system was highly innovative and that the company did not violate its societal obligations by failing to implement this technology prior to the end of the 1965-71 claim period. (b) Operating Procedures. Rooth stated that, “due to the crust breaking and expected irregularities of operation, the long-run collection efficiency as a monthly or yearly average is between 60 and 80 percent.” Ex. 932 at 12. “Thirty percent [escaping] is quite representative.” Ex. 932a at 179. Because Rooth had not visited the plant before 1973, however, he could not testify as to its collection efficiency during the 1965-71 period. Ex. 932a at 69-70. Byrne stated that the plant experienced about one anode effect per day for every two cells, or about 150 anode effects per day in the plant’s 300 total cells. Ex. 339 at 73. Ryssdal testified that in the “early days” the plant’s cells experienced anode effects every day or even every shift but that the period between anode effects had been “stretched out to a week or at times even longer” in any particular cell by the institution of work schedule changes in September, 1962, January, 1965, and January, 1968. Ex. 930a at 20; Ex. 930 at 6. According to the company, the number of anode effects per cell per day decreased as follows: *1013 1962 1.75 1966 .59 1969 .42 1963 1.14 1967 .56 1970 .32 1964 .78 1968 .55 1971 42 1965 .66 Schulein suggested that better maintenance of the cells could reduce emissions into the potrooms. TR 309-11. Schulein visited the plant in 1965 and 1966 and prior to his testimony before the arbitrators in 1970 and 1971. 11 TR 341, 358. “[A]t the times I was there and observing the pots, I definitely noticed they should have had more manpower.” TR 317. He observed no substantial change in cell maintenance by 1971. TR 359. Zeh also suggested that “real tight operating procedures” could have reduced fluoride emissions into the potrooms, but he admitted that he had not actually observed the plant’s operating procedures during the 1965-71 period. Ex. 1103 at 34-35. Rooth noted that the Japanese have claimed 90% collection efficiency for the Sumitomo system of VSS cell operating procedures. Ex. 932 at 12. Byrne placed the efficiency at 95%. Ex. 935a at 34. The Sumitomo procedures maintain better heat balance and cell stability by closely controlling the addition of alumina feed, which keeps the cathode clean and the side freeze from intruding under the anode and obstructing the passage of electric current. Ex. 930a at 30. After 1975 the company began to consider purchasing the Sumitomo know-how; an order was placed in March, 1977. Ex. 935 at 16. Ryssdal expects these procedures to be implemented at the plant by the end of 1979. Ex. 930 at 13-14. Based upon the testimony of Schulein that his inspections of the potrooms during the 1965-71 period revealed inadequate cell maintenance and the fact that other plants have been able to increase their primary system collection efficiencies through careful operating procedures, I find that the company could efficiently and economically have increased the proportion of cell emissions captured and directed into the primary control system. I do not agree that the company had to wait until aluminum producers in Japan implemented these procedures and obtained 90% primary collection efficiency. (2) Treatment Efficiency. Having been captured by the skirt system, the pot emissions are directed to the primary treatment system, which may consist of cyclones, multicyclones, wet or dry scrubbers, wet or dry electrostatic precipitators, bag filters or other devices. The plant was constructed in 1958 with a primary treatment system consisting of a burner at each cell to combust carbon dioxide and tar emanating from the cell, a multicyclone centrifugal dust collector, a humidifier, a wet spray scrubber tower and three 40-horsepower fans to draw cell emissions through the system. The plant substituted three 60-horsepower fans in 1960, added a bubbler at the base of the scrubber tower in 1961 and a second burner at each cell in 1962. Ex. 780a. The company tried out various additional primary emission control techniques between 1963 and 1970, including a Venturi scrubber, bag houses, a ping-pong ball scrubber, a ceilcote cross-flow scrubber and a Peterson separator, but found nothing practical or suitable. Ex. 935 at 13-14; TR 831-37. These devices were tested as prototypes in small installations, not by hooking up the entire plant. Ex. 339 at 96-97. LeGault was familiar with the Rheinfelden primary treatment system including electrostatic precipitation described at the 1962 AIME meeting which he attended. Ex. 336 at 71. In 1963 he testified to his knowledge of three aluminum plants that had installed such systems. Ex. 336 at 96, 112, 124. No such device was incorporated into the plant’s primary system before or during the 1965-71 claim period. The company completed installation of wet electrostatic precipitators in the primary control system in March, 1972, Ex. 780d, and began its practice of disengaging the spray scrubber tower except when the precipitators were being serviced. Ex. 911 at 11. *1014 Rooth concluded in 1977 that the state of the art permitted 99 + % efficiency in removing total fluorides from the primary emission stream. Ex. 932 at 15. This same high level of removal was practical in 1967 and even earlier. Ex. 932a at 88. If you go back to ... Figure A, you will see that some smelters have kept their primary systems unchanged since 1950..... [Scrubbing] out pot gas from a vertical stud Soderberg plant ... has been done with a very high degree of efficiency all the time. Ex. 932a at 88-89. Rooth stated that the plant’s wet electrostatic precipitation system effected a reduction in fluoride emissions, but he could not quantify the reduction. Ex. 932a at 57-58, 163-64. Electrostatic precipitators had previously been installed at the Hytte-Rheinfelden VSS plant in 1959, the Pechiney-Noguerre VSS plant in 1961, the Ardal og Sunndal Verk Ardal VSS plant in 1962 and Sunndalsdra VSS plant in 1969, and the Mosjden Aluminiumswerk VSS plant in 1972. In 1963 Judge Kilkenny stated his belief in . .. the feasibility of the introduction of electrostatic precipitators for the removal of the minute or small particulates which are not removed by the other processes. ... The great weight of the evidence points to the conclusion that the installation of the cell hoods and the employment of electrostatic precipitators would greatly reduce, if not entirely eliminate, the escape of excessive material now damaging the orchards of the plaintiffs. Renken, 226 F.Supp. at 172. The October 1955 Stanford Research Institute study proposal to the company noted: ... A number of methods may be employed to eliminate, or minimize the effects of, pollutants contained in pot gases collected by hoods. Among those are scrubbing with water, bag filtration, electrostatic precipitation, combustion (for the destruction of tar fog) and discharge from a tall stack. Ex. 220d at 2. Zeh noted that electrostatic precipitators and bag houses (dry systems) have been available for 15-20 years but were not installed at the plant due to cost considerations. Ex. 1103 at 29-30; ZTR 21. Rooth noted that dry systems were installed at the Graenges-Sundsvall VSS plant in 1973 and the Árdal og Sunndal Verk-Árdal VSS plant in 1974. Ex. 932 at 20. Schulein suggested that the exhaust from the primary treatment system could have been run through the secondary system for an additional increment of emission reduction. TR 311-12. I need not consider this proposed stacking of primary and secondary systems, nor the dry systems suggested by Zeh. It is sufficient to establish a breach of societal obligations that the company failed to utilize a wet electrostatic precipitator during this period. New EPA isokinetic tests of the primary system after installation of the wet precipitator in 1972 showed fluoride releases from the primary outlet ranging from .0006 lb./ton of aluminum produced (October 2-3) to .027 lb./ton (October 4-5). The six test sessions indicated an average primary system emission of .012-016 lb./ton. The company’s own 1972 tests showed a comparable average of .028 lb./ton. Ex. 904 at 11-13. Under the assumptions that the cells emanated 46 lb./ton, and that the skirt system collected 70% of these emissions, the primary treatment system would have been receiving fluoride at a rate of 32.3 lb./ton. EPA measured receipt of 36.43 lb./ton. Ex. 903 at 23. The average primary system outlet emission of about .014 lb./ton indicates that the system was achieving 99.95% removal efficiency. Elsewhere the company in 1975 reported an average primary system emission of .02 lb./ton and a primary system fluoride removal efficiency of 99.7% during March through December, 1972, after installation of the wet electrostatic precipitators. Ex. 310 at 12. The EPA itself calculated 99.30-99.97% removal efficiency. Ex. 903 at 23. In sharp contrast, an EPA test conducted prior to the installation of the wet electrostatic precipitator revealed fluoride emissions from the primary outlet of 1.0-1.4 lb./ton, Ex. 903 at 29, a removal efficiency *1015 of only 96-97%. In addition, Byrne stipulated in 1971 that the plant’s primary control system was achieving 99.75% efficiency in removing gaseous fluoride but only 85% efficiency in removing fluoride entrained on particles, Ex. 339 at 178, which may constitute 25% of total fluoride emanating from the cells. Ex. 932 at 9-10. If so, the primary outlet was releasing in gaseous form about 0.2% of total fluoride entering the primary system [75% X .25% = .1875%]; in particulate form about 3.75% of total fluoride entering the primary system [25% X 15% = 3.75%]. Thus, the overall treatment efficiency of the primary system in 1971 was, according to Byrne, about 96%^-which corresponds very closely with the 96-97% figure calculated by the EPA. In sum, the primary treatment system technology employed at the plant during the 1965-71 period removed about 96-97% of the fluoride emissions captured by the primary collection system, even though wet electrostatic precipitator technology had been employed in other VSS plants since 1959 and, when finally installed at the plant in 1972, achieved 99.3%-99.97% removal efficiency. Failure to implement this reasonably efficient and economic technology resulted in an unnecessary additional fluoride emission from the primary control system outlet of about 1 lb./ton and amounted, in my judgment, to a breach of the company’s societal obligations. c. Secondary Emission Control System. (1) Collection Efficiency. The 20^40% of emissions from the cells that are not captured by the skirt system and directed into the primary control system are released into the potrooms. Until 4-5 months prior to the end of the 1965-71 claim period the plant’s secondary control system wet scrubbers, installed in 1963 and located in the roof of each building, would treat only those heated potroom pollutants that rose up to the roof by convection. The remaining emissions would escape the pot-rooms through louvers and doors in the walls of the buildings. The secondary system’s collection efficiency expresses the percentage of the potroom pollutants that were actually treated by the system’s wet scrubbers there; the remaining potroom contaminants escaped without treatment. As of 1965 the plant had the only secondary emission control system in a VSS aluminum plant that did not employ forced draft fans to direct the emissions escaping from the skirts to the secondary system’s scrubbers. Ex. 339 at 99-104; Ex. 932 at 20. The potrooms at the plant were ventilated by large louvers, and fluoride escaping from the skirts could also escape to the atmosphere through these openings in the buildings. Byrne stated his belief that fluorides actually “escaped into” the potrooms through these louvers due to the high concentration of fluoride in the plant’s courtyards caused by the local settling of the cooled emissions from the primary control system stacks, although no tests were conducted to confirm this belief. Ex. 339 at 127-28. The plant’s chief engineer George Youngmeister noted that the plant had forced-draft fans drawing air into the cell buildings through the floors but that this air, after becoming laden with emissions from the cells, could escape through the cell buildings’ doors and louvers; the collection efficiency of the secondary system was basically dependent upon thermal convection. Ex. 334 at 663, 675. The fans feeding air in through the floors were removed or turned off when fans were added to the secondary control system itself. Ex. 334 at 676-78. Teller visited the plant during the winter of 1964-65 and inspected the secondary emission control system. Ex. 335 at 11. The spray-screen system then in place was not “a highly efficient system, relative to other possible methods in 1963, for the recovery of hydrogen fluoride from a gas stream like [that emanating from the plant].” Ex. 335 at 14-15. Due to the absence of an induced air fan, not all of the fluoride emissions escaping the skirt collection system arrived at the spray-screen secondary system in the roof of each building. “[T]he inference I made from the data was that all the gas was not being treated at all times. In other words, a lot of the gas was *1016 going out the louvers.” Ex. 335 at 19. He calculated that at times only a third of the fluoride emissions from the cells was reaching the roof secondary control system. Due to insufficient irrigation, the screens became plugged, “forcing more of the air out the sides [of the buildings] rather than through the [secondary] system.” Ex. 335 at 22. Schulein concluded that the emission control system operated by the plant during the 1965-71 period “was far below the best available, in concept, to achieve the minimization of fluoride emissions.” Ex. 348 at 22. The collection efficiency of the plant’s secondary system as of May, 1972, can be derived from data presented in the 1972 report of Valentine, Fisher & Tomlinson (Ex. 777) and volume 1 of the 1974 report of EPA (Ex. 903). The EPA contractor’s testing of the plant’s pollution control systems in May, 1972, found a secondary control system fluoride intake of 2.5-3.28 lb./ton; EPA later calculated the average secondary system fluoride intake at 2.85 lb./ton. Ex. 777 at 16; Ex. 903 at 23 (Plant D). Assuming that the cells emanated 46 lb./ton, Ex. 932a-l at 19; Ex. 1102 at Table 1, and that the primary system collected 36.43 lb./ton, Ex. 903 at 23, the secondary system was collecting only 30% of the 9.57 lb./ton escaping into the potrooms. The November, 1974, report on ventilation at the plant by Rooth’s firm, Norsk Viftefabrikk, indicated that polluted air continued to leave the cell buildings through the upper part of the large doors, a situation that could have been corrected by reducing the side wall and door openings. Ex. 932a-5 at 5, 10. This contradicted the earlier conclusion of Byrne that fluoride “escaped into” the potrooms through these openings and the subsequent statement of Zeh that, during his visit to the plant in 1978, the flow of air was primarily from outside the buildings in through the open louvers. Ex. 1103 at 27. The Norsk Viftefabrikk report also noted that the clogging of the demisters prevented efficient suction of the potroom pollutants into the secondary scrubbing system. Ex. 932a-5 at 22. I find this November, 1974 report to be relevant to the pollution control efforts of the company during the 1965-71 claim period, because the company has not sought to introduce any evidence that its plant ventilation practices deteriorated from 1965 to 1974. For the same reason I also find relevant this statement of Zeh after his 1978 visit to the plant: Well, at The Dalles it seemed that the haze inside the potrooms sort of drifted and hung, like it wasn’t being sucked out very forcibly. That was not the case at Goldendale. The draft was quite substantial, and there wasn’t that hanging of the haze in the potroom. Well, it made me wonder for a while whether the roof fans were even on. Later in the day we found out that they were on, but there still was not the movement of air in the potrooms at The Dalles as there was at Goldendale. Ex. 1103 at 28, 29. See ZTR 20-21. The percentage of potroom pollutants reaching the secondary scrubbers could have been increased by the efficient use of forced-draft fans to suck in the potroom air. The Pechiney-Noguerre VSS plant installed such a forced-draft system in 1959, as did the Hytte-Rheinfelden VSS plant in 1959, the Mosjden Aluminiumswerk VSS plant in 1963, the Graenges-Sundsvall VSS plant in 1963, the Alnor -KarmeSy (Norsk Hydro) VSS plant in 1967, and the company’s plant in November, 1970, the system becoming operational in 1971. Ex. 932 at 20. The Reynolds-Troutdale prebake plant installed a roof spray control system in 1946 and in 1948 began converting to a 200-fan forced-draft roof system that became operational in 1950. TR 339, 364. Instead of utilizing this more effective forced-draft method, the company in 1963 installed a roof spray-screen system without fans. In his August 4,1966, deposition, Schulein suggested the installation of a forced-draft secondary system in a horizontal configuration, which is what the company imple *1017 mented in 1970-71. TR 349-50. Schulein stated that the forced-draft secondary control system that the company began to install in 1970 was “old”; “the concept and the technology were known in 1965.” Ex. 348 at 23. Byrne admitted that the 1970-71 system could have been implemented in 1961 or 1963. Ex. 339 at 142-49. By 1971 this scrubbing technique had been known for 20 years. TR 913. Youngmeister didn’t know why the company hadn’t installed the. forced-draft tunnel control system earlier; the design was not “exotic.” “We just didn’t think of it.” Ex. 334 at 672. He knew about every element in the system as early as 1959. Ex. 334 at 674. The system was not installed until 1971 because Byrne and others at the company believed that the fluoride pollution problem had been “solved” by the system implemented in 1963. Ex. 339 at 136; TR 860. LeGault attributed the company’s failure to install the 1970-71 system earlier to the “limitations of creativity of this domain.” Ex. 336 at 93. “That knowledge was not available.” Ex. 336 at 94. Rooth stated his “opinion that the addition of fans in the potroom gas cleaning system would not have had a significant effect on the efficiency of emission control at The Dalles plant.” Ex. 932 at 21. The meaning of Rooth’s phrase “efficiency of emission control” was clarified during cross-examination. A. Basically, the fans installed in scrubbing systems do not increase the scrubbing as such. Q. You would concede, would you not, Mr. Rooth, that the installation of fans in a secondary control system that depends solely upon [convection] for sucking a fluid into the system would improve the chances of that system taking in as much as it could get? A. That’s right. You would secure the volume going through the scrubber by fans; that is right. TR 499, 523-24. Rooth also agreed with the statement that the secondary system installed at the plant during 1970 “improved its pollution control system as to fluoride emissions” by providing easier access to the scrubbers by placing them in the courtyard and by “improvpng] the possibility of keeping good control, better control of the scrubber technology compared to what you have it in the roof.” Ex. 932a at 56-57. He agreed that “with such an improvement the actual emissions would be reduced,” but he could not quantify the reduction. Ex. 932a at 57, 163-64. I conclude that the system’s record of collecting only about a third of the potroom emissions during Teller’s visit immediately preceding the 1965-71 claim period and during the EPA contractor's 1972 tests could have been improved during the 1965-71 period by the installation of forced-draft fans, by regular maintenance to prevent clogging of the screens and by closing at least some of the cell building doors and louvers from which up to two-thirds of the potroom emissions were escaping untreated into the atmosphere. Because about 30% of the total fluoride emanating from the cells was released into the potrooms, even a small improvement in the efficiency of the secondary collection system could have substantially reduced fluoride emissions from the plant. For example, if the secondary control system had collected 60% of the potroom emissions instead of 33%, the magnitude of untreated emissions escaping from the plant would have been reduced from about 9.2 lb./ton [46 lb./ton X .30 to potroom air X .67 escaping untreated] to about 5.5 lb./ton [46 lb./ton X .30 to pot-room air X .40 escaping untreated]. Failure to implement reasonably efficient and economic forced-draft fan technology until the very end of the 1965-71 claim period constituted a violation of the company’s societal obligations. (2) Treatment Efficiency. During the 1965-71 period the plant’s secondary control system consisted of water sprays and screens in the cell building roofs that had been installed in 1963. A problem with roof gutters was corrected by lining *1018 them with fiberglass in 1964. TR 828-30. This secondary control system was not utilized during the months November through February of each year from 1963 through 1970 because of concern that the scrubbing system would freeze during the cold weather. Ex. 339 at 126; Ex. 435 at 12; TR 860. The scrubbers initially removed about 67-70% of the fluorides entering the system. Renken, 226 F.Supp. at 171. The efficiency of this system declined after 1963 due to increased emission of smaller fluoride particles from the cells. When operational, the system’s fluoride treatment efficiency varied from 21 to 68% during tests conducted by the company in 1964 and 1965, which included other readings of 28%, 33%, 42%, 54% and 57%. Ex. 339 at 137-41. As previously noted, the company did not test its emissions again until 1970. Following his visit to the plant in 1964-65, Teller concluded that the roof scrubbers were removing about 25-35%, sometimes even 50%, of the fluoride in the air passing through the system. Ex. 335 at 20-21. The efficiency of the system was diminished because it used only one-tenth the appropriate amount of water for the spray. Ex. 335 at 21-22. Rooth confirmed that such systems can be hampered by clogging of the water spray nozzles and the screens by tar and dust from the potroom air. Ex. 932 at 17. Rooth explained that 75% of the fluoride escaping into the potrooms is in gaseous form, while the other 25% is fluoride entrained on particles, about half of which are coarse (2 microns dihmeter and larger) and the other half fine (smaller than 1 micron diameter). Ex. 932 at 9-10. In low pressure wet [secondary] scrubbers the [gaseous] hydrogen fluoride will be cleaned out to a reasonable efficiency. Nearly all the coarse dust will be scrubbed out but almost all of the fine submicron dust will pass through because it is so fine.. .. As a monthly or yearly average, not more than 80% efficiency on total fluoride has been achieved. Despite regular maintenance, water failure or clogging may occur. Secondary scrubbers normally give 80% or below as efficiency on a long-run average. ... At 10 inches of water pressure loss, only approximately 20% of the fine particulate fluoride will be collected, and this scrubbing pressure will give a prohibitive energy consumption for the scrubber with the vast gas quantities in question. (At 10 inch water gauge approximately 1550 kWh/ton of aluminum.) Ex. 932 at 15-16. The 80% of fine particulate fluoride remaining represents about 10% of the fluoride in the potroom air-the equivalent of about 3% of the total fluoride emanating from the pots, assuming a primary system collection efficiency of 70% [.30 (potroom emissions) X .25 (fluoride on particles) X .50 (submicron particles) X .80 (untreated by secondary scrubber) = .03 (3%)]. In other words, 3% of the fluorides generated in the plant would be emitted raw from the secondary system, even if the secondary system were to treat all of the pollutants in the pot-rooms and were to be maintained at 10 inches of water pressure loss. Rooth concluded that present technology permits 80% removal of fluorides by a secondary scrubbing system in a VSS plant. Ex. 932a at 170. This degree of treatment efficiency has not increased appreciably since 1958, when the Mosjden Aluminiumswerk plant was built incorporating a secondary scrubbing system. This plant has presumably been able to achieve about 80% total fluoride removal efficiency with its secondary system, as has the Pechiney - Noguerre plant since 1959, the Hytte- Rheinfelden plant since 1959, the Graenges-Sundsvall plant since 1963, the Alnor-Karmdy (Norsk Hydro) plant since 1967 and, theoretically, the company’s plant since 1963. Ex. 932 at 20; Ex. 932a at 165-66, 172. A plant can achieve the 80% secondary system removal efficiency only if its system is properly maintained. Ex. 932a at 170. Byrne stated after the fans had been instituted in 1970-71 that the plant’s secondary scrubbing system removed fluoride gas with 88% efficiency and fluoride on *1019 particles with 65% efficiency. Ex. 339 at 178. He estimated the overall fluoride removal efficiency of the secondary system at 75%. Ex. 339 at 148. This corresponds roughly to the 72% efficiency reported by the company for 1972, Ex. 310 at 12, and represents a considerable improvement over the 1964-65 company findings of 21-68% efficiency and Teller’s 1964-65 appraisal of 25-35% efficiency. The improvement achieved by the new secondary system was confirmed by EPA measurements taken in late 1971 and in 1972 showing emissions from the secondary system outlets ranging from .643 lb./ton to 2.93 lb./ton. The nine test sessions indicated an average secondary emission of .806-2.05 lb./ton, comparable to the company’s own reported 1972 average test result of 1.37-.52 lb./ton, Ex. 904 at 14-18, but somewhat less than the 2.86 lb./ton reported elsewhere by the company for 1972. Ex. 310 at 12. It appears that the fluoride removal efficiency improved appreciably after installation of the tunnel system in 1970-71, despite Byrne’s contention that the new system did not reduce the overall level of emissions from the plant. TR 831, 840. Byrne himself stated his familiarity with a paper by H. Schmitt 12 that described the Hytte-Rheinfelden plant’s secondary treatment system consisting of roof-mounted sprays, screens and fans that achieved, according to Schmitt, 90% efficiency in absorbing gaseous fluorine. Ex. 339 at 123. Teller stated that a spray scrubber can achieve 99% efficiency in removing gaseous fluoride if operated with a sufficient flow of water through the sprays. Ex. 335 at 28. I conclude that the company failed to fulfill its societal obligations when it delayed adopting the forced-draft fan, more efficient secondary control system until 1970-71. I need not consider the suggestion of Hatchard that the company could have achieved 95-99% fluoride removal efficiency in its secondary system as early as 1961 by utilizing electrostatic precipitators, bag houses and high-energy Venturi scrubbers therein. Ex. 346 at 11. d. Overall Performance. A comparison of the actual emission control system in operation at the plant during the 1965-71 period with an improved system that the company could have been operating during that period is presented in Figure 2. From the data presented in this lawsuit, my best estimate of the amount of fluoride that the plant was emitting per ton of aluminum production is about 12.5 lb. (Figure 2:A). Implementation of reasonably efficient and economic systems previously described in this opinion could have reduced those emissions to about 2.5 lb./ton (Figure 2:B)-an overall reduction in fluoride releases of 80%. A preponderance of the evidence indicates that the company could have adopted these systems before 1965, with the possible exception of the careful cell operating procedures required to achieve 90% primary system collection efficiency. The company in 1977 purchased the Sumitomo know-how, which may constitute a significant, non-obvious innovation in VSS cell operation technique beyond the imaginative capability that could reasonably have been expected from the company before or during the 1965-71 period. If careful cell operations short of the Sumitomo procedures could have increased primary collection efficiency to 80%, then the plant’s fluoride emissions could have been decreased to about 5.1 lb./ton (Figure 2:C)-an overall reduction of 60%. And even if the possibility of reducing emissions through careful cell operating procedures is ignored, the company still could have decreased its fluoride releases to about 7.6 lb./ton (Figure 2:D)-an overall reduction of 40%. This somewhat mechanical, mathematical approach to the company’s emission control performance is supported by the less precisely expressed opinions of various witnesses. *1020 Barney McPhillips, former chairman of the Oregon Environmental Quality Commission, stated that fluoride emissions from the plant could have been limited to 1 lb./ton but that, starting as of 1965, the company had not voluntarily minimized its emissions *1021 to the fullest possible extent. MTR at 10, 14-15. Richard Hatchard, chief of the state air pollution control office in the Oregon Board of Health and State Sanitary Authority during 1952-64, concluded that the damages incurred during the 1958-73 period were “unnecessary ... because adequate controls were not installed.” Ex. 346 at 14. “I have every confidence that if they sought the services of an experienced design firm that a system could have been designed and installed in The Dalles in 1961. . . . ” TR 266. The emission control systems eventually installed at the plant subsequent to the 1965-71 claim period “were available for many years before the plant came into existence.” Although one or two or maybe three different vertical stud Soderberg plants in the United States didn’t have some of these systems, still the technology was there on other plants and other industrial processes which required pollution control. The basic technology was available. Ex. 346 at 284. Reid Iverson of EPA stated that as of 1972-73 the plant was “one of the better controlled in the United States” due to its combination of primary and secondary systems. Ex. 1100 at 26. Iverson’s paper presented at the 1973 AIME Metallurgical Society meeting showed that the plant achieved the highest overall fluoride control efficiency of any plant listed, Ex. 1102 at Table 1, based upon EPA’s 1972 — 73 emission testing. The company’s plant was the only VSS plant listed. The EPA report itself confirmed that the plant had the most efficient primary system but indicated, however, that the plant’s emissions from the secondary system exceeded those from the prebake plants. Ex. 904 at 8-9. And EPA did not “determine what percentage of potroom gases or emissions were not captured by either the primary or the secondary system.” Ex. 1100 at 43. In other words, EPA did not measure the amount of fluoride escaping through the cell building doors and louvers. Finally, Iverson replied in the negative when asked whether an increase in plant fluoride emissions from 640 lb./day in 1964 to 750 lb./day in 1972 indicated that the plant had been utilizing the “best achievable technology.” Ex. 1100 at 53. Frederick Skirvin, supervisor of the Oregon Department of Environmental Quality’s Air Quality Control Division, testified that during the 1967-71 period the company “was a leader in emissions control operations and technology,” TR 619, though he had never visited the other VSS plants in Montana, Texas, Tennessee, British Columbia or Mexico, TR 620, nor did he appear particularly familiar with the relevant literature. TR 654-56. The preponderance of the technical and evaluative testimony supports the conclusion that the company did not implement reasonably efficient and economical pollution control measures that could have substantially reduced the fluoride emissions from the plant during the 1965-71 period. 4. Mitigation Measures. It is possible that the adverse effects of emissions from the plant might have been reduced by a strategy of mitigation, including the installation of tall stacks to propel the fluorides through the occasional atmospheric inversion layer and the spraying of susceptible fruit with a calcium chloride solution. a. Tall Stacks. The Stanford Research Institute 1955 proposal to the company “for an investigation of potential air pollution conditions in the vicinity of your future aluminum reduction plant at The Dalles” noted among the methods for minimizing the effects of pollutants “discharge from a tall stack.” In fact, the proposal stated: ... If particulate fluorides and tar fog are present in [pot room] air to an excessive degree, the most feasible method of reducing their influence in the vicinity may be the installation of a tall or high-velocity stack for the discharge of ventilating air. Ex. 220d at 2. Despite this early advice, LeGault stated that the company did not explore the tech *1022 ñique of using tall stacks to prevent the fluoride emissions from reaching the nearby orchards. Ex. 336 at 126. b. Application of Lime Spray. In 1961 MCAES researchers applied to J. H. Hale peach trees at The Dalles Experiment Station a spray of calcium chloride (lime) concentrated at 2 lb. per 100 gallons. Trees not sprayed showed 26.8% soft suture, while those sprayed twice showed 23.3%, sprayed three times showed 5.2%, and sprayed four times showed 4.5%, even though the fluoride content of the peaches increased with the number of spray applications. Ex. 121 at 23. The report noted, however, that this additional measured fluoride content “may represent surface contamination and not absorption per se.” Ex. 121 at 5. Three or four applications of the calcium chloride spray appeared to decrease the incidence of soft suture by a factor of about five in J. H. Hale peaches. Bailey testified that the company’s horticultural consultant Benson had reported no reduction in soft suture in peaches receiving the lime spray treatments. TR 380. As recently as 1977 Scholes stated his belief that lime spray prevents soft suture but that it is no longer necessary in light of present insignificant incidence of soft suture. Ex. 933 at 4. The 1966 Renken consent decree relieved the company of any obligation to compensate the orchardists for soft suture caused by fluoride emissions from the plant following the 1966 growing season “... unless the orchard owner involved sprays his peach orchards with a calcium chloride or lime spray or other spray, as designated from time to time by [the company], in accordance with methods prescribed by [the company], for which spraying [the company] agrees to pay the reasonable cost. In 1971 the claim arbitration panel made awards to the orchardists for damage to peaches even though they had not “always sprayed their peaches with calcium chloride as they had agreed to in the consent decree.” Arbitrators’ Findings, Decision and Award at 2. The Oregon Supreme Court in Byers v. Santiam Ford, Inc., 281 Or. 411, 574 P.2d 1122 (1978), stated, “Whether the plaintiff has adopted a conciliatory attitude in resolving the controversy is not relevant in determining if defendant should pay punitive damages.” 281 Or. at 417, 574 P.2d at 1125. Thus, the orchardists’ failure always to spray their peaches does not excuse the conduct of the company in causing the injury to their peaches. Nevertheless, I find that the company’s agreement to provide the orchardists with the calcium chloride spray and to pay the reasonable cost of its application to peach trees to have been constructive and in conformity with the company’s societal obligations. 5. Conclusion. The company did not fulfill its societal obligation to adopt and maintain reasonable, efficient pollution control measures. Having located the plant in a rich agricultural district, the company did not diligently monitor the plant’s emissions nor the ambient concentrations of fluoride in the surrounding orchards. Nor did the company implement before or during the 1965-71 claim period efficient available methods for reducing the emissions. In particular, the company did not adopt cell operating procedures to minimize the escape of fumes from the primary collection system until several years after the end of the claim period, did not install wet electrostatic precipitators in the primary treatment system until 1972, and did not utilize a forced-draft secondary collection system until 1971. All of these measures could have been taken prior to the start of the claim period and would have substantially reduced the plant’s emissions of fluoride, perhaps by as much as 80% and at least by 40%. In addition, use of tall or high-velocity stacks might have prevented the occasional concentration of emissions beneath the atmospheric inversion layer. The company’s sponsorship of calcium chloride spraying of peach trees was laudable but not sufficient to overcome the preponderance of evidence showing that the com *1023 pany faltered in carrying out its social responsibility to control its harmful emissions. C. Arranging to Compensate for the Remaining Harm. As the court in determining the propriety of a punitive damage award may consider evidence of harm by the defendant’s conduct to persons other than the plaintiff, see Orchard View Farms, Inc. v. Martin Marietta Aluminum, Inc., No. 73-3080, slip op. at 1-2 (9th Cir. June 23, 1975), so should the court take note of the defendant’s efforts to neutralize that harm by voluntary payment of compensation, even though this compensation did not extend to the damage for which the jury in this case made a compensatory award. In 1961 the company agreed to compensate peach and apricot growers in the vicinity of the plant for soft suture damage to their peaches and for apricot leaf necrosis believed to inhibit tree growth. TR 795-800. According to Byrne, the company paid out about $100,000 in settlements during the 1961-66 period. TR 887, 889-90. On November 3, 1966, the United States District Court entered a consent decree settling the claims of 15 orchardists filed against the company in federal court and providing for the dismissal of 17 other cases then pending in the Oregon state courts. The company agreed to . . . pay each of the plaintiff peach orchard owners the then prevailing market price of his peach fruit which has been or is made unmarketable by soft suture caused by fluorides emitted from defendant’s plant; provided, however, that the defendant will have no such obligation for any soft suture occurring in the future unless the orchard owner involved sprays his peach orchards with a calcium chloride or lime spray or other spray, as designated from time to time by defendant, in accordance with methods prescribed by defendant, for which spraying defendant agrees to pay the reasonable cost. Consent Decree ¶ 2. For damage other than soft suture the company agreed to ... pay to the respective plaintiffs such amounts as may be necessary to compensate them for past or future economic damage (other than soft suture in peaches) in their respective orchards, caused by fluorides emitted from defendant’s plant; subject, however, to the terms and conditions herein stated, including arbitration pursuant to paragraph 10. Consent Decree ¶ 3. Paragraph 10 of the decree provided for a three-member claim arbitration panel, one member selected by the company, one by the orchardists and the third member by the other two, to settle claims according to Oregon statutory provisions governing arbitration. On February 11, 1971, the claim arbitration panel awarded the 15 orchardist plaintiffs a total of $942,305 for damage to their crops and trees during the 1960-69 period. On February 19, 1972, the panel awarded $120,900 to five orchardists for damage during 1970 and 1971. The orchardists then terminated the arbitration agreement, and no further voluntary settlements have been reached. The company’s agreement to recognize that the plant’s emissions were damaging the orchards and to compensate the orchardists for the damage under an arbitration arrangement is to be complimented, and future such agreements to be encouraged. Such conduct is strong evidence that the company was attempting to fulfill its societal obligations by accounting for the damage its operations were causing to its neighbors. Though laudable, this conduct does not entirely shield the company from punitive damages liability, for it came about after some eight years of the plant’s operation and after the company was faced with numerous lawsuits claiming damages. As the Oregon Supreme Court noted in Byers v. Santiam Ford, Inc., 281 Or. 411, 574 P.2d 1122 (1978): In the case here at issue the evidence of contrition and a conciliatory attitude of one of defendant’s agents after the complaint was filed has scant relevance *1024 respecting the state of mind of other agents of defendant at the time the car was sold to plaintiff. Assuming the evidence established the good faith and good will of defendant’s president toward plaintiff, such conduct came as response to the complaint, which prayed for substantial punitive damages. The evidence shows a desire to “buy peace” and minimize the risk of an award of punitive damages and not that defendant dealt in good faith with plaintiff in selling the car. 281 Or. at 417, 574 P.2d at 1125. Unlike Byers, however, in this case the company’s agreement to arbitrate and compensate came during the period for which punitive damages are sought, not entirely after the events giving rise to liability. The Renken consent decree was entered after the 1965 and 1966 growing seasons .but before the 1967-71 seasons. Therefore, I consider the arbitration agreement relevant to the determination of punitive damages liability in this case. If the company during the 1965-71 claim period had cooperated fully in ascertaining the harm from the plant’s emissions and in effecting some combination of efficient emission control combined with compensation for the remaining harm, I would rule against the plaintiff’s request for punitive damages. The company’s participation in the arbitration system is certainly indicative of corporate social responsibility but is insufficient to overcome its failure in the other two respects. IV. AWARD OF PUNITIVE DAMAGES. I am satisfied by the evidence in this case that an award of punitive damages is appropriate for the earlier portion of the claim period. It is difficult to put a precise date on the watershed of the company’s conduct showing a sufficient compliance with societal obligation so as to rule out punitive damages. In this regard, I rely heavily upon the testimony of Barney McPhillips 13 , who may almost be regarded as the father of Oregon’s pollution control progress. While his testimony was generalized, and did not contain any particular dates, nonetheless it furnishes more than adequate support for a finding that midway through the 1965-71 claim period a change occurred in both the attitude of the company and its efforts to carry out pollution control measures so as to behave like a good neighbor. One cannot look at a single event alone, since the attitude of society (both private and governmental) was in a state of substantial change. And as society’s attitude changed, as was evidenced by the movement toward a more careful attention to the earth around us and the necessity of its preservation, so also did society’s laws and regulations, and with that the response of its components-both of the antagonists here, aluminum company and orchardist. I conclude, therefore, that a punitive damage award is available for the claim years 1965 through 1968. If the claim years here were only 1969-71, I would not award punitive damages. By this time-the late 60s and early 70s-on the record in this case, it cannot be said that the company was in sufficient disregard of its societal obligations so as to be liable for punitive damages.- And while, of course, the only period of time before me in this case is 1965-71, and the only orchards involved are the ones of this plaintiff, nonetheless it is difficult to see how any claim for punitive damages would succeed as to any period of time in and after about 1969, in view of *1025 all of the developments during the 1969-71 period and since that time. Indeed, though of course I need not and do not decide the question, it seems most unlikely that a punitive damage claim for any period of time after 1971 based on the record in this case would even go to a jury. Because the company did not cooperate in ascertaining the nature, severity and scope of the harm inflicted upon the plaintiff by the plant’s emissions or in arranging to prevent this damage or to neutralize it through voluntary compensation arrangements, the company is liable to the plaintiff for an award of punitive damages. Previous judicial opinions provide little guidance as to the proper amount of such an award. Courts often state that such an award should be sufficient to deter continuation or repetition of the offending conduct. In this case the offending conduct was the company’s refusal either to implement economically efficient emission control measures or voluntarily to compensate the plaintiff for the damage caused by the plant’s emissions. Thus, punitive damages should be awarded in an amount that will deter this and other companies from attempting to impose a portion of their costs of production upon their neighbors by compelling those damaged by the emissions to resort to the uncertainties of the legal process in order to obtain compensation. Under the circumstances here, I believe an appropriate and measured award for punitive damages is $200,000 for the claim period here through the year 1968, but none thereafter. The foregoing constitutes findings of fact and conclusions of law, pursuant to Rule 52, Fed.R.Civ.P. Footnotes [1] Organizing a plaintiff class is hindered by the fact that the benefit of -a successful lawsuit against the polluter for compensation is not limited to the plaintiffs. Persons damaged by the pollution but not contributing to the legal action also benefit due to the collateral estoppel effects of the initial lawsuit in subsequent actions and because the first plaintiff or group of plaintiffs has already done the work of organizing some relevant evidence and locating experts willing to testify. Thus, each person damaged by the pollution has an economic incentive to let someone else bring the first lawsuit and then to take a “free ride” or at least a discount excursion to obtaining his own compensation. [2] “TR” refers to the transcript of the 1977 retrial. “MTR” refers to the separately-numbered transcript of the testimony of Barney McPhillips presented later in 1977. “ZTR” refers to the separately-numbered transcript of the testimony of Harold Zeh presented in 1978. [3] The company submitted as exhibits only those few pages of these reports (Exs. 501-07) containing remarks primarily applicable to the plaintiffs orchards. [4] Information received by the company about fluoride injury to fruit in orchards other than the plaintiffs can nevertheless provide notice that the plant’s emissions are causing harm that should be eliminated through pollution controls or be compensated for. See Orchard View Farms, Inc. v. Martin Marietta Aluminum, Inc., No. 73-3080, slip op. at 1-2 (9th Cir. June 23, 1975). [5] The notes of Scholes, Treshow and Blodgett submitted as Exs. 502d, 50 lh and 503f-k do not cover the Francois and Curtis orchards. [6] The 1961 study finding stylar dimple and a leaf fluorine content of 119 ppm in cherry trees sprayed with ammonium fluoride did not utilize the procedure of washing the leaves prior to crushing and analysis. N.or did the MCAES surveys of foliage fluorine contents in The Dalles area during the years 1953-62. As noted earlier, the MCAES researchers began to wash the leaves prior to crushing and analysis in 1963, so a direct comparison between cherry leaf fluorine levels associated with stylar dimple in the 1961 experiment and cherry leaf fluorine levels in the area around the plant after 1962 is not possible with the data at hand. [7] Due to the sketchiness of the supporting evidence, this conclusion does not rely upon the alleged incidents of Byrne’s interference with federal funding for additional MCAES studies or his alleged attempts to have extension agent Thienes fired and to have extension agent Smith’s 1968 report on cherry blossom petal browning and cupping suppressed. See Ex. 339 at 169-70; TR 154-55, 205. [8] In a later statement Byrne stated that he did not know the concentration of fluoride required to produce soft suture. Ex. 935a at 26. At the 1977 retrial, however, Byrne agreed that 1 mg/m 3 would be sufficient to cause soft suture “under some conditions, maybe.” TR 871. [9] The company hired a consultant, Richard Hatchard, to test the efficiency of the primary treatment system in 1966. TR 366. [10] No sound reason has been advanced by defendants why hoods, similar to those employed by Reynolds, should not be installed. While it is true that a substantial portion of the gasses and particulates escape at the time when the new aluminum ore is being introduced into the pot or the liquid metal is removed, I am convinced that such an escape could be prevented by a properly designed hood over the open area. I agree with the expert that, after the installation of the hood, the small amount of gasses which might escape on the introduction of ore or the removal of liquid metal would be inconsequential. Renken v. Harvey Aluminum, Inc., 226 F.Supp. 169, 171-72 (D.Or.1963). [11] Prior to being reminded of the dates of his visits to the plant, Schulein erroneously testified that he had not been there during the 1965-71 period. See TR 317, 341. [12] H. Schmitt, the Fluorine Problem in Aluminum Plants, in 2 Extractive Metallurgy of Aluminum 93 103 (1963). [13] McPHILLIPS: [A]fter the plant was purchased by Martin Marietta we seemed to get more cooperation than we did from the Harveys. MTR 19. THE COURT: [W]ould you say that the attitude of the company, the policy of the company, could be described before 1968 as being not that of a good citizen and after 1968 that of a good citizen? McPHILLIPS: Your Honor, I would say that there was a definite improvement. MTR 30. THE COURT: But would it be fair to say that, that as of about that time, there seemed to be a better attitude on the part of the aluminum company? McPHILLIPS: Yes. I would say so. THE COURT: And you are talking now roughly of 1967, 1968? McPHILLIPS: Probably in that area. MTR 31.