The "Shotgun Wedding" of Science and Law: Risk Assessment and Judicial Review Devra Lee Davis* During its environmental heyday, Congress passed a number of laws which may be thought of as "basic science-forcing." Includ- ing the Clean Air Act Amendments of 1970' and the Federal Water Pollution Control Act Amendments of 1972,2 these laws authorize agencies to take regulatory action on the basis of agency findings made at the very frontiers of scientific inquiry. 3 To make such findings, public administrators increasingly rely on * Executive Director, Board on Toxicology and Environmental Health Hazards, Na- tional Academy of Sciences; B.S., University of Pittsburgh (1966); M.A., University of Pitts- burgh (1967); Ph.D., University of Chicago (1972); M.P.H., Johns Hopkins University (1982). The research for this article was conducted for the Integrated Environmental Management Program of the United States Environmental Protection Agency while Dr. Davis served as Science Policy Director of the Environmental Law Institute. Mary Ellen Scheckenbach and Harry Lewis served as research associates. Michael Gruber, Jeffrey Trauberman, Philip Reed, Gordon Apple, Richard Merrill, Tom McGarity and Sara Davis Buss provided critical review. The author would like to thank Victor Miller for his substan- tial editorial assistance. The views expressed herein are not necessarily those of the Envi- ronmental Protection Agency, the Environmental Law Institute, or the National Academy of Sciences. 1. Clean Air Amendments of 1970, Pub. L. No. 91-604, 84 Stat. 1676 (1970), codified at 42 U.S.C. §§ 7401-7642 (1982). 2. Federal Water Pollution Control Act Amendments of 1972, Pub. L. No. 92-500, 86 Stat. 816 (1972) (amending 33 U.S.C. §§ 1251-1376 (1982)). 3. See, e.g., 33 U.S.C. §§ 1251(d), 1252, 1254, 1361 (1982); 42 U.S.C. §§ 7403(b), 7404(b), 7407(c), 7408, 7601 (1982). For a summary of provisions in the major environ- mental statutes which authorize agencies to regulate at the frontiers of scientific knowl- edge, see infra Appendix I, Environmental Protection Agency, Chemical Substances Designation (1981). 68 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 a variety of scientific techniques for assessing risks to human health.4 Many risk assessment techniques are highly speculative, and almost all rely upon multiple assumptions of fact, some of which may be entirely untestable. 5 The administrative agencies' reliance on risk assessment engen- ders a fragile and uneasy partnership 6 between science and law, 7 -a veritable "shotgun wedding" according to former EPA Administrator William Ruckelshaus.8 Administrative agencies are committed to the traditional scientific paradigm as the basis for finding facts, but cannot rely on it to select the best regulatory action. Questions of fact may be stated in the language of sci- ence, but administrative decisions based on "fact" demand the exercise of judgment and discretion. Thus, agencies repeatedly find themselves forced to resolve scientific questions to which the scientific community has only incomplete answers. To the consternation of both scientists and judges, not science but the judiciary must rule on the "correctness" of an administra- tive agency's resolution of questions lying at the frontiers of sci- ence. 9 The proof of "correctness" required by a reviewing court differs both qualitatively and quantitatively from that which would be demanded in a scientific forum.' 0 Regardless of the applicable 4. The agencies' use of risk assessment in regulatory decisionmaking has been the sub- ject of considerable commentary by both scientists and legal scholars. See generally, QUAN- TITATIVE RISK ASSESSMENT IN REGULATION (1982); NATIONAL RESEARCH COUNCIL, RISK ASSESSMENT IN THE FEDERAL GOVERNMENT: MANAGING THE PROCESS (1983). 5. But see infra, note 196 (proposed guidelines for the uniform application of risk assess- ment techniques to agency decisionmaking). 6. The term "uneasy parnership" originated with Judge Friendly, who used it to de- scribe the relationship between administrative agencies to which Congress has delegated decisionmaking of a legislative character, and the federal courts, which Congress has charged with the task of reviewing agency exercise of the delegated power. Associated Indus. of N.Y.S. v. U. S. Dep't of Labor, 487 F.2d 342, 354 (2d Cir. 1973). The term is equally appropriate as a description of the relationship between science and law. 7. Alvin Weinberg has coined the term "trans-science" to describe the interface be- tween these divergent sectors. Weinberg, Science and Trans-Science, MINERVA, April 1972, 209. 8. W. Ruckelshaus, Science, Risk and Public Policy, Speech at National Academy of Sci- ences 3, (June 22, 1983) (available in office of Columbia Journal of Environmental Law). 9. See, e.g., Lead Industries Ass'n v. EPA, 647 F.2d 1130 (D.C. Cir. 1980), cert. denied, 449 U.S. 1042 (1982) (noting that the task of review in a case involving complicated scientific questions is difficult for the court). 10. See infra notes 22-32 and accompanying text for discussion of the standards ofjudi- cial review. Scientists develop statements of probability at specified levels of statistical sig- nificance that a particular fact is likely to be true. Many toxicological studies set the level of statistical significance as that at which there are only five chances out of one hundred that a particular finding occurred due to chance. Risk Assessment standard of review, judges often find themselves engaged in a hands-on examination of incomplete scientific data regarding risks to human health. Not surprisingly, review of agency action under science-forcing laws has strained judicial expertise, and created a bull market, in certain courts, for law clerks with train- ing in the sciences." This article will examine how courts have faced the "shotgun wedding" of science and the law when reviewing agency decisions based on risk assessments.' 2 In particular, this article will ex- amine whether courts have applied standards ofjudicial review in risk assessment cases in a consistent fashion. Cases involving agency regulations of asbestos, polyvinyl chlorides, benzene, for- maldehyde, leaded gasoline, ozone, taconite ore, and dioxin will be discussed in detail. The discussion will highlight some major policy questions involving the "shotgun wedding": What proof of harm must the agency provide to support its regulation? Given that science can provide only probabilities of harm, how much certainty of risk can a reviewing court demand? And when must politically based policy decisions replace scientifically based risk assessments? Section I of the article will review basic risk assessment method- ologies. Section II will introduce the scope of judicial review under the "substantial evidence" and "arbitrary and capricious" standards. Section III will examine cases involving risk assess- ments subject to the substantial evidence and arbitrary and capri- cious standards of review. Section IV will examine how courts have reviewed risk assessments which have not been used by an agency to promulgate a regulation, but which are presented as 11. The late Judge Leventhal of the United States Court of Appeals for the District of Columbia Circuit suggested in an informal conference sponsored by the Environmental Law Institute in 1978, that judges should have specially-designated science clerks to assist them with technical matters lying beyond the bounds of traditional legal training. Judge Leventhal believed that the courts have a "central role in ensuring the principled integra- tion and balanced assessment of both environmental and nonenvironmental considera- tions in federal agency decisionmaking." Leventhal, Environmental Decisionmaking and the Role of the Courts, 122 U. PA. L. REV. 509, 555 (1974). He subscribed to the talmudic notion that judges should have some competence with which to review results. His colleagues, Judges Bazelon and Wright, disagree with this view. Judge Bazelon maintains that the very complexity of the evidence submitted makes it dangerously unreliable to involve " illiter- ate judges" in the review process. Ethyl Corp. v. EPA, 541 F.2d 1, (D.C. Cir.) (en banc) (Bazelon, J., concurring), cert. denied, 426 U.S. 941 (1976). 12. See also McGarity,Judicial Review of Scientific Rulemaking, 9 SCIENCE, TECHNOLOGY AND HUMAN VALUES 97 (1984). 19851 70 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 evidence by plaintiffs seeking to enjoin a particular polluting activity. I. RISK ASSESSMENT TECHNIQUES Toxicology, epidemiology and clinical research are the basic tools employed in assessing risks to human health.' 3 Toxicology attempts to predict future risks to humans by measuring the ef- fects of exposure to a suspect substance on test mammals such as mice and rats. Because toxicological research is conducted through controlled experiments, it produces findings, such as dose-response curves which are, in and of themselves, relatively precise. Extrapolating from animals to humans is more problem- atic.' 4 Epidemiologic research documents past risks in human populations exposed to toxic substances under "natural condi- tions." A major defect of epidemiological research is that it can rarely determine the precise level or "dose" to which a given pop- ulation has been exposed. 15 In theory, clinical research on human subjects would solve many of the problems encountered in both toxicology and epidemiology, but for obvious ethical and practical reasons, clinical tests are of limited use. Faulty design or sloppy implementation of all these investigative approaches can invalidate expensive and time-consuming studies. Finally, risk assessment models apply mathematical formulae to the raw data produced by toxicological, epidemiological and clinical research, in order to answer the question typically posed by the regulatory agency: What are the potential consequences of exposure at dif- ferent levels? II. THE SCOPE OF JUDICIAL REVIEW Environmental and health statutes generally confer broad dis- cretion on agencies to choose the means for determining the existence of hazards which would trigger regulatory action.' 6 Statutory descriptions of those triggering points may, themselves, be sufficiently broad to afford agencies wide discretion to make 13. See, e.g., NATIONAL RESEARCH COUNCIL, supra note 4. 14. See generallyJ. DOULL, C.D. KLAASSEN & M.O. AMDUR, CASARETr AND DOULL'S ToXI- COLOGY: THE BASIC SCIENCE OF POISONS (2d ed. 1980). 15. See generally A. M. LILIENFELD & D.E. LILIENFELD, FOUNDATIONS OF EPIDEMIOLOGY (1980); J.S. MAUSNER & A.K. BAHN, EPIDEMIOLOGY: AN INTRODUCTORY TEXT (1974); B. MACMAHON & T.F. PUGH, EPIDEMIOLOGY: PRINCIPLES AND METHODS (1970). 16. See, e.g., Occupational Safety and Health Act of 1970, 29 U.S.C. § 655(b)(5) (1982). Risk Assessment the determination. For example, section 6(a) of the Toxic Sub- stances Control Act requires the EPA to regulate chemicals when necessary to "adequately" protect against any "unreasonable risk of injury to health or the environment" posed by the chemical's manufacture, processing, use or disposal.' 7 The extent of an agency's discretion depends on the applicable scope of judicial review. Review may be specified either in the statute authorizing the particular agency action at issue,' 8 or in the Administrative Procedure Act ("APA").' 9 Under the APA, agency decisions made after "formal" or trial-type proceedings 20 must be supported by "substantial evidence in the record." 2' This standard has been defined as "more than a mere scintilla. It means such relevant evidence that a reasonable mind might ac- cept as adequate to support a conclusion. ' 22 "Informal" agency decisions, those not required by statute to be made on the record after a formal agency hearing, are set aside by a reviewing court if found to be "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 23 This standard is in- tended, in the abstract, to be more deferential than the substan- tial evidence standard, but, as applied, may involve just as rigorous an inquiry by reviewing judges. According to the Supreme Court's decision in Citizens To Preserve Overton Park v. Volpe, 24 a court applying the arbitrary and capricious standard must make a searching and careful "substantial inquiry" into the facts presented to the agency to determine whether the agency 17. Toxic Substances Control Act § 6(a), 15 U.S.C. § 2605(a) (1982). 18. See, e.g., Clean Air Act § 307(d)(9), 42 U.S.C. § 7607(d)(9) (1982). 19. Administrative Procedure Act, 5 U.S.C. §§ 551-559; 701-706 (1982). 20. 5 U.S.C. §§ 556, 557 (1982). Section 556 refers to hearings in front of an adminis- trative law judge, where interested parties may be present, have the right to submit evi- dence, and cross-examine witnesses; the transcript in the proceeding is the "exclusive record for decision." Section 557 provides that the agency must cotsider findings and conclusions proposed by interested parties, and rule on the proposals "on the record." All decisions including initial, recommended and tentative decisions are a part of the rec- ord and must include a statement of findings and conclusions, and the reasons or basis therefor. 21. 5 U.S.C. § 706(2)(e) (1982). 22. Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229 (1938). This definition was adopted prior to the APA, but has been held to apply to the substantial evidence standard under the APA, with the additional requirement that "substantial evidence" must be based on all the evidence in the record. See Universal Camera Corp. v. NLRB, 340 U.S. 474, 477 (1951). 23. 5 U.S.C. § 706(2)(a) (1982). 24. 401 U.S. 402 (1971). 1985] 72 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 considered all the relevant factors, or made a "clear error of judgment.''25 Whether the substantial evidence and arbitrary and capricious standards, as defined above, require reviewing courts to engage in different inquiries, at least, in the context of agency regulations based on risk assessments, is questionable. To determine whether an agency made "a clear error of judgment," a reviewing court essentially determines whether the agency had an adequate fac- tual basis for its decision. This inquiry is not far removed from that which lies at the heart of substantial evidence review: was the agency's decision based on evidence that a reasonable mind might accept as sufficient to support a conclusion? 2 6 Moreover, given the increased complexity of modern environmental deci- sionmaking, the record in an informal proceeding may be as ex- tensive as that generated under the requirements of formal rulemaking. 27 In an attempt to subject agency proceedings to greater and more uniform scrutiny, some major environmental statutes provide explicitly for substantial evidence review of agency decisions which are arrived at on the basis of proceedings which, under the APA, would be reviewed under the "arbitrary and capricious" test.28 A recent decision by the Supreme Court in Chevron v. United States,29 may set a precedent for greater deferential judicial review of all agency decisions. 30 The Supreme Court in Chevron held that an agency's regulation-subject to arbitrary and capricious re- view-which is based on a "reasonable" interpretation of statu- tory language should be upheld by a reviewing court absent 25. Id. at 415-16. 26. See also Associated Industries, 487 F.2d at 349-50 ("[Iun the class of cases in which the ground for challenging the agency action is the inadequacy of its evidentiary basis, it is difficult to imagine a decision having no substantial evidence to support it which is not 'arbitrary', or a decision struck down as arbitrary which is in fact supported by 'substantial evidence,' " citing Scalia & Goodman, Procedural Aspects of the Consumer Product Safety Act, 20 U.C.L.A. L. REV. 899, 935 n.138 (1973). 27. See, e.g., Lead Industries Ass'n v. EPA, 647 F.2d 1130, 1181 (D.C. Cir. 1980); Indus- trial Union Dep't, AFL-CIO v. Hodgson, 499 F.2d 467, 474 (D.C. Cir. 1974). 28. See, e.g., Occupational Safety and Health Act of 1970, 29 U.S.C. § 655 (1982); Con- sumer Product Safety Act, 15 U.S.C. § 2501 (1982). Such statutes have been referred to as "hybrid" statutes. See, Industrial Union, 499 F.2d at 473. 29. - U.S. -, 104 S. Ct. 2778 (1984). 30. See, e.g., Wald, Negotiation of Environmental Disputes: A New Role for the Courts? 10 COLUM.J. ENVTL. L. 1 (1985); and Reed, Three Strikes and the Umpire is Out: The Supreme Court Throws the D.C. Circuit Out of the Bubble Review Came, 14 ENVrL. L. REP. (ENVrL. L. INST.) 10,338 (1984). Risk Assessment evidence of Congress' specific intent on the meaning of such lan- guage, and given evidence of conflicting policies which the regu- lation is directed to promote. 3' The regulation at issue in Chevron was an interpretation of the Clean Air Act's definition of a pollut- ing "source" as constituting an entire industrial plant, rather than each pollution-emitting unit within the plant.32 What impact the Court's holding will have on judicial review of agency regulations based on risk assessments is uncertain. If nothing else, the Court's emphasis on the reasonableness of an agency's decision re- inforces the point made above that the arbitrary and capricious and substantial evidence standards often require a reviewing court to undertake the same kind of inquiry. III. CASE LAW ON RISK ASSESSMENT A. Regulating Hazards in the Home and the Workplace; the Substantial Evidence Test The Occupational Safety and Health Act of 1970 33 delegates broad authority to the Secretary of Labor and, through the Secre- tary, to the Occupational Safety and Health Administration ("OSHA"), to promulgate "occupational safety and health stan- dards" to protect American workers. 34 The Act defines such stan- dards as those required to establish working conditions and practices that are "reasonably necessary or appropriate to provide safe or healthful" places of employment. 35 The Act further di- rects that, when promulgating standards for "toxic materials" or "harmful physical" agents, the Secretary must "set the standard which most adequately assures, to the extent feasible, and on the basis of the best available evidence, that no employee will suffer material impairment of health or functional capacity." 36 These standards are to be developed on the basis of research, demon- strations, experiments and other "appropriate" information. 37 In setting the standard, the Secretary is to consider the Act's goal of attaining the highest degree of workplace safety and health, the 31. Chevron, 104 S. Ct. at 2783. 32. Chevron, 104 S. Ct. at 2782-83; Reed, supra note 30 at 10,338-40. 33. Occupational Safety and Health Act of 1970, 29 U.S.C. §§ 651-678 (1982). 34. 29 U.S.C. § 655(a) (1982). 35. Id. § 652(8) (1982). 36. Id. § 655(b)(5) (1982). 37. Id. 1985] 74 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 latest available scientific data in the field, the feasibility of the standards, and experience gained under health and safety laws generally. 38 Although the procedures required for the promulga- tion of OSHA standards fall short of those required for "formal" rulemaking under the APA, 3 9 the Act directs courts of appeals to apply a substantial evidence test in reviewing the Secretary's standards.40 In Industrial Union Dep', A.F.L.-C.LO. v. Hodgson,4 1 unions whose members were affected by the health hazards of exposure to asbestos dust challenged OSHA regulations of the atmospheric concentration of asbestos dust in the workplace. The health hazards posed by asbestos were a matter of concern even before OSHA came into being, and one of the agency's first acts was to establish an emergency standard to control atmospheric concen- trations of asbestos. 4 2 Under that standard, the maximum per- missible eight hour time-weighted average airborne concentration of asbestos dust had been limited to five fibers greater than five microns in length per milliliter of air.4 3 For its permanent standard, the Secretary reduced the permissible con- centration to two fibers, but retained the five-fiber limit for four years in order to give employers a grace period in which to meet the stricter limit. The unions challenged the four-year delay on two grounds: that the five-fiber standard endangered the health of employees and that the employers did not need four years to comply with the more protective standard.4 4 In assessing the health impacts of the Secretary's decision to allow four years before imposing the lower two-fiber standard, the court drew attention to the widely disparate conclusions from risk assessments conducted by experts with whom the Secretary had consulted. 4 5 The then Assistant Surgeon General of the 38. Id. 39. These procedures are detailed in 29 U.S.C. § 655(b)(1)-(4). See Synthetic Organic Chem. Mfrs. Ass'n v. Brennan, 503 F.2d 1155, 1160 (3d Cir. 1974)(discussing the scope of judicial review of OSHA regulation limiting employee exposure to ethyleneimine); Indus- trial Union, 499 F.2d at 474-75 (comparing "substantial evidence" review to review of a "legislative policy determination"). 40. 29 U.S.C. § 665(f), 667(g) (1982). 41. 499 F.2d 467. 42. Id. at 471. 43. 36 Fed. Reg. 23,207-08 (1971). 44. 499 F.2d at 479. 45. Id. ("The experts differed sharply in some of their opinions, but their responses are generally cautious and reflect deficiencies in available data."). Risk Assessment United States had stated that some not significant increase in the effects of asbestosis could be expected from the delay in imple- menting the two-fiber standard.46 The chief of medical research at St. Luke's Hospital in Cleveland had stated that no evidence supported the proposition that asbestos is per se carcinogenic. 47 Conversely, a research professor at Mount Sinai School of Medicine in New York had indicated that any concentration higher than two fibers could not be justified.48 Judge McGowan of the District of Columbia Circuit described the problems facing the court under the Act's substantial evi- dence standard of review. According to Judge McGowan, the agency's determination could not be reviewed under a substantial evidence standard as traditionally applied, because sufficient data had not been available upon which the agency could make a "fully informed factual determination.' 4 9 Because "no precise predic- tion of increased harm can be made at this time," Judge Mc- Gowan could not conclude whether the Secretary had erred in determining that the imposition of the two-fiber standard could safely be delayed for four years, or whether employees would be subject to an additional risk from continued exposure to the five- fiber concentration. 50 Moreover, the agency had been forced to make an "essentially legislative policy judgment" in the face of conflicting and inconclusive evidence. 5 1 Accordingly, the court's task was to determine not whether substantial factual evidence supported the agency's decision, but whether the agency had car- ried out its job "in a manner calculated to negate the dangers of arbitrariness and irrationality in the formulation of rules for gen- eral application in the future." 52 Such a test required, at a minimum, that the agency provide a careful identification of the policy considerations which affected its choices. 53 The Secretary's consideration of the chance that health hazards would result from continued exposure to the higher standard, and of the Act's "overriding concern" for the 46. Id. at 479 n.27. 47. Id. 48. Id. 49. Id. at 474. 50. Id. at 479. 51. Id. at 474. 52. Id. at 475, citing Automotive Parts & Accessories Ass'n v. Boyd, 407 F.2d 330, 338 (D.C. Cir. 1968). 53. 499 F.2d at 475-76. 19851 76 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 protection of employees' health, were deemed by the court to have been a sufficient indication of the policies supporting the Secretary's choice of a two-fiber standard. 5 4 However, the court found that the Secretary's decision to allow a four-year delay for all industries, regardless of the capacity of each industry to imple- ment the two-fiber standard immediately, was not supported by sufficient policy considerations. Rather, the record left "nagging questions-even for the inexpert observer" as to the reason and rationale for the agency's decision. 55 Industrial Union highlights the problem ofjudicial review which is presented when experts in a medical or scientific field draw sharply conflicting conclusions from risk assessment analyses. The District of Columbia Circuit's solution was a practical one and one which respected the agency's legislative role: having de- termined that the factual evidence was conflicting, the District of Columbia Circuit upheld those parts of the Secretary's decision which were sufficiently supported by policy considerations. 56 In Society of the Plastics Industry, Inc. v. Occupational Health and Safety Administration,57 the Second Circuit accorded similar defer- ence to OSHA regulations of employers producing vinyl chloride monomer (VCM) and its products, notably polyvinyl chloride (PVC). By the time the case was decided, thirteen workers ex- posed to VCM had died from angiosarcoma, a rare form of can- cer. 58 Thus, the workers' deaths lent a sense of urgency to the court's decision. Retired Supreme Court Justice Clark, sitting by designation, wrote the opinion for the Second Circuit, which upheld the regu- lation. At the outset, Justice Clark noted his agreement with Judge McGowan of the District of Columbia Circuit that "the traditional 'substantial evidence' test is almost impossible of ap- plication where, as here, the Secretary's decision is essentially leg- 54. Id. at 474. 55. Id. at 488. 56. See also Merrill, The Legal System 's Response to Scientific Uncertainty: The Role of Judicial Review, 4 FUNDAMENTAL AND APPLIED ToxICOLOGY S418 (1984). Merrill describes Industrial Union as characteristic of an era of "judicial reticence" to review agency decisions on health risks. Id. at S522. Merrill applauds the court's review in Industrial Union as a "frank recognition that where facts do not carry you all the way, judgment is needed, and that reviewing courts cannot expect certainty where science cannot provide definitive an- swers." Id. at S423. 57. 509 F.2d 1301 (2d Cir. 1975), stay denied, 420 U.S. 1002 (1975). 58. Id. at 1306. Risk Assessment islative in character." 59 Justice Clark explained that, where agency decisions are based on evidence lying at the frontiers of scientific inquiry, the court must essentially apply an arbitrary and capricious standard of review. 60 As background, the court detailed the "morbid chronology" of events associated with vinyl chloride. 6' A study conducted in the Soviet Union in 1949 had first found liver irregularities in rats and rabbits subjected to VCM at a concentration of 100 parts per million (ppm). Recurring reports of the softening of the finger- tips and bone of VCM/PVC workers prompted the Manufacturing Chemists Association ("MCA") to sponsor a study, but the results were inconclusive as to the exact cause of the malady. Further toxicological studies continued to show that VCM exposure caused cancer in rats. 62 The first death of an American worker exposed to VCM was reported in 1971; on March 30, 1972, MCA financed an epidemiological study; by 1974, thirteen workers in the PVC and fabricating industries had died from angiosarcoma. 63 In response to these developments, the Secretary promulgated an emergency standard of fifty ppm time-weighted average to re- place the 500 ppm standard then prevailing in the industry. Four days after a hearing was held on this standard, it was discovered that angiosarcoma of the liver had been produced in mice at a level of fifty ppm of VCM. OSHA abandoned the fifty ppm stan- dard in favor of a one ppm standard, the so-called "no detecta- ble" level. The agency's permanent standard was one ppm averaged over an eight-hour period, but allowing for peaks of up to five ppm during periods of not longer than fifteen minutes. 64 That thirteen human lives had already been lost as a result of inadequate regulation of VCM greatly affected Justice Clark's de- cision. 65 The Society of the Plastics Industries and members of 59. Id. at 1304. 60. Id. 61. Id. at 1305-06. "We need not outline in detail the morbid 'Vinyl Chloride Chronol- ogy,' published by an industry spokesman. . . in order to illustrate the mounting evidence ofVCM's carcinogenity. Indeed, the record shows what can only be described as a course of procrastination on the part of the industry to protect the lives of its employees." 62. Id. 63. Id. at 1306. 64. Id. at 1307. 65. See, e.g., id. at 1308 ("[I1t must be remembered that we are dealing here with human lives, and the record reveals that eleven manufacturing plant workers and two fabrication plant workers have already died from the effects of this potent chemical."). 1985] 78 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 the vinyl chloride industry argued that the one ppm standard was not justified by the available scientific evidence because none of the expert witnesses who testified at the agency's rulemaking hearing could say with certainty that exposure to VCM at levels lower than fifty ppm was unsafe. 66 Justice Clark, however, pointed to evidence including a toxicology study conducted by the industry's trade association, and the recommendation of "ex- pert after expert" that exposure to the carcinogen be restricted to the "lowest detectable level." Given the Secretary's role under OSHA to protect employees, and that the Secretary's decision was at the "frontiers of scientific knowledge," Justice Clark found the evidence "quite sufficient" to support the Secretary's decision. 67 OSHA's effort in 1978 to reduce exposure to benzene met with less judicial sanction than its efforts in the Society of Plastics case. In what has come to be known as the "Benzene Case," 68 industry representatives challenged the Secretary's proposed permanent standard for this known carcinogen. The standard would have lowered the permissible level for airborne concentrations of ben- zene from ten ppm to one ppm. The Secretary based the lower standard on a combination of epidemiological data and recent toxicological studies showing that exposure to benzene caused leukemia, chronic nonmalignant blood disorders, chromosomal aberrations and other long-term health effects in humans. 6 9 Be- cause no level of exposure of benzene had been shown by these studies to be safe, the agency reduced permissible exposure to 1 ppm on the basis of its general policy of reducing exposure to all carcinogens to the "lowest feasible level." 70 In the court's view, the primary inquiry was whether the agency acted within the bounds of its statutory authority. 7' According to the court, for a standard to be "reasonably necessary" to provide a safe and healthful place of employment, as required by the Act, the Secretary had to provide substantial evidence showing that the the benefits of lowering the standard bore a reasonable rela- 66. Id. 67. Id. 68. American Petroleum Inst. v. OSHA, 581 F.2d 493 (5th Cir. 1978), aff'd, 448 U.S. 607 (1980). 69. 581 F.2d at 498 & n.10. 70. Id. at 501. 71. Id. at 497. Risk Assessment tionship to the costs imposed upon industry of complying with the standard. 72 In this case, the estimated dollar costs of compli- ance were quite high,73 while the benefit to workers was indeter- minable. The Secretary argued that it was sufficient to assume, on the basis of scientific hypothesis, that those benefits "may be appreciable," 74 but that it was impossible to estimate such bene- fits because there was a lack of knowledge on the effects of low- level exposure to benzene. 75 The Fifth Circuit disagreed with the Secretary's claim that estimates could not be based on the ex- isting level of scientific knowledge. 76 The court also felt that the agency's estimation of benefits could not be supported by sub- stantial evidence simply because it was based on a rational hy- pothesis. 77 Rather, the Secretary had to provide "some factual basis for an estimate of expected benefits" to show that the stan- dard was reasonably necessary. 78 The court distinguished the In- dustrial Union and Society of Plastics decisions as not having considered the "reasonably necessary" statutory requirement. 79 By a plurality vote,80 the Supreme Court affirmed the Fifth Cir- cuit's decision to invalidate the 1 ppm standard, but did not ad- dress whether the Act required that OSHA standards be supported by a reasonable correlation between costs and bene- fits. 8 ' Instead, the Court construed the "necessary and appropri- ate" language of the Act 82 to contain a threshold requirement for revising health and safety standards: the Secretary had to show, on the basis of substantial evidence, that the existing standard 72. Id. at 502-05. 73. Id. at 503. According to OSHA's estimates, first-year operating costs for all affected industries would be $187 to 205 million; engineering control costs would be $266 million, and recurring annual costs would be $34 million. 74. Id. 75. Id. at 504. 76. Id. 77. Id. at 503. 78. Id. at 504. 79. Id. at 505. 80. Justice Stevens wrote the opinion for the Court; the ChiefJustice and Justice Stew- art joined in the opinion and Justice Powell joined it in part. Justice Rehnquist concurred in the judgment; Justice Marshall, joined by Justices Brennan, White and Blackmun, dissented. 81. 448 U.S. at 639-40. That question was later resolved in American Textile Mfrs. Inst. v. Donovan, 452 U.S. 490 (1981), where the Court held that the Act does not require the Secretary of Labor to determine that the costs imposed by a health and safety standard bear a reasonable relationship to its benefits for workers. 82. 29 U.S.C. § 652(8) (1982). 1985] 80 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 "more likely than not" posed a "significant risk of material health impairment" such that lowering the exposure limit would signifi- cantly increase worker health.83 As in the Fifth Circuit case cited below, the Secretary's policy of reducing exposures to the "lowest feasible level" failed to pass the Court's substantial evidence standard of review. The Court argued that the policy placed the burden on industry of proving a "safe" level of exposure, while the Act imposed a burden on the Secretary to establish a need for stricter standards. 84 The Secre- tary had found that exposures to between twenty-five and forty ppm of benzene indisputably caused nonmalignant blood disor- ders; but it had not provided "direct support" that such disorders occured at exposures below ten ppm. 8 5 The Secretary had not estimated the risk of contracting nonmalignant disease from ex- posures below ten ppm, because, the Secretary argued, the data linking low-level exposures and blood abnormalities was inade- quate to construct a dose-response curve at such levels.8 6 The evidence that low-level exposure to benzene caused leukemia was "even sketchier," according to the Court. Only one epidemiolog- ical study supported such a conclusion, and its authors had ex- pressly stated that the study was not conclusive because workers' deaths might have been caused by other carcinogens.8 7 In the Secretary's view, the policy was justified lest the agency be forced to wait for deaths to occur before it could regulate to prevent them.88 The Court denied that its new threshold require- ment put the agency in such an untenable position. The agency could still determine what constituted a "significant" risk, a test which was not intended to impose a "mathematical straitjacket" on the agency to calculate the exact probability of harm.8 9 However, the court at one point urged the agency to use math- ematical methods of calculating risk, and to extrapolate from toxi- cological and epidemiological evidence to show that it would be "more likely than not" that a significant risk exists under the pre- vailing standard.90 The Court also seemed to be promoting a so- 83. 448 U.S. at 653. 84. Id. at 659. 85. Id. at 631. 86. Id. at 631-32. 87. Id. at 633. 88. Id. at 652. 89. Id. at 655. 90. Id. Risk Assessment called "reasonableness" standard for determining significance: a one in one thousand chance of death from exposure might appear significant to a "reasonable person," a one in one billion chance would not.9 ' Despite the Court's assurances to the contrary, the very choice of a reasonableness test for estimating "significance" reveals a reflexive, unthinking reliance on mathematical assess- ments. 92 For example, the Court's set of reasonably significant and insignificant numerical risks ignores the frequency of expo- sure. Yet it is essential that a probable frequency of the event be incorporated in any calculation of risk. Risk assessment requires two distinct estimations: toxicity and exposure. For example, it is conceivable that of 240 million persons experiencing a daily risk of one in a million from drinking water, 87,600 would die each year. But if 1,000 persons experienced a cumulative lifetime risk of only one in a thousand from exposure to a rare airborne contami- nant, only one person would die in seventy-three years. The Court's numerical examples only indicate toxicity, but suitable risk assessments must also reflect exposure, that is, they must be restricted to doses over time in specific populations. In his dissent, Justice Marshall decried the reliance on quantifi- cation of risks which the plurality opinion seemed to foster. Jus- tice Marshall explained that, for carcinogens, the assumptions required for quantification are "necessarily arbitrary." Regula- tory action based on quantification, therefore, would "deceive the public . . .[that] realistic assessments of the relevant risks" were being made. 93 The plurality assured the Secretary that "substantial evidence" did not mean "scientific certainty" but simply "best available evi- dence." 94 The Court cited the District of Columbia and Second 91. Id. at 656, 657 n.64. 92. Merrill describes the Benzene case as marking the "hard look" era ofjudicial review, supra note 48, at S423-24. Merrill criticizes the Court's review in this case as "betray[ing] little recognition of the limits of scientific certainty." Id. at S424. 93. Id. at 716. See also Ruckelshaus, Risk in a Free Society, 14 ENVTL. L. REP. (ENvTL. L. INST.) 10,190, 10,193 (1984) ("[W]e should understand the limits of quantification; there are some cherished values that will resist being squeezed into a benefits column, but are no less real because of it. Walter Lippman once pointed out that in a democracy "the people" as in "We the people," refers not only to the working majority that actually makes current decisions, and not only to the whole living population, but to those who came before us, who provided our traditions and our physical patrimony as a nation, and to those who will come after us, and inherit. Many of the major decisions we make on envi- ronmental affairs touch on this broader sense of public responsibility.") 94. Id. at 656. 1985] 82 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 10:67 Circuits, and stated that "the Agency is free to use conservative assumptions interpreting the data with respect to carcinogens, risking error on the side of over-protection rather than under- protection," in construing the evidence, "so long as those as- sumptions are supported by a body of reputable scientific thought."9 5 In this case, however, the plurality felt that the Secre- tary had not used all the scientific techniques currently available for assessing risks. 9 6 The Court attributed the Secretary's refusal to make a dose-response estimate not to a lack of data to make a meaningful estimate, but--"at least in part"-to the Secretary's view that "nothing less than absolute safety would suffice." 9 7 It is difficult to reconcile the Court's assurances that the Secre- tary retained substantial discretion with the Court's actual deci- sion in the Benzene case. The plurality denied having made its own fact findings, or having rejected those made by the Secre- tary.9 8 But the Court's assumption that certain kinds of scientific evidence must be employed is, nevertheless, an intrusion on the Secretary's role of determining risks at the frontiers of scientific knowledge. Because the Benzene decision provides no uniform criteria for evaluating exposure standards promulgated pursuant to the Act, it is Uncertain how meaningful the Court's assurances will be. A more recent decision in the Fifth Circuit, Gulf South Insulation v. Consumer Product. Safety Commission,9" adds to the confusion over judicial review of risk assessment. A markedly activist Fifth Cir- cuit rejected a risk assessment of the adverse health effects of urea-formaldehyde foam insulation (UFFI), which was used by the Consumer Product Safety Commission (CPSC) as a basis for reg- ulation under the Consumer Product Safety Act. 0 0 In Gulf South, manufacturers sought to overturn the Consumer Product Safety Commission's ban on the use of UFFI in schools and residences. The Commission supported the ban with a finding, as required under the Act, that UFFI presented an unreasonable risk of injury to human health, and that no feasible product standard could be 95. Id. 96. Id. "[T]he record ... and OSHA's own rulings on other carcinogens indicate that there are a number of ways in which the Agency can make a rational judgment about the relative significance of the risks associated with exposure to a particular carcinogen." 97. Id. at 656 n.63. 98. Id. at 659. 99. 701 F.2d 1137 (5th Cir. 1983). 100. Consumer Product Safety Act, 15 U.S.C. § 2057 (1982). Risk Assessment devised that would adequately protect the public. 0 1 Petitioners challenged that finding under Section 2060(a) of the Act, which provides for review of the Commission's action in the courts of appeals according to a substantial evidence standard. 0 2 The court's opinion began by reviewing the agency's investiga- tions into the health effects of UFFI, specifically, those effects caused by the substance's propensity to emit formaldehyde gas. The agency had first undertaken a three-year investigation of 350 UFFI homes in which residents had complained of acute irritant symptoms such as nausea, headaches, respiratory distress and skin irritation. The agency then commissioned the National Acad- emy of Sciences to determine whether there was a threshold level of formaldehyde exposure below which no acute symptoms will be experienced. The Academy found, on the basis of available scientific literature, that no such threshold existed. From the Academy's finding, the Commission concluded that formaldehyde gas released from UFFI posed an unreasonable risk of acute irri- tant effects.10 3 The Commission had also received data from UFFI manufac- turers linking nasal cancer in rats to high levels of formaldehyde exposure. The Commission assembled a sixteen-member panel of government scientists to study the data. The panel concluded that the data were valid and that "formaldehyde should be pre- sumed to pose a carcinogenic risk to humans."' 1 4 The Commis- sion then extrapolated from the high-dose animal studies to quantify the human cancer risk posed by low levels of exposure to UFFI. Using a computerized mathematical risk assessment model called Global 79, the agency predicted that it was 95% possible that the increased risk of cancer to a person living in a UFFI home for his or her lifetime would range from zero to fifty-one in 1,000,000. From this prediction, the Commission concluded that UFFI posed an unreasonable risk of cancer to humans. The Comission's subsequent ban was based on its findings with re- spect to both acute irritation and carcinogenicity. 0 5 The UFFI manufacturers alleged that the Commission's carci- nogenicity findings were unconvincing, primarily because of inad- 101. Id. 102. Id. § 2060(a). 103. 701 F.2d 1141. 104. Id. 105. Id. at 1142. 19851 84 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 equacies in the data base that had produced the Global 79 prediction. One of the assumptions underlying that prediction was that the level of formaldehyde present in a UFFI home aver- aged .08 ppm over nine years. The Commission had derived this average from tests taken in 1,164 homes, and from laboratory tests conducted under simulated conditions. Petitioners argued that the 1,164 sample homes were not selected randomly, that the tests were not conducted consistently or, in some cases, accu- rately, and that the laboratory tests were conducted under condi- tions that did not resemble those in the average home. 0 6 The Fifth Circuit found that the studies relied on by the Com- mission did "suggest" that UFFI appreciably raises in-home for- maldehyde levels. However, the court felt that the Commission had erred by incorporating its data into "an exacting, precise and extremely complicated risk assessment model."' 1 7 The court considered the model useless without reliable data on what con- stituted an average UFFI home.'0 8 The court also faulted the model for relying on empirical data on formaldehyde carcinoge- nicity derived from a single study involving only 240 rats. The court noted that if twenty fewer or twenty more rats had been used, the risks predicted by the model might have been drastically altered. 09 The court also criticized the Commission's conclusion regard- ing the acute irritant effects produced by UFFI, because the Com- mission's study of 350 homes did not indicate the degree of likelihood that such symptoms would occur. Without determin- ing the degree of risk involved, in the court's view, the Commis- sion could not make a finding of "unreasonable risk of injury." The Commission had relied on the degree of risk posited in the National Academy of Sciences study, that "somewhat less than 20% of healthy adults may respond to the irritant effects of for- maldehyde at 25 ppm." But as the court pointed out, 25 ppm was considerably greater than what the Commission had concluded to be the formaldehyde level in the average UFFI home; moreover, the NAS study had not indicated whether the predicted responses were severe or slight. In view of these shortcomings, the court concluded that substantial evidence did not support the Commis- 106. Id. at 1143-44. 107. Id. at 1145. 108. Id. 109. Id. at 1146. 1985] Risk Assessment sion's finding of "unreasonable risk" of injury from acute irritant effects.' I o The reviewing court's determination to undertake a scientific re-analysis of the reliability of the CPSC's risk assessment in Gulf South is cause for concern. For sound reasons the decision leaves legal analysts troubled."' Of the many issues in the case, the court concentrated on a scientific re-analysis of the agency's risk assessment. The court provided a footnote criticizing the agency's assumption that human beings are at least as vulnerable as rodents, and its use of conservative assumptions for develop- ing standards. 1 2 The court also chided the agency for failing to consider the absence of evidence on human harm. 1 3 Merrill cor- rectly describes this decision as the first to embody even greater uncertainty than that associated with the risk assessments them- selves. 114 As McGarity cautions, such decisions will find the courts inheriting the kind of unpleasant attention previously re- served to the agencies. 1"5 The decision stands simply as a remark- able judicial probe of an agency's record on a narrow question. It is unlikely to set an important precedent.' 16 The above analysis indicates the erratic application of the sub- stantial evidence review to agency risk assessments." 7 The courts in Industrial Union and Society of Plastics deferred to the agency's policy choices based on the limitations of those assessments. The 110. Id. at 1148. 111. See, e.g., Merrill, supra note 56, at S424-25; McGarity, supra note 12, at 103. 112. 701 F.2d 1145-46. 113. Id. 114. Merrill, supra note 56, at S425. "The opinion's close scrutiny of an exercise that is fraught with uncertainty, but yet promises improvement in regulation of health hazards, is disconcerting." Id. Merrill cautions that the Gulf South decision may discourage agencies from using risk assessments without data which is usually not available, and may lead agen- cies to conclude that risk assessments are simply not a sufficient legal basis for regulating. Id. 115. McGarity, supra note 12 at 103. 116. But see Merrill, supra note 56, at S424-25 (noting that Gulf South marks a new era of increased judicial scrutiny over agency decisions based on risk assessments). 117. Merrill breaks the courts' decisions down into four eras ofjudicial review. Cases in the first era-"forced deliberation"-marked courts' willingness to review agency inaction. Merrill, supra note 56, at S421-22. In the second era of"judicial reticence," as exemplified by Industrial Union, the courts gave agencies broad discretion to find facts and make policy decisions related to health risks. Id. at S422. The third era- "hard look"- includes the Benzene case and reflects courts' "aggressive scrutiny" of agencies' fact-findings. Id. at S423-24. The fourth era, which Merrill labels "substituted judgment," consists solely of ;the Gulf South decision and represents an even closer judicial scrutiny of agency decisions based, in part, on risk assessments. Id. at S424-25. 86 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 courts in the Benzene case and Gulf South, on the other hand, closely scrutinized the agencies' risk assessment methods. Such scrutiny, itself, led to possibly inconsistent results. The Supreme Court in the Benzene case urged the agency to quantify its risk assessments, despite inadequacies in the data base for such quan- tification; the Fifth Circuit in Gulf South chastised the agency for over-relying on quantification of risks when the underlying data was weak. Given that risk assessments are conducted at the fore- front of scientific knowledge, courts cannot routinely scrutinize agencies' methodologies in order to fine-tune such methods, and provide a clear course for agencies to follow. 2. Controlling Air Pollution: The Arbitrary and Capricious Test In Ethyl Corporation v. EPA,'"8 manufacturers of lead additives and refiners of gasoline sought review under the Clean Air Act (CAA) 1 19 of the EPA's regulation requiring a reduction of lead in all gasoline to an average of 0.5 grams per gallon.' 20 Section 2 11 (c) (1) (A) of the CAA authorizes the Administrator of the EPA to promulgate regulations prohibiting the use of fuel additives if emission products of fuel or fuel additives "endanger the public health or welfare."' 12 1 Because the Clean Air Act did not pre- scribe a standard of review for emissions regulations, the District of Columbia Circuit applied the APA's prescription that "infor- mal" rulemaking procedures must not be "arbitrary, capricious, or an abuse of discretion."'' 2 2 A division of the Court found the regulation to be arbitrary and capricious but, in a rehearing en banc, the full Court vacated its decision. Judge Skelly Wright, who had filed a vigorous dissent when the case was first decided, wrote the majority opinion for a sharply divided court. The focus of the en banc court's inquiry was the level of proof required under the CAA for a finding of "endangerment" to the public health or welfare. The majority held that the standard did not require proof of actual harm, but only proof of a "significant risk of harm."' 2 3 Indeed, the agency was not even required to 118. Ethyl Corp. v. EPA, 541 F.2d I (D.C. Cir.) (en banc), cert. denied, 426 U.S. 941 (1976). 119. 42 U.S.C. § 7401 and scattered sections of Title 42 (1982). 120. 541 F.2d at 9-10. 121. 42 U.S.C. § 7401 (1982). 122. 5 U.S.C. § 706(2)(A) (1982). 123. 541 F.2d at 13. Risk Assessment prove that harm is "probable," if a probability could not be deter- mined from the available evidence.' 24 According to the majority, because the standard is essentially precautionary and preventive, it requires the Administrator to assess risks without relying solely on facts and, where necessary, to make "an essentially legislative policy judgment, rather than a factual determination, concerning the relative risks of underprotection as compared to overprotec- tion." 25 Its own task, the court emphasized, was not to decide whether the agency's finding was based on substantial evidence in the record, or even on a preponderance of the evidence, but rather to determine whether that finding had a "rational basis" in the evidence.' 26 To make such a determination, the court waded through the massive record generated by the rule-making process. The rec- ord revealed that the EPA had relied on three types of evidence in establishing the new standard for lead in gasoline: theoretical work on lead dust-fall, epidemiological studies of exposed popu- lations, and clinical studies of exposed individuals. A number of laboratory studies had established that young animals were more susceptible than adults to the effects of lead. Molecular biochem- istry analysis showed that animals deficient in zinc, calcium and iron-a condition very common among children in poor fami- lies-absorb lead much more readily than other animals. 2 7 After immersion in the record which contained over 10,000 pages, and which all parties agreed was incomplete, the court ad- mitted that the evidence was so inconclusive that it could as easily have upheld a decision not to regulate lead as a decision to re- quire phased-in elimination of lead in gasoline. 128 Under the applicable standard of review, however, EPA's action was upheld. The agency's reliance on the best available estimate of the risks of lead to the health of children was within the defer- ence afforded by the "will endanger" standard, and its handling of the risk assessment process was entirely rational.12 9 Central to the court's decision was its own risk assessment formula: the mag- nitude of the risk sufficient to justify regulation is inversely pro- 124. Id. at 18. 125. See id. at 13-18. 126. See id. at 33-37. 127. See id. at 37-48. 128. Id. at 37. 129. See id. at 47-48. 1985] 88 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 portional to the harm to be avoided.' 30 In the court's view, the severity of harm threatened was of great importance in judging the rationality of the agency's conclusions. 13 In this case, the po- tential harm was severe, particularly in relation to the most vul- nerable sector of the population-inner-city children. This potential made it less significant that the agency could not conclu- sively prove the degree to which emissions from leaded gasoline, as opposed to exposure from other lead sources, caused the re- ported damage to human health.' 32 Judge Wright's formula translated into legal doctrine an emi- nently sound value judgment with which most scientists in the public health field readily agree: It is better to prevent disease, rather than try to cure it. The court's review in Ethyl Corp. was both exhaustive and deferential. The court considered every study in the record and the objections made to them. Yet the court was equally impressed by the agency's thorough assessment and handling of a "complicated problem with great ease and can- dor," and respectful of the "flexibility" afforded the agency's as- sessments by the "will endanger" standard in the CAA.'3 Perhaps the best example of the exhaustive evidentiary review under the arbitrary and capricious standard which courts have made in the past decade is Lead Industries Ass 'n. v. EPA 13 4 In Lead Industries, the District of Columbia Circuit upheld EPA's national ambient air quality standards for lead, which were promulgated under Section 109(a)(2) of the Clean Air Act.' 3 5 As in Ethyl Corp., the primary evidentiary issue arose from the problems inherent in determining the health effects of low-level, cumulative exposure to lead. As required by statute, EPA had prepared a "criteria docu- ment" to support its ambient air quality standards for lead. 13 6 130. See id. at 19. 131. See id. 132. See id. at 13. 133. See id. at 47. 134. 647 F.2d 1130 (D.C. Cir. 1980), cert. denied, 449 U.S. 1042 (1982). 135. 42 U.S.C. § 7409 (1982). EPA plans to issue final regulations to eliminate lead from gasoline by 1990. The Federal Centers for Disease Control of the Department of Health and Human Services is recommending that the action level for blood lead be set at 25 micrograms per decileter. 136. 42 U.S.C. § 7408(a)(2) (1982) provides that criteria documents 'shall accurately reflect the latest scientific knowledge useful in indicating the kind and extents of all identi- fiable effects on public health and welfare which may be expected from the presence of such pollutants in the ambient air in varying quantities." Risk Assessment The document considered a range of health effects related to lead, and concluded that children with more than forty micro- grams of lead per liter of blood risked developing anemia. A range of subclinical effects on the blood forming system were also noted, including elevation of erythrocyte protoprophyrin (EP), which impairs functioning at the subcellular level of the blood. 3 7 Preschool children and pregnant women were found to be partic- ularly susceptible to adverse health effects of lead exposure. 138 In addition to considering the qualitative health effects of expo- sure to lead, the criteria document laid the groundwork for quan- tifying a relationship between air lead levels and blood lead levels. After a detailed examination of relevant studies, the crite- ria document concluded that air lead/blood lead ratios encoun- tered in the general population fell within a range of 1:1 to 1:2 (micrograms of lead per cubic meter of air to micrograms of lead per deciliter of blood). No safe level for lead in the blood was identified. 1 3 9 The existence of a hypersusceptible population of children and pregnant women was central to the development of the final am- bient air standards for lead. The proposed standards were specif- ically designed to prevent the elevation of EP and the resulting effects on cellular functions in children. Thus a target mean pop- ulation blood lead level was selected as the lowest reported threshold of lead levels for EP elevation in children. On the basis of the information in the criteria document, the EPA selected a ratio of 1:2 to calculate the effect of air lead exposure on blood lead levels and arrived at the final standard of 1.5 micrograms of lead per cubic meter of air. 140 Petitioners contended that nothing in the record supported the Administrator's choice of the level of thirty micrograms lead per deciliter of blood, particularly since no adverse health effects had been shown to occur at that level. Under the APA's "arbitrary and capricious" standard of review, however, the court found the record "adequate" to support the EPA's decision. 14 1 The EPA had relied primarily upon the findings contained in its criteria document which, the court noted, was the product of a process 137. 647 F.2d at 1138-40. 138. Id. 139. Id. at 1140-41. 140. Id. at 1143-44. 141. Id. at 1156-60. 1985] 90 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 that allowed for rigorous scientific and public review. 142 It was significant that various experts who testified at the rulemaking proceedings supported the Administrator's choice of a target blood lead level. The court also found that the agency's reliance on the practices of the Centers for Disease Control was well- placed. 14 3 According to the court, because "there is evidence" in the record to support the agency's judgment, and because the agency had properly explained both the factual and policy bases for its decision, the selection of a target blood lead level must be upheld. 144 The court in Lead Industries was also forced to grapple with methodological challenges to the agency's use of three studies in establishing an air lead/blood lead ratio of 1:2. The studies in- volved both children and adults; since the standards were designed to protect children, petitioners contended, only studies focussing on children should have been used. The court dis- missed this argument on the basis of the criteria document which cited studies reporting ratios for children at 1:1.2 to 1:2.3.145 Pe- titioners also argued that the agency's use of these studies was "inconsistent and designed solely" to support EPA's predeter- mined choice of an air quality standard for lead. The court re- jected this characterization and accepted EPA's explanation that differences in its approaches to these studies were warranted by apparent errors in the studies themselves. The court noted that the criteria document had treated these studies as reliable, that the agency's choice of a ratio was endorsed by several experts, and that the entire issue of air lead/blood lead ratios had been the subject of extensive discussion in the rulemaking process.' 4 6 Ethyl Corp. and Lead Industries reveal the mixed nature of review which courts undertake pursuant to the arbitrary and capricious standard. The District of Columbia Circuit in Lead Industries cited Ethyl Corp. to justify its "substantial inquiry" into the facts (which in a case such as this will consist primarily of a body of scientific 142. The Administrator had found that the initial adverse health effects of lead expo- sure occurred at the thirty microgram level in children, and that the thirty microgram level provided a margin of safety to protect children against more serious consequences of lead exposure. Id. at 1144. 143. Id. at 1157-58. 144. Id. at 1158. 145. Id. at 1162-63. 146. Id. Risk Assessment literature).1 47 In Ethyl Corp. the court's perusal of the record was equally as rigorous. Yet in both cases the court of appeals placed less weight on the existence of conflicting evidence which its vig- orous inquiry revealed than on the existence of some evidence to provide a rational basis for the agency's decision. The purpose of the court's "substantial inquiry" was to "educate" the court, not so that it could second-guess the agency's conclusions and meth- ods, but merely to adequately assess whether such conclusions and methods were rational and within the discretion afforded by Congress. 1 48 The flexibility with which the arbitrary and capricious standard may be applied is also illustrated by the District of Columbia Cir- cuit's opinion in American Petroleum Inst. v. Costle.149 At issue were EPA's revised national ambient air quality standards for ozone. In revising its prior standard of 0.08 ppm to 0.12 ppm, the EPA had relied on a mix of scientific evidence, including clinical studies of healthy subjects exposed to ozone, toxicological studies, and an environmental model of the formation of ozone. In addition, a risk assessment study summarized medical opinions regarding the dose-response relationship between ozone and chronic diseases such as emphysema. 150 Among the ten petitioners, the Natural Resources Defense Council contended that the revised regulation was too lenient. Challenging the regulation as too stringent, the American Petro- leum Institute ("API") presented the now familiar argument that because no adverse health effects had been proven below 0.25 ppm, the 0.12 standard was irrational. 15 The API also cited EPA's admission that its own risk assessment was not completely reliable. ' 52 Under the arbitrary and capricious standard of review,' 53 the court considered its "proper function . . .not to weigh the evi- dence anew and make technical judgments; . . . but to determin[e] if the Administrator made a rational judgment."' 154 147. Id. at 1145-46. 148. Id. 149. 665 F.2d 1176 (D.C. Cir. 1981), cert. denied, 455 U.S. 1034 (1982). 150. Revisions to the National Air Quality Standards for Photochemical Oxidants, 44 Fed. Reg. 8202-17 (1979). 151. 665 F.2d at 1184-85. 152. Id. at 1185. 153. 42 U.S.C. § 7607(d)(9)(A) (1982). See 665 F.2d at 1184. 154. 665 F.2d at 1185. 1985] 92 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 To make such a finding, the court did not consider itself obligated to determine that each study relied on by the EPA was accurate. Numerous studies showed disruption of normal body functions at "low" (0.15 to 0.39 ppm) ozone levels, and the court saw no rea- son to second-guess the Administrator's reliance on the studies, "even given the acknowledged uncertainties in some of the conclusions." 55 The District of Columbia's Circuit's approach in American Petro- leum seems hardly consistent with the court's painstaking inquiries in Ethyl Corp. and Lead Industries under the same standard of re- view. Analysis of judicial review under the arbitrary and capri- cious standard also reveals that such review is not fundamentally different from review of agency risk assessments at the forefront of scientific knowledge when the substantial evidence standard applies. Under both standards, review is limited to the rationality or reasonableness of the agencies' conclusions. How deeply the reviewing court will delve into the record to make such a determi- nation seems more a function of that court's predilections than of the applicable standard of review. IV. THE COURT AS AGENCY: THE ENDANGERMENT STANDARD Several environmental statutes provide that the United States and private individuals may bring actions in the federal district courts to enjoin unlawful polluting activities.i 5 6 A court called upon to grant relief under such provisions often finds itself in the same position as an administrative agency with basic science-forc- ing authority - both are confronted with inconclusive risk assess- ments on which some decisive action must be taken. Reserve Mining Co. v. EPA 157 is a classic illustration of the courts' role as an independent assessor of uncertain scientific evidence. The EPA had sought an injunction ordering Reserve Mining a taconite processing company, to cease discharging taconite tail- ings from its facility into the waters of Lake Superior and the air surrounding the community of Silver Bay, Minnesota. Taconite tailings, which result when taconite (low-grade iron) ore is processed into iron-rich pellets, were being released into the wa- ters of Lake Superior at the rate of 67,000 tons daily when the suit 155. Id. 156. See, e.g., Federal Water Pollution Control Act, 33 U.S.C. §§ 1319(b), § 1365(a)(1) (1982); Clean Air Act, 42 U.S.C. § 7604(a)(1) (1982). 157. 514 F.2d 492 (8th Cir. 1975) (en banc). Risk Assessment was commenced in 1972.158 The EPA, joined by three states and several environmental groups, charged that Reserve was violating a number of federal and state environmental laws and regula- tions. 159 The trial court concluded that Reserve's air and water discharges were unlawful and posed a "substantial danger" to public health. The court granted an injunction, and ordered the Silver Bay facility to be closed immediately. 60 Reserve appealed the order and the Eighth Circuit granted a stay of the injunction pending the appeal. 16' After further pro- ceedings in the district court, the case was decided by the court of appeals sitting en banc. Judge Bright, in an elaborate opinion, concluded that Reserve's discharges did so endanger the public health as to require abatement. However, the endangerment had not been shown to be so substantial as to justify the immediate shut-down of the facility. 162 Factual disputes dominated every aspect of this complex case,' 63 including the central issue of whether the discharges posed a "substantial endangerment" to the public. Studies dem- onstrating the health damage caused by inhalation of asbestos fi- bers provided one of the sources of contention. Most of the studies were conducted among workers in asbestos mills and mines at exposures much higher than that presumed to face Silver Bay residents. 164 A few of the studies showed that low level expo- sure to asbestos fibers might result in mesothelioma, but because the level of exposure was not quantified in those studies, and be- cause the concentration of particles in the air around Reserve's processing facility could not be precisely measured, the trial court had found it impossible to draw firm conclusions from these studies. 165 158. Id. at 500. 159. Id. at 501. 160. United States v. Reserve Mining Co., 380 F. Supp. 11 (D. Minn. 1974). 161. Reserve Mining Co. v. EPA, 498 F.2d 1073 (8th Cir. 1974). 162. 514 F.2d at 500. 163. A threshold question was whether the particular ore mined by Reserve contained amosite asbestos, a material with a demonstrated capacity to endanger human health. At trial, the evidence did not show conclusively that Reserve's tailings contained asbestos, but only that a "portion of" the cummingtonite-grunerite contained in the ore could not be meaningfully distinguished from amosite asbestos. 380 F. Supp. at 33 (summarized by the court of appeals at 514 F.2d at 510). 164. 514 F.2d at 510. 165. Id. at 510-12. 1985] 94 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 Assessing the danger posed by the discharge of tailings into wa- ters which were the source of the Silver Bay public drinking water system raised even more difficult questions of proof. Reliable evi- dence as to whether the ingestion, as opposed to inhalation, of asbestiform particles could cause damage to human health was limited to studies based on experimentation with animals, and studies of gastrointestinal effects in workers exposed to high levels of asbestos. These studies could not be applied to the Silver Bay community with any degree of precision to draw conclusions other than that the ingestion of asbestos fibers posed some unde- termined health risk. 16 6 The trial court had heard testimony from over 100 witnesses in the course of a 139-day trial which generated 18,000 pages of transcript. 16 7 The court of appeals, in turn, scrutinized this numbing record. It was obvious that the medical and scientific questions in dispute lay at the frontiers of scientific knowledge: the very nature of the pollutant could not be conclusively estab- lished; there was no proof that actual harm to human health would result if Silver Bay residents continued to be exposed to taconite tailings in the air and water; indeed, the court of appeals could not say that the probability of harm from continued dis- charge was more likely than not.' 68 Although the Reserve case involved a request for injunctive re- lief, the court of appeals viewed its function as similar to one of reviewing an agency regulation based on the same evidence. The court cited Industrial Union in indicating that its evidence was "clearly . . . on the frontiers of scientific knowledge."' 169 Next the court invoked Ethyl Corp. where the court used the same "en- dangerment" threshold as a basis for taking preventive measures. Under the "endangerment" test, the court reasoned that potential harm must be considered.170 FollowingJudge Wright's formula in his dissent in the first Ethyl Corp. decision, 17' the Eighth Circuit saw the severity of potential harm as a crucial element in its deter- mination of endangerment. 72 The EPA's purported hazards were 166. Id. at 520. The opinion summarizes the available information on ingestion of as- bestos fibers at 514-19. 167. 380 F. Supp. at 15. 168. 514 F.2d at 520. 169. Id. at 519. 170. Id. at 519-20. 171. See supra text accompanying notes 112-126. 172. 514 F.2d at 519-20. Risk Assessment not based on proven scientific facts but on a medical theory (es- sentially, that the pollutant was a form of asbestos which endan- gered the public health even at low levels of exposure), but the consequences that could result if the theory proved true would be severe. The court concluded that the existence of a public health risk justified an injunctive decree requiring abatement of the haz- ard "on reasonable terms as a precautionary and preventive mea- sure to protect the public health."' 173 Although the district court's order for an immediate shut-down of the taconite facility was re- versed as an abuse of discretion, 174 the court of appeals gave de- tailed directions for the formulation of an order on remand requiring Reserve Mining to take immediate action to abate its discharges. 17 5 Similar problems of proof and an even more toxic substance were at issue in United States v. Vertac Chemical Corp., decided by the District Court for the Eastern District of Arkansas in 1980.176 The United States sought injunctive relief under the Resource Conservation and Recovery Act, 177 Federal Water Pollution Con- trol Act, 178 and Refuse Ac0 79 requiring Vertac to cease discharg- ing toxic chemicals, including dioxin,iso into the air, water, and soil surrounding its Jacksonville, Arkansas, chemical manufactur- ing plant. Vertac had undertaken considerable effort in coopera- tion with federal and state authorities to control releases of toxic wastes from its 92-acre site. Nevertheless, at the time the suit was brought, dioxin persisted at the parts per billion level in the soil and sediment surrounding the plant, in the cooling pond, and in the Jacksonville sewage treatment plant.1si The sediments of Rocky Branch Creek at the plant contained 1090 parts per billion of dioxin. The question for the court was whether the presence of such quantities of dioxin constituted an "imminent and sub- 173. Id. at 520. 174. Id. at 537. 175. Id. at 538-40. 176. 489 F. Supp. 870 (E.D. Ark. 1980). 177. 33 U.S.C. § 1364 (1982). 178. 42 U.S.C. § 6973 (1982). 179. 33 U.S.C. § 407 (1982). The court rejected this claim because Vertac had ceased producing dioxin. 489 F. Supp. at 876-77. 180. The district court referred to dioxin as "the most acutely toxic substance yet syn- thesized by man." 489 F. Supp. at 876. 181. Id. at 876-77. 1985] 96 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 stantial endangerment" to human health for purposes of the ap- plicable federal statutory injunction provisions. 18 2 The twin elements of endangerment-the severity of harm and the probability of its occurence-were as difficult to quantify here as they had been in Ethyl Corp. and Reserve Mining. The dioxin con- centrations present at the Vertac site were "far below the thresh- old for acute or single-dose toxic effects" of dioxin, but the United States contended that the long-term effects of chronic ex- posure to low levels of dioxin could be severe. 183 Toxicity studies demonstrated that harmful effects could be produced in animals exposed to dioxin at low levels. No safe detectable level of dioxin in the environment was known to exist. 184 The district court granted a preliminary injunction and re- quired Vertac to undertake specified abatement measures. 8 5 As justification for its decision, the court cited evidence from 1971 EPA hearings on the dioxin-containing chemicals 2-4-5-T and TCDD, after which the EPA canceled its registration of the chemi- cals under the Federal Insecticide, Fungicide and Rodenticide Act.' 8 6 Such evidence included epidemiological data from Ore- gon and Vietnam which showed an increased incidence of miscar- riage and birth defects in populations exposed to dioxin in defoliant sprays.' 8 7 The evidence also included laboratory stud- ies showing that dioxin caused birth defects in mice and rats.' 8 8 The district court acknowledged that "while there may be low probability of harm from dioxin as defendants contend, there is a serious and dire risk from exposure to dioxin should the hypothe- sis advanced by the plaintiff prove to be valid."' 8 9 Citing Reserve Mining, the court acknowledged that "endangerment" need not be proven with certainty, but merely as a probability. Moreover, because the evidence gave rise to a "reasonable medical concern for the public health," the court found such evidence of risk to present an "imminent and substantial endangerment."' 9 0 182. Id. at 884-85. 183. Id. at 876. 184. Id. 185. Id. at 888-89. 186. Id. at 881 n.7. 187. Id. at 880-81, 884. 188. Id. 189. Id. at 885. 190. Id. (quoting 33 U.S.C. § 1364 (1976) and 42 U.S.C. § 6973 (1972)). Risk Assessment From the above discussion, it appears that judicial review of risk assessments, presented in actions for injunctive relief under the endangerment standard, is quite similar to the review con- ducted in the two other contexts discussed earlier. The reviewing court is faced with inherently inconclusive evidence, and a statu- tory policy dictating what consequences should follow from an apparent-albeit highly uncertain-risk of harm. The policy in Reserve Mining and Vertac of, essentially, erring on the side of pre- caution or prevention, was derived from the District of Columbia Circuit's formula in Ethyl Corp. for assessing the rationality of an agency's decision. The policy also appears more deferential to the proponent of the requested protective action than that which the Supreme Court drew from OSHA, in the Benzene case. CONCLUSIONS According to a major school of the philosophy of science, scien- tific evidence cannot "prove" anything, but can only establish the probability that something is likely to be true.19 ' A proposition can be falsified, that is, disproved, but can never be confirmed. According to one author, the level of proof required to convince a reviewing court that a regulatory action was supported by sub- stantial evidence, or was not arbitrary and capricious, would demonstrate to a scientist only that the proposition had a forty and thirty percent chance, respectively, of being true.'9 2 Recog- nizing the impossibility of proving certain scientific propositions even under the less rigorous legal standards, the Supreme Court and most federal courts have demanded that agencies provide neither rigorous step-by-step proof of cause and effect, nor scien- tific consensus, to support agencies' efforts to protect the public health. Confronted with issues on the frontiers of science, and regardless of the applicable standard of review, courts have granted considerable weight to experimental or animal evidence, and to efforts by experts to estimate risks from such evidence. This is especially true where evidence of human harm, albeit in- complete, supplements experimental evidence.' 93 191. K. POPPER, THE LOGIC OF SCIENTIFIC DISCOVERY (1959). 192. Hills, Legal Decisions and Opinions in Pollution Cases, 10 ENVrL. SCI. AND TECH. 234-35 (1976). 193. Agency and judicial approval of risk assessment techniques remains a source of public debate. In one debate, ex-EPA Administrator William Ruckelshaus argues that risk assessment is necessary to direct agencies' attention toward "significant problems" to help 1985] 98 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW [Vol. 10:67 Analysis of judicial review of risk assessments under the sub- stantial evidence, arbitrary and capricious, and endangerment standards reveals that the applicable standard does not strictly de- termine the scrutiny which such review will entail. 194 Professor Rodgers views the uneven application of standards of judicial review on environmental matters as grounds for sus- pecting that "whether the court will dig deeply or bow cursorily depends exclusively on whether the judge agrees with the result of the administrative decision."' 9 5 This cynicism does not seem completely justified, the Fifth Circuit notwithstanding. Uneven application of standards of review by the courts may simply be a consequence of the inherently imprecise nature of the risk assess- ment process, and of the varied circumstances in which that pro- cess is presented to the courts. As a basic example, judicial acceptance of the agencies' reliance on tentative findings, extrap- olations, and experimental models, appears much more likely when such evidence is supplemented by the actual occurence of human harm from exposure to the chemical at issue. 196 A risk assessment is hardly the linchpin in the disposition of an issue of such widespread public concern as the effects on children agencies prioritize public resources for the protection of public health. Ruckelshaus, supra note 93, at 10,190-93. Ruckelshaus cautions, however, against over-reliance on risk as- sessments because of their uncertainties, and urges that the public be educated as to the uncertainties and the policy assumptions underlying the risk estimates. Id. at 10,190-191. David Doniger, a senior attorney with the Natural Resources Defense Council, argues that reliance on risk assessments even to the extent urged by Ruckelshaus is unwarranted. According to Doniger, risk assessments are simply "too uncertain and fragile" to be a basis for agency regulations in all but a few limited circumstances. Doniger, The Gospel of Risk Management: Should We be Converted?, 14 ENVTL. L. REP. (ENVrL. L. INST.) 10,222, 10,223 (1984). 194. Whether the strong pronouncements by the Supreme Court in Chevron, 104 S. Ct. 2778, on deference to agencies' choice among conflicting statutory policies will lead to a more consistent review of agency's policy choices remains to be seen. 195. Rodgers, Judicial Review of Risk Assessment: The Role of Decision Theory in Unscrambling the Benzene Decision, II ENVTL. L. 301, 302 (1981). The EPA recently proposed guidelines for the uniform application of risk assessment techniques to agency decision-making. See N.Y. Times, Nov. 20, 1984, at A21, col. 1. Adoption of uniform guidelines by administrative agencies may, in turn, lead to a more uniform judicial review of agency decisions based on risk assessments. 196. The lesson of the Society of Plastics Industry, 509 F.2d 1301, where the deaths of thirteen workers exposed to vinyl chloride preceded the court's consideration of OSHA's strict regulatory standard for the substance, remains striking in this regard. The EPA recently proposed guidelines for the uniform application of risk assessment techniques to agency decision-making. See 49 Fed. Reg. 46,294-46,331 (1984). Adoption of uniform guidelines by administrative agencies may, in turn, lead to a more uniform judicial review of agency decisions based on risk assessments. Risk Assessment of flaking asbestos insulation in schools. However, while it is con- venient to differentiate between the political management and the scientific assessment of risk, the latter cannot be removed from politics. The very decision to assess the risk associated with a par- ticular product or activity can derive from public pressure. Some years ago, Max Weber aptly characterized this dilemma: "Strictly speaking, objectivity cannot be applied to the selection of problems, but only to their solutions. What a society deems worth resolution becomes the measure of that society, but this worth cannot be scientifically demonstrated."'' 9 7 Moreover, risk assessment techniques, themselves, are artful constructions, based on science but entailing numerous assump- tions of fact which are based on politically derived policy choices. For example, the choice of a margin of safety-which is a compo- nent of risk assessments-is also a keenly political question. Sci- ence alone cannot rationalize a regulatory standard which is a 10th, 100th, or 1000th of the level at which no effect has been observed. As ex-EPA Administrator Ruckelshaus cautions, "We should remember that risk assessment data can be like a captured spy: if you torture it long enough, it will tell you anything you want to know."' 198 The cases examined above all attest to this problem. No readily discernible logic yet allows regulatory agencies to anticipate fu- ture risks. The agencies continue to mount a reactive response to externally generated pressures. This often leaves them short on science and long on speculation. Yet to delay occupational or en- vironmental regulation in the name of better science because no human harm has yet been detected makes experimental subjects of those exposed in the meantime. To the fore comes risk assess- ment, garbed as the neutral arbiter. With the exception of Gulf South, the cases discussed illustrate judicial tolerance for risk assessment as an aid to agencies strug- gling with an imperfect data base. In the face of scientific uncer- tainty regarding the effects of highly toxic substances, courts are willing to concede that prevention of future harm demands infer- ences and leaps of faith. Unwavering scientific consensus is not a prerequisite to regulatory action. What remains largely unde- cided among Congress, agencies and the courts is just how far 197. M. WEBER, THE METHODOLOGY OF THE SOCIAL SCIENCES (1949). 198. Ruckelshaus, supra note 93, at 10,190. See id. at 10,190-01 for a brief discussion of policy based assumptions. 1985] 100 COLUMBIA JOURNAL OF ENVIRONMENTAL LAw [Vol. 10:67 such inferences should be tolerated as a basis for regulatory ac- tion. At a minimum, courts can be expected to continue to re- quire not certain .evidence, but the best available evidence. Agencies must marshall the best science, rationally and systemat- ically. Limitations on judicial review are imposed both by courts' lack of expertise and by the agencies' assigned role of protecting the public health in the face of scientific uncertainty. 1985] Risk Assessment 101 -2 E 0 E E . 0. 0 0 0 .00 -E LN. 5L.0 ~.0 ~0 > > czcz M~ 00 0 C)C)- ;,= 0 z 0 -z M0 1 0 -~~ 0 .CzjO. cn V) 0. 102 COLUMBIA JOURNAL OF ENVIRONMENTAL L-Aw [Vol. 10:67 - 0: , :zz ZZ 2 k' 0 ,0 qj E 0'0 0 ' C) 0 X Z 0 0. Co' CL, P-4 1985] 0 0 =u u4 0 o- a0. :150 u Risk Assessment 0 C 0 j =~ Mu 4 0 *0 ., vJM-c, , , *00 c 0 1-~M. uuu . 41.E r 0.~" 5- 11 104 COLUMBIA JOURNAL OF~ ENVIRONMENTAL LAW [Vol. 10:67 m E E~ v jC w r C Q.0 q 00 E >- z 00 to "m Z 0J mi Ua 0 - = : C, - - U 0~ 0 -m 2- m 0 b sj E . .0 bOc bD <0 Cb 0 0 Risk Assessment r. : -L =-0 0 :6 E 0 E (4 r Iq 0 u 00 0 Cc c CL C > > r I s 0 >J 0. GJ> CI V ~ i ~ : cc V . = - OL. 0.0 -C j ~ CJ CIL> C 00 CIC M C w I CJ *~G~ w uI 0 G c EL o o > 0I 0i 19851 106 COLUMBIA JOURNAL OF ENVIRONMENTAL L-AW [Vol. 10:67 E E 00 0 0 ouo v a P-4- bp~ 0: 4, 0 00 gm - 1 - a U.4 z 0 0" a 0 00 E- E. 0 < 9L. 0CL -1 Go < < < U. U. 1985] Risk Assessment 107 0 o EE .0 0a 0 .0. re0.. -0 0 .0 r 2: 06 0 0 0P-4 o~ ac o..2 >< M 03 U ~ c~ 0 E* -. > w -0 u . 0 0 ~- .. 0 .0 . 0. 0 0 . 3: M blo b 0 M ow 2M.: U w t 0 0- S Uz 4 C6 0.o U 1 >,Z ~ C 2 ~. 2 ~ ~ .U ~ . Ci 0 0< -0 ~ U0 COLUMBIA JOURNAL OF ENVIRONMENTAL LAW 0,, r e a 0 bj", M .- E. E E..S ,- . - S bk cd U -" 0 0 0 5 [Vol. 10:67 Risk Assessment M 0 c o o 0 0 C3 0 t 0- 0 r- c0 M - b -C13 C .- v b 0 - 0 A~ A 7; E .. 0 0iM A.c u CL z' 1985]