John S. Applegate

Professor of Law and John S. Hastings Faculty Fellow

Indiana University School of Law - Bloomington

211 S. Indiana Ave.

Bloomington, IN 47405

Phone: 812-855-9198

Fax: 812-855-0555


Professor Cross describes the phenomenon of hormesis and its application to the relevant legal standards for environmental protection1 with characteristic lucidity and force. The hormesis hypothesis asserts that most (if not all) chemicals that are toxic in low doses are in fact beneficial at even lower levels, below the no observed adverse effect level (NOAEL), because these very low doses stimulate positive defense mechanisms in the affected cell, organ, or organism. The U- or J-shaped dose-response curve that results from hormesis, Cross argues, should be integrated into current regulatory practice. If it were, he believes, many regulatory standards for toxic substances would allow greater human exposure than they presently do, to take advantage of the beneficial effects of very low exposure levels. This, of course, contradicts the basic assumption underlying regulation of toxic substances in the United States that toxic substances either have a threshold below which they have no effect, or that they are actually or potentially deleterious at all levels of exposure.

The bulk of Cross' paper applies the hormetic dose-response curve to the various legal standards found in environmental legislation. However, to reach those issues, Cross passes over some fundamental questions about hormesis that must be addressed before it is useful to talk about its legal or regulatory implications. Appealing as it is, the clear picture that Cross paints is not a fair rendering of either the science or the law of hormesis. In fact, hormesis is a complex and uncertain biological phenomenon. Its stimulatory effect cannot be reduced simply to beneficial and harmful effects ("good for you" or "bad for you," so to speak). While the existence of a stimulatory response seems (at least to a non-scientist like me) to be likely, the range, variability, and reliability of the effect for toxic chemicals outside the laboratory remain unknown. At the same time, the law is not simply a transmission belt for applying novel scientific theories to regulation. It must assure the validity of the scientific claims, apply them in service of important public values, and construct a workable regulatory system under conditions of uncertainty. This comment will address these evaluative aspects of the law as they apply to the complexities and unanswered questions about hormesis.


The law, first, must assure that scientific assertions are valid and applicable to the problem at issue according to "the methods and procedures of science."2 The so-called gatekeeping function is far more apparent in private litigation where private parties present all of the information and there is no external check on validity or applicability. In Daubert v. Merrell Dow Pharmaceuticals, Inc., the Supreme Court ruled that courts "must ensure that any and all scientific testimony or evidence admitted is not only relevant, but reliable," and it offered a non-exclusive list of criteria for evaluating the evidence.

Hormesis would not be immune to the Daubert analysis. On the one hand, the hypothesis is testable, and it has been published in scientific journals. On the other hand, Calabrese & Baldwin's 1997 report,3 which appears to be the principal basis for the claim that hormesis is a universal phenomenon, raises a number of methodological issues. It is a study of many studies which were not designed to identify hormesis but apparently found it anyway. The comparability of these studies, especially when drawing quantitative conclusions, is an obvious issue. Moreover, Calabrese & Baldwin examined only studies that appeared to provide evidence for the hormetic hypothesis, making it impossible to claim that other explanations had been excluded.4 In addition, the low doses involved pose real difficulties in resolving even large hormetic effects,5 so the margin of error is presumably relatively high. Finally, hormesis does not appear to have received general acceptance in the toxicologic community, a fact which supporters of the hypothesis have gone to some lengths to explain.6,7

In public law cases, including those applying the legal standards that Cross addresses, the courts' gatekeeping role is naturally more limited. The Daubert analysis does not (and should not) apply.8 With a few egregious exceptions in which reviewing courts have taken upon themselves the task of deciding what is and is not "good science,"9 courts regularly and properly10 rely on the regulatory agency's expertise to perform the gatekeeping role. With hormesis, the relevant federal agencies seem to be performing this function. Government scientists are clearly aware of the hypothesis, but they have been circumspect in their response to it.11 Regulators are clearly entitled indeed, expected to recognize the existence of a novel scientific theory and yet decline to adopt it because its validity or applicability has not yet been demonstrated.8

Do these gatekeeping concerns mean that the hormesis hypothesis is "wrong" or irrelevant to toxics regulation? No and it is certainly not for a non-scientist like me to make such declarations. But it does appear that, as the matter now stands, hormesis may be either a fundamentally new paradigm for toxicology or an interesting but legally insignificant biological phenomenon. Before plunging into revision of regulatory standards, the law obliges agencies and courts to determine which it is.

Expression of Values

The law is not only or mostly a vehicle for the dissemination and application of scientific theory. The purpose of toxics legislation, for instance, is not to express and implement a particular dose-response model, but rather to protect the public from involuntary exposure to hazardous substances, using dose-response information where appropriate to formulate precise requirements. The law's principal function is to express and implement important social values, such as who should bear the risk of harm in society and what levels of risk are acceptable. These are not scientific questions.12

U.S. environmental law takes as its basic goal the prevention of harm from toxic chemicals, as opposed to either permitting such harm to occur or relying on ex post compensation.13 Given this basic perspective, it is important to remember what kinds of chemical substances we are talking about. We are not talking about salt or sugar or aspirin. These chemicals may exhibit hormetic effects, but they are of little interest to environmental regulation. Rather, hormesis is important to environmental regulation to the extent that it claims application to the kinds of chemicals that the environmental laws regulate: chemical substances which are believed to pose a risk of death at very low levels of exposure, like dioxins, PCBs, asbestos, and heavy metals. Regulatory practice not unreasonably starts from the position that uncontrolled exposure to such materials is dangerous and that involuntary human exposure to them should be sharply limited or eliminated where appropriate. Federal legislation expressly encourages this approach.14,15,16

This legal context suggests several reasons for not rushing to replace current understandings of toxic risks with hormetic theory. First, at a fundamental level, it is not really accurate to characterize the effects of hormesis as a simple dichotomy between "healthiness" and "declining health." Hormesis claims to cover many different endpoints at many different levels of operation, from cells to organs to entire organisms, so it clearly operates through numerous mechanisms simultaneously. Indeed, the same dose can have both good and bad effects (Cross suggests the example of selenium). More importantly, hormesis is a stress reaction to an external agent. While "That which does not kill me makes me stronger" has a certain intuitive appeal and a certain degree of truth, some growth responses to stress are not desirable (e.g., hypertrophy of the heart muscle) or may overlap with increased cancer risk (e.g., cellular hyperplasia).17 As a result, good-for-you and bad-for-you are necessarily net effects, and the netting process is a matter of choosing among desirable and undesirable endpoints.11 The move from a stimulatory response to health and declining health is a value judgment, not an objective benefit that can readily be plugged into dose-response calculations.

Second, we have extremely limited direct data on the hormetic effect. Calabrese & Baldwin's 1997 article, on which advocates of hormesis place great weight, is not only a study of studies, but the underlying studies were not designed to identify, to say nothing of measure, hormesis. Such an indirect approach may assure lack of bias in the original studies, as Calabrese & Baldwin emphasize, but it is hardly the ideal scientific method for examining the details of a phenomenon. In fact, very few studies have actually set out to measure hormesis, and the very low doses involved will make it unusually difficult to do so.5 The protective purposes of environmental law would certainly justify reluctance to rely on hormesis without knowing where the hormetic "tipping point" is.

Third, the size of the hormetic effect is not actually the "reciprocal" of the deleterious effect. Hormesis operates in the region below the NOAEL of chemicals that already pose toxic risks at low levels of exposure (micrograms and parts per million or billion). This is a very small range, and any amount above that poses a demonstrable (the lowest observed adverse effect level by definition designates an demonstrable effect) toxic risk. Consequently, in order to take advantage of a hormetic effect in setting regulatory standards, one must be able to calibrate exposure to those levels allowing enough exposure to reach these very low levels, but not more. While such calibration may be possible in the laboratory setting, it is utterly unrealistic to think that exposure can be so carefully controlled in the ambient environment.11 The chemicals of greatest environmental concern, the so-called persistent bioaccumulative toxic (PBT) chemicals and persistent organic pollutants (POPs), remain in the environment for a long time and many have been produced and used in large quantities. As a result, many are ubiquitous at low doses, and humans (at least in industrialized countries) have existing body burdens of them. Given the very low doses at which hormesis works, it is entirely possible, if not likely, that the hormetic increment has already been taken up by body burden, background exposures, or cumulative sources of the same chemical.

Finally, we know nothing of the long-term effects of sub-NOAEL doses. Basic hormetic theory says that the stimulatory response to the external stressor is overwhelmed as levels of exposure increase (thus the U- or J-curve). The same effect may occur over time. Long-term stress may exhaust the initial stimulatory response or repair mechanism, and the low doses will begin to have deleterious effects.4 Similarly, the existence of other stressors in the environment (and there are many) of the same or a similar type could also exhaust the cell, organ, or organism, resulting in deleterious effects at levels that appeared to be hormetic in the laboratory.20

Again, my point is not that hormesis is wrong or entirely inapplicable to environmental regulation, but rather that no one knows yet. Cross argues that the linear dose-response default assumption is also unproven, but this criticism misses the point. EPA has not adopted the linear default because it is necessarily or universally "correct" (it is not), but rather because it is biologically plausible and it guards against underestimation of health risks. Thus, the primary cost of error in using the linear assumption is economic unnecessarily stringent regulation rather than risk to human health. With hormesis, by contrast, the cost of error is risk to human health. If the phenomenon does not exist, or if its range is smaller than was thought, or if other sources of exposure take up the hormetic increment, permissive regulation would accept higher risks of a deadly disease. And there is little countervailing benefit to permitting hormetic doses. While there is a certain vagueness on this point, none of reports I have seen claims an actual therapeutic effect from low doses of the toxic chemicals about which environmental law is concerned. While it may be true that selenium, for example, is an essential mineral, it is a large step from that observation to a claim that exposure to selenium pollution is an essential part of balanced diet. (I suspect that the day will never come when children are exhorted to take their daily PCB supplement.) This balancing of the costs of error and the dispreference for health risks is by no means incontrovertible, but it reflects the important social values embodied in protective environmental legislation.

Managing Uncertainty

The protective purpose of the law cannot be fulfilled if barriers to regulatory action are set so high that no protective action can be taken. The law must craft rules and legal structures that are capable of implementing the values it expresses. In the regulation of toxic substances, not only are the basic mechanisms of disease not fully understood, but there is an astonishing absence of long-term, low-dose toxicology data of any kind about the chemicals to which humans are regularly exposed.18,19 To implement the protective goal of the environmental laws, courts have consistently permitted regulation on the basis of less-than-complete scientific information.21,22 They have also accepted regulators' reliance on conservative default assumptions in the absence of better evidence to the contrary.23,24

While more information and greater precision would be ideal, real-world regulatory systems tolerate some degree of inaccuracy as the cost of taking prompt, protective action.25 For example, in considering the 1990 Clean Air Act Amendments for hazardous air pollutants, Congress was faced with two decades of nearly total EPA inaction under the original "margin of safety" provision, but it was also committed to achieving a risk-based level of health protection. So, in the interim between the enactment of the amendments and the completion of the complex process of developing risk-based regulations, Congress required EPA to adopt technology-based regulations.15,18,26 They are an imprecise surrogate for risk, but they are better than permitting uncontrolled exposure.

The continued use of a linear dose-response model until the nature and extent of the hormetic effect is better defined should be seen in the same light. The "extraordinary study design demands" for determining hormetic effects5 strongly suggests that not only do we have nothing like the data that would define the range of the hormetic effect in individual chemicals, but also that we are unlikely to obtain significant quantities of such data any time soon. Even if not precisely accurate, low-dose linearity at least implements the main protective purpose of the environmental laws. (As noted above, the downside health risk of the linear model is minimal or non-existent.) The alternative, requiring a full accounting of hormetic effects in proposals for protective regulation, would leave humans exposed to uncontrolled levels of highly toxic chemicals while such data are gathered. In the meantime, human beings, the objects of the law's concern, would be its test subjects. Thus, the protective legal response to the hormesis hypothesis may well be to await clearer demonstrations of the nature and extent of the effect. Only then will regulators have occasion to apply hormesis to the legal standards that they administer.


By relying on a stylized account of the science of hormesis and the law's response to scientific innovation, Professor Cross makes his analysis accessible and compelling, but he also passes too quickly over some fundamental issues. Our legal system demands that we take a hard look at critical scientific questions about hormesis and its relationship to the public values that the law embodies before we incorporate it into regulatory standards for highly toxic chemicals. That said, here, as so often in his writings, Cross brings new ideas and new perspectives to old problems, challenging our assumptions and predispositions. A sensible precautionary approach to environmental regulation imposes stringent controls in situations where scientific uncertainty predominates, but it supersedes the provisional response if newer and better information warrants it. EPA takes this position with its default assumptions in risk assessment, including low-dose linearity.27 Cross performs a real service by calling to our attention the emerging theory of hormesis and its potential legal implications, and in doing so he helps to begin and to frame this important inquiry.


1. Cross F. The Legal Implications of Hormesis, [this issue]

2. Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993)

3. Calabrese EJ, Baldwin LA. The dose determines the stimulation (and poison): development of a chemical hormesis database. International Journal of Toxicology 1997; 16: 545-559

4. Elliott, KC. A case for caution: an evaluation of Calabrese and Baldwin's studies of chemical hormesis. Risk: Health, Safety & Environment 2000; 11: 177-196

5. Calabrese EJ, Baldwin, LA. Can the concept of hormesis be generalized to carcinogens? Regulatory Toxicology and Pharmacology 1998; 28: 230-241

6. Calabrese EJ, Baldwin LA. The marginalization of hormesis. Toxicologic Pathology 1999; 27: 187-194

7. Calabrese EJ, Baldwin LA. Reevaluation of the fundamental dose-response relationship. Bioscience 1999; 49: 725-732

8. Sierra Club v. Marita, 46 F.3d 606 (7th Cir. 1995)

9. Gulf South Insulation v. Consumer Product Safety Commission, 701 F.2d 1137 (5th Cir. 1983)

10. Cross FB. Pragmatic pathologies of judicial review of administrative rulemaking. North Carolina Law Review 2000; 78: 1013-1064

11. Davis JM, Farland WH. Biological effects of low-level exposures: a perspective from U.S. EPA scientists. Environmental Health Perspectives 1998; 106 Supp. 1: 379-381

12. Wagner WE. The science charade in toxic risk regulation. Columbia Law Review 1995; 95:1613-1723

13. Applegate, JS. The Precautionary Preference: An American Perspective on the Precautionary Principle, Human & Ecological Risk Assessment 2000; 6: 413-443

14. Clean Water Act, 33 U.S.C. § 1314

15. Clean Air Act, 42 U.S.C. § 7412

16. Toxic Substances Control Act, 15 U.S.C. § 2605

17. De Rosa CT et al. Public health implications of environmental exposures. Environmental Health Perspectives 1998; 106 (Supp. 1): 369-378

18. Applegate JS, Laitos JG, Campbell-Mohn, C. The Regulation of Toxic Substances and Hazardous Wastes. Foundation Press: New York 2000

19. Campbell-Mohn C, Applegate JS. Learning from NEPA: Some Guidelines for Responsible Federal Risk Legislation. Harvard Environmental Law Review 1999; 23:93-139

20. Foran J. Regulatory implications of hormesis. Human & Experimental Toxicology 1998; 17: 441-443

21. Ethyl Corp. v. Environmental Protection Agency, 541 F.2d 1 (D.C. Cir. 1976) (en banc), cert. denied, 426 U.S. 941 (1976)

22. Industrial Union Department, AFL-CIO v. Hodgson, 499 F.2d 467 (D.C. Cir. 1974)

23. Natural Resources Defense Council v. EPA, 824 F.2d 1211 (D.C. Cir. 1987)

24. Chlorine Chemistry Council v. EPA, 206 F.3d 1206 (D.C. Cir. 2000)

25. Latin H. Ideal versus real regulatory efficiency: implementation of uniform standards and "fine-tuning" regulatory reforms. Stanford Law Review 1985; 37:1267-1332

26. Wagner WE. The triumph of technology-based standards. University of Illinois Law Review 2000; 2000:83-113

27. U.S. Environmental Protection Agency. Proposed guidelines for carcinogen risk assessment. Federal Register 1996; 61:17960-18011