Document 1Q8GGbOZ0Rw6xge22QpbVwJ8E

The Regulation of Carcinogenic Hazards Gio Batta Gori In earlier times economic develop ment, an ethical imperative in the West ern world, took precedence over con cern for individual or collective safety. Rational audacity is a distinct human characteristic, and many have argued that too much regard for safety goes against human nature and happiness. But today new dangers of chronic illness and disability give new dimensions to fear, and there is a clear need to better define the odds involved. Carcinogens have been recognized latory issue (J), but in the United States regulation was born of the social dialec tic of the turn of the century and retains some of the romantic intransigence of that movement. American laws do not give explicit directions for considering risk-benefit factors, and economic argu ments usually are contested with the im plicit assertion that health has a supreme value whose economic costs need not, should not, be measured. Undeniably, for an individual life has transcendent value. But excessive regu Summary. In the United States, statutes for controlling carcinogens are largely motivated by the ideal of absolute safety at all costs. There is no requirement for riskbenefit assessment. Bias is inherent in the prescribed bioassays; test findings may have restricted meaning within the specific experiments but cannot be translated into quantitative assessment of human risk. Such data are improperly used in resolving regulatory questions case by case. It is suggested that a system of relative standards of utility be formulated which, paired to standards of tolerable risk or safety, would define a range of use restrictions. Substances intended for a certain use would then be regulated according to those standards. since Percival Pott some 200 years ago (/), and regulatory initiatives have flour ished for the last three decades, as anx ieties gradually shifted from vanishing acute infections to chronic diseases made more prevalent by longevity gains (2), and in the wake of concern over ubiquitous industrialization. The need for regulation of carcinogens in developed societies is undisputed, but the premises and practices of such regu lation are not. In the United States we are experiencing a transitional period that reflects general trends of social evo lution: regulatory agencies, which in ac cord with centuries-old traditions have been allowed to wield quasi-autonomous . normative powers, find themselves in creasingly at odds with expanding de mands for due process in the resolution of uncertain perceptions and conflicting values. In many countries the balance of toler able risks and benefits is the central regu- lation hampers technological develop ment and thereby denies its fruits to the poor in our own society and elsewhere in the world. Public policy is fundamentally an economic exercise. It cannot evade the balancing of risks and benefits with out incurring gross inequities. Gradually, loftier views are giving way to a realism that expects regulation to improve the quality of life for the living, not merely to extend life expectancy. When it is accepted that absolute safety is not a reasonable goal, it becomes the business of regulation to define tolerable levels of risk; to this end, explicit proce dures for benefit assessment need to be introduced into regulatory statutes. Al so, revisions of current practices appear to be in order for a consistent approach to the determination of risk from poten tial carcinogens. In general, normative regulation has relied on the definition of standards, usu ally of empirical origin but then upgraded as use and newly acquired information suggests: the standards define reference compounds, testing procedures, process flows, as well as tolerable or permissible doses and levels. In the late 1950"s, in the heat of public concern over food additives and new pharmaceuticals. U.S. legislators sought the advice of science in the definition of carcinogen standards. Unfortunately, at that time understanding in this field was too problematic to produce even a sug gestion of standards. Legislators were left w ith the alternatives of either intran sigent policies, such as the Delaney amendment adopted at that time, or vague statements of intent, which have intrigued dialecticians ever since. It now appears that these legislative precedents have been largely responsible for the po larization and ambiguity which plague the regulation of carcinogens in this country, and for a climate of opinion that has discouraged intermediate solutions in favor of all-or-none pronouncements. The public would be surprised to note that different potential hazards, docu mented with comparable scientific meth ods and data, are regulated by widely dif ferent criteria. The explanation is simple. There is an implicit necessity to tolerate certain conditions where intran sigent regulation would mean a drastic alteration of traditional life-styles. There are many examples: exposure to sunlighi: ingestion of fats, proteins, and ex cessive calories, or of foods containing natural and apparently unavoidable po tential carcinogens such as aflatoxins: and the paradox of potential risk from the ingestion of our own saliva, at times very rich in nitrites, precursors of carci nogenic nitrosamines. The need to tolerate such hazards has not been seriously challenged even by the most ardent proponents of regula tion. But the absence of explicit statu tory rules for risk and benefit assessment has led to a situation where each sub stance is considered separately, and widely disparate outcomes are influ enced more by adversary emotions than by real values. Testing procedures have proliferated and become more com plicated. logistic resources have been virtually exhausted, and testing as now envisioned may be precluded for the ma jority of environmentally significant sub stances. Moreover, intrinsic uncertainties in current procedures make it impossible to prove safety beyond doubt: on this basis prognostications of doom have flour- The author is Deputy Director. Division of Cancer Cause and Prevention. National Cancer Institute, Bethesda. Maryland 20205. 256 0Q36*8075/8(V0418*0236$01.50/0 Copyright Q 1980 AAAS SCIENCE. VOL. 208. 18 APRIL 1980 AP00024154 atom may take place, (iii) multiphoton free electron-free electron transitions (bremsstrahlung, inverse bremsstrahlung) can occur, and (iv) the close dy namics of the scattering process can be affected by n-photon transitions in the CN + l)-electron negative atom system. The theoretical treatment of these com plex processes has been limited to spe cial situations where one particular inter action dominates and the others can be neglected or treated as perturbations. Experimental investigations in this area have been initialed only very re cently and can be divided into two cate gories: (i) electron scattering by laser-ex cited species and (ii) free-free radiative transitions. In the first category, tunable (low- power) dye lasers are utilized to excite the target atom and scattering by the ex cited species is studied. Here the inter action of the electromagnetic field with the target dominates and its interaction with the free electron can be neglected. Register ei al. (53) excited the (6p) 'P state of Ba with a wide-band laser (iso tope and hyperfinc levels unresolved) in a dense Ba beam (where multiple absorp tion and reemission cause all degenerate magnetic sublevels to be evenly popu lated). Electron scattering was then ob served by the 'P Ba as well as cascadepopulated 3P, 'D, and 3D Ba. Energyloss spectra with the laser off and on are shown in Fig. 13. With the laser on, addi tional features are observed correspond ing to superelastic scattering and to tran sitions from excited state to excited state. From this and similar spectra cross sections (averaged over isotope, hyperfine, and magnetic sublevels) were deter mined. When a single-frequency dye laser is used, the energy resolution (10'* eV) is good enough to selectively pump an indi vidual hyperfine level of a specific iso tope, and by selecting the proper polar ization of the laser light the population of the magnetic sublevels can be con trolled. In other words, a polarized tar get (aligned or oriented) can be prepared for electron scattering. These types of experiments make it possible to study electron scattering with an energy reso lution of 10'* eV and nuclear spin ef fects, and they yield scattering ampli tudes for individual magnetic sublevel excitations. The excitation with polar ized light and subsequent measurement of the superelastic electron scattering is equivalent to the time inverse electronphoton coincidence measurement {34). Experiments in the second category have been concerned with free-free radi ative transitions in strong electromagnet ic fields. Here the photon energy is cho- Entrgy in units < Isstr photons Fig. 14. Elastic scattering by Ar (a) without and (b) with a high-intensity CO, laser field. sen to be far from resonance (even for multiphoton processes) and the major in teraction occurs between the photon field and the free electron. The present situation in this area has been summa rized by Gavrila and Van der Wiel (55). Because of emission or absorption of photons by the scattering "free" elec tron (induced and inverse bremsstrah lung), the elastic (and inelastic) features appearing in energy-loss spectra have side lobes on both sides separated by an energy corresponding to the photon en ergy. Elastic scattering in a high-power pulsed CO, laser field has been studied by Weingartshofer el al. (36); an energyloss spectrum for the laser off and laser on cases is shown in Fig. 14, a and b, respectively. The symbols 6, i/i, and g in Fig. 14 refer to the electron scattering angle, the laser beam angle, and the elec tron impact energy, respectively. Here SP = e-K/2 k0, where is the laser beam polarization vector, K is the momentum transfer vector, and *0 is the initial mo mentum of the electron. Interesting situations concerning reso nances can develop in connection with free-free radiative transitions. Reso nances occur not only when the electron kinetic energy is proper but also when the photon energy or the electron-photon combined energy is appropriate. Such situations have been studied by Langhans (37) and Langendam and Van der Wiel (?S). I hope the examples of electron-atom (molecule) collision processes discussed in this article have given the reader a prospective view of this special, but very active, area of atomic and molecular physics. The investigations in this area yield information that is pertinent to our understanding of the basic laws of phys ics. and at the same time they are of con siderable practical value for understand- f ing the behavior of systems where free electrons are present. Reference! and Note* 1. S. Trajmar. Acc. Chem. Res.* in press. 2. H. S. W. Massey and E. H. S. Burhop, Elec tronic and tonic impact Phenomena (Claren don, Oxford, 1969), vol. l; H. S. W. Massey, ibid.* vol. 2. 2. S. Trajmar and A. Chutjian, unpublished re sults. 4. S. Trajmar and R. 1. Hall, unpublished results. 5. S. F. Wong and G. J. Schulz, unpublished re sults; G. F. Schulz, in Principles oj Laser Plasmas, G. Bekefy, Ed. (Wiley, New York, 1976), chap. 2. 6. D. F. Register, S. Trajmar, S. K. Srivastava, Phys. Rev., in press. 7. S. K. Srivastava, A. Chutjian. S. Tnymar. J. Chem. Phys. 63, 2639 (1973). 8. G. J. Schulz, Phys. Rev. Leu. 10. 104 (1963). 9. For recent reviews see G. J. Schulz. Rev. Mod. Phys. 45. 378 (1973); Wd.. p. 423; D. E. Golden. Aav. At. Mot. Phys. 14, I (1978). 10. V. Fano, Phys. Rev. 124, 1866 (1961);_____ and J. W. Cooper, Phys. Rev. A 137, 1364 (1965). 11. P. J. Chantry, in Electronic and Atomic Colli sions, G. Watel, Ed. (North-Holland, Amster dam, 1978), pp. 23-24. 12. S. Trajmar and R. I. Hall, J. Phys. B 7, L458 (1974). 13. R. 1. Hall, in Electronic and Atomic Collisions. G. Watel. Ed. (North-Holland, Amsterdam, 1978), pp. 25-41. 14. R. A. Bonham and M. Fink, High-Energy Elec tron Scattering (Van Nostrand Reinhold, New York, 1974); R. A. Bonham, in Electron Spec troscopy: Theory, Techniques and Applications, C. R. Brundle and A. D. Baker. Eds. (Academic Press, London, 1979), pp. 127-187. 15. H. A. Bethe, Ann. Phys. 5, 325 (1930). 16. M. Inokuti, Rev. Mod. Phys. 43. 297 (1971). 17. E. N. Lassettre, A. Skerbele, M. A. Dillon. J. Chem. Phys. 56. 1829 (1969). 18. S. Trajmar. in Electronic and Atomic Collisions, G. Watel. Ed. (North-Holland, Amsterdam. 1978), pp. 113-128. 19. L. S. Vuskovic, S. K. Srivastava, S. Trajmar, D. Bozinis, Cienc. Cult. (Sdo Paulo) 31, 1018 (1979). 20. W. Williams. J. C. Cheeseborough, III, S. Traj mar,/. Phys. B 11, 2031 (1978). 21. S. Trajmar, W. Williams. S. K. Srivastava, ibid. 10.3323 0977). 22. W. Williams, S. Trajmar. D. G. Bozinis, ibid. 8. L96 (1975). 23. S. Trajmar and W. Williams, unpublished re sults. 24. D. C. Cartwright, et al., Phys. Rev. A 16, 1013 (1977);ibid., p. 1041; /bid., p. 1052. 25. See, for example, j. P. Doering and J. H. Moore, Jr.,/. Chem. Phys. 56,2176(1972); J. P. Doering. ibid. 71, 20 (1979). 26. See, for example, W. M. Flicker, O. A. Mosher, A. Kuppermann, ib/d. 70. 1986 (1979); ibid., p. 2003. 27. R. H. Huebner, D. L. Bushnell, Jr., R. J. Celotta, S. R. Miekrzarek, C. E. kuyatt, Nature (London) 257. 376 (1975). 28. H. Ehrhardt, K. H. Hasselbacher, K. Jung. K. Willman. Case Stud. At. Phys. 2, 161 (1972). 29. I. E. McCarthy and E. Weigold, Phys. Lett. C 27,276(1976). 30. R. E. Imhof and F. W. Read, J. Phys. B 4. 450 (1971). 31. K. Blum and H. Kieinpoppcn, Phys. Rep. 4, 203 (1979). 32. M. J. Van der Wiel and G. Wiebes, P/tysica (Utrecht) 54,411 (1971). 33. D. R. Register, S. Trajmar, S. W. Jensen, R. T. Poe, Phys. Rev. Lett. 41, 749 (1978). 34. I. V. Heriel and W. Stoll, Adv. At. Mol. Phvs. 13, 113 (1977). 35. M. Gavrila and M. Van der Wiel, Comments At. Mol. Phys. S. 1 (1978). 36. A. Weingartshofer. E. M. Clarke, J. K. Holmes, C. Jung, Phys. Rev. A 19, 2371 (1979). I am grateful to these authors for supplying a preprint of their paper and for permission to use their data in rig. 14. 37. L. Langhans./. Phys. B 11. 2361 (1978). 38. P. J. K. Langendam and M. J. Van der Wiel, ibid., p. 3603. 39. I wish to thank my collaborators whose work has been cited. This article represents the re sults of one phase of research conducted at the Jet Propulsion Laboratory under NASA con tract NAS 7-100. AP00024155 ished. Such prophecies fly in the face of a little-publicized circumstance, namely that in the United States and other ad vanced countries age-adjusted cancer rates in general have remained nearly stationary or have declined over the past several decades, except for some few cancers of recognized etiology (/, 4). Yet, because regulation is an essential safeguard of civilized living, we must re solve the conflict between the need to improve living standards and the need to preserve health and the natural environ ment. We must find more rational and defensible regulatory options. Of prime importance will be a reevalu ation of the scientific framework for the appraisal of carcinogenic risks. For this purpose society needs to depend on the objectivity of scientists, free of political pressures in experimental choices, de sign, and interpretation (5). Today such pressures are not absent; scientists have often been forced to produce clear-cut statements that, however convenient for the regulator, may not have scientific justification. Many of the debates on regulation of carcinogens have relied on the notion that animal tests can provide meaningful data for extrapolation of human risk, but evidence reviewed in this article sug gests that this notion should be modified. What Is a Carcinogen? It is commonly observed that higher organisms are naturally affected by tu mors. Because the information to deter mine their origins is lacking, it has been customary to define as baseline the natu ral rate of incidence of tumors in popu lations that have not been disturbed by known challenges. Thus, the current def inition of carcinogens refers to insults that increase the incidence of all tumors or of certain tumors or that shorten their customary time of appearance {6-8). This teleologic definition identifies the biologic consequences of carcinogens but not the mechanisms of action in volved, which at present arc still un known (9). Under such definition, over crowding, noise, and circadian and other stresses (10) could come to be defined as carcinogens even though they may be merely modulating factors in the assay system used. It has been argued that for regulatory purposes it does not matter that the defi nition is only teleologic, because it is the final outcome, increased cancer in cidence, that regulation seeks to control.. This would be true if simple and repro ducible interactions existed within the IS APRIL 1980 modulating factors and carcinogens that determine response in a given assay, but even in those instances where semiquantitative outcomes can be experi mentally reproduced, some artificial con trivance is necessary, such as the use of compounds that are strong carcinogens for the species or strain selected--but not necessarily for others--or the use of the maximum doses the test animals can tolerate. Both instances represent limit ing situations, where the carcinogenic in sult is artificially made to overpower oth er factors. It is well known that general toxicity and carcinogenicity do not go hand in hand and that they vary from species to species and with the method of adminis tration or intake; therefore, maximum tolerated doses--generally the highest intake that an animal can sustain for its lifetime without significant signs of acute or subchronic toxicity--will result in widely disparate testing levels, quite at odds with real-life conditions. It is apparent that the current defini tion of carcinogens confines the validity of data to a specific experiment, restricts the opportunities for generalization, and makes it difficult to distinguish between direct carcinogens and modifying fac tors. The Carcinogenesis Process Essentially nothing is known of the ul timate molecular events that determine the transformation of normal cells into cancer cells (II. p. 7: 12). Tumors can originate spontaneously, thus suggesting a natural instability of the cell. Whether this is truly an intrinsic phenomenon has yet to be finally settled: more often trans formation is observed after the appli cation of an insult external to the cell. Today it is generally believed that a carcinogen entering an animal may un dergo various metabolic manipulations before reaching a target cell: there it may find various conditions of susceptibility, resistance, competition, or repair, and may be subject to additional modifica tion before finding molecular receptors to determine, directly or not, transfor mation and the eventual appearance of a cancer cell (13), either as a single hit or after cumulation of progressive insults and damage. Not all cancer cells devel oped through this process have the op portunity of progressing to overt disease, because natural defenses may suppress the onset of asymptomatic cancer. It is estimated (//, p. 11) that the po tency of carcinogens in animal experi ments can vary by a factor of 10T. Hu man exposure to environmental insults can vary by a factor of 10" (14). If one were to add the attenuations occurring between exposure and actual intake of a compound, it is conceivable that the range of effectiveness of a carcinogen could vary by several orders of magni tude, and the overall probability of a giv en molecule's being effective could be come very small indeed. As the outcome of cancer appears to be determined by the balance against the effectiveness of a single insult entity and the frequency of available entities, the regulatory) process attempts to identify the quantity of insult to which man can be exposed without an unacceptable chance of developing cancer. Because it would be unethical to con duct prospective testing in man. animal experimentation has been used as an al ternative, with the implication that car cinogenesis data from animal experi ments can be translated to human condi tions. Testing for Carcinogens in Animals Using animals to test carcinogens has its roots in basic research where the prin cipal concern was not and is not assess ment of real-life risk but the study of the phenomenon of carcinogenesis. For that purpose negative results are unfruitful and there is understandable pre occupation with increasing the odds of inducing cancer. Hence, the practice de veloped of using maximum tolerated doses, and this in species and strains chosen for their susceptibility. So it happens that current guidelines for carcinogen bioassay are replete with precise directions for the control of room temperature, air changes, humidity, and other easily controlled conditions but of ten suggest the introduction of deliberate bias into the experimental design (7, 15. 16), for example: "Both sexes of each of at least two species of animals should be used in the test throughout their lifespan. In most cases these species would be rats and mice. Hamsters and dogs might be suitable, but guinea pigs, for ex ample, appear to be resistant to some known carcinogens" (17, p. 8) and pre sumably should not be used; and again ". . . considerations in selecting the proper species and strains should in-! elude . . . sensitivity to tumor induc tion . . ." (15, p. 4), or "Generally such decisions have been made on the basis of the most sensitive species tested" (16, p. 3). Also, testing guidelines in general pre scribe feeding whenever that 'is more 257 AP00024156 convenient than other modes of adminis tration. Agents that in reality are ab sorbed by respiration or through the skin are thus subjected to abnormal metabolic processing and may impinge on cellular and organ systems that are not their nat ural targets; in general, feeding results in higher maximum-tolerated doses, be cause other routes offer less protected, more direct and rapid access to receptors that determine acute toxicity. Such rec ommendations are justified in the design of a research experiment of self-contained validity, but are difficult to reconcile with the need to obtain results of general value, particularly when other powerful obstacles exist. For instance, if seems reasonable to conclude that, because the probability of developing cancer from natural exposure is related to the number of cells present in an animal and to the duration of its life, aged mice should have natural can cer incidences much lower than aged hu mans. In fact they have comparable in cidences, which suggests that mice could be from 3 x to4 to 10" times more can cer-prone than humans (18, 19). The compelling power of such an argument cannot be dismissed simply because of its simplicity. Further bias derives from the usual prescription of nearly toxic "maximum tolerated" doses. Although these may not appreciably affect the animal's vis ible condition during the experiment, they are known to cause metabolic over loads that may unprcdictably promote or retard a carcinogenic process, with out comes that differ from species to species. Disturbing questions on this issue have been raised by many reports and studies (//. 16, 17, 20-26) but usually go unan swered when actual recommendations for testing are made (7, 11, 15-17, 20, 23 , 27). Diet is likely to be a major source of experimental variation. Early observa tions (25) on the effect of caloric intake, of dietary fat and protein, have been fol lowed by an even broader appreciation of the enzyme inducers and toxicants that may act, for example, on the im mune system (29-31). Moreover, when the agent being tested is a promoter, the presence of carcinogenic contaminants in the diet may erroneously result in its classification as a carcinogen, with out comes that may vary from species to species and from diet to diet. Concern with dietary disturbances has been voiced in many reports: ", . . some natural constituents of the diet or even an essential hutrient, such as selenium, may constitute a carcinogenic risk. Clearly these substances cannot be corn- 258 plctely excluded from the diet" (17, p. 5): or "What can be the significance of the incidence of. . . tumors in susceptible strains when one is not certain about the presence of carcinogenic contaminants in the diet on which animals have been maintained?" (23, p. 427). The need to control diets, although frequently recog nized (23), is still an unresolved problem in the official guidelines for carcinogen esis experiments (15). Another source of difficulty is the translation of animal pathology into terms of human significance. Individual agents can produce different tumors in different species or only in certain spe cies, thereby implying a variety of organ otropisms probably related to widely dif ferent metabolic conditions and homeo static mechanisms of cell proliferation and repair in different species. Hepa tomas, for instance, are very frequent in rodent tests but remarkably rare in man, and the oncogenic viruses commonly in festing small rodents may be one reason for the unusually high frequency of lym phomas in these animals. The difficulties of comparison under these conditions are further complicated when tumors arise from tissues of different embryologic origin in different species. The biologic implication is that different agents may be carcinogenic for certain species or particular organs but relatively harmless for others, for reasons that are not yet apparent to science. Kraybill (29) lists a number of other sources of quantitative uncertainty in an imal testing, including inappropriate routes of administration, enhancement of susceptibility by deliberate immuno suppression or induced hormonal action, contaminants in the agents being tested, accumulation of a burden of the agent or other uncontrolled compounds in certain tissues, the theoretical and practical dif ficulties in matching duration of ex posure in man and animals, and dif ferences in the time required for tumor formation. One could add recent findings on the quantitative disturbances caused by various environmental and chemical stresses (10,52), the use of rodent strains contaminated with endemic oncogenic viruses and of selected or inbred strains (25), and the effects of transient infec tious contaminants (35). In general, one can only conclude that current guidelines for the testing of car cinogens frequently introduce deliberate bias in order to enhance the probability of a positive response and that they ig nore a number of sources of variability that cannot be controlled or are difficult to control with available technology. Un der current testing a carcinogen may go undetected in a particular test, but just as likely a positive result may be valid only for the particular species and test conditions utilized; current science can not predict or explain the outcome. Current Methods of Assessing Human Risk In present regulation, data from ani mal tests have been used, first, to define a particular hazard as a carcinogen--ac cording to the general definition dis cussed above--and, second, as a basis for extrapolating to presumed conditions of human exposure, The first use has been challenged be cause maximum tolerated doses may in hibit the appearance of tumors, as in the case of vinyl chloride (25) and other compounds, and because the metabolic overload created by such doses is likely to derange normal homeostasis and create physiologic conditions with no real-life counterpart (16, 17, 21, 23, 24, 26,30,34, 35). However, since false neg atives are difficult to count, popular con vention has it that current practices en hance the probability of detecting carcin ogens in animals and prudence dictates that those detected should be deemed potential carcinogens for man. While the latter argument may be defensible (36), it does not provide scientific justification for the codified practice of using maxi mum tolerated doses (37, 38). Of course, identification of an animal carcinogen is only a first step in the regu latory process, unless it happens to come under the provisions of the Delaney clause (59). In that case the regulatory verdict is unequivocal, because the law states that any substance against which there is evidence must be banned. This legislation has endured for over 20 years, but lately increased analytical sophisti cation and expanded testing activities have begun to raise public opinion in fa vor of a less intransigent approach. Congressional action, and the tempo rary suspension of the Delaney require ment, in the case of saccharin is the most recent example of this trend (40). The real-life question in this case is whether the risk from exposure to artificial sweet eners is balanced by risks that users would incur without them from diabetes, excessive calorie intake, dental caries, and so on, or simply by hedonistic re wards. Similar questions are likely to become a major issue of regulatory action in the near future, influenced by emerging atti tudes toward no-effect thresholds and to ward the limitations of animal test data. SCIENCE, VOL. 20* AP00024157 The issue of no-effect thresholds will in evitably assume importance as smaller and smaller quantities of potentially haz ardous compounds become identifiable through advances in chemical methods. Up to now, the probable occurrence of thresholds has usually been ignored, and some regulatory guidelines specifically prevent considering them (7, 15). Such an attitude largely results from avoiding the distinction between the practical and the theoretical. Difficulties in conceiving or measuring thresholds in cellular and molecular contexts have been taken as reason to question the reality of those practical levels below which adverse ef fects cannot be measured epidemiologically. Tolerable limits of exposure (TLV) are a common concept in regulation, and while it is true that epidemiologic defini tion of practical thresholds has been dif ficult except in rare instances (41), their presence is suggested by much evidence (II, p. 10:42. 43) which parallels univer sally accepted concepts in chemistry, physiology, and pharmacology. Deliber ate laboratory and epidemiologic studies on this problem could supply informa tion of direct significance to regulation. Regarding the use of animal test data for human risk assessment, severe obsta cles were recognized very early in sever al documents (6, 7, 11, 15, 17, 20-24, 27, 37-39), but the logical conclusions were not drawn: experimental practices quite valid in a basic research setting were adopted for regulatory purposes without a critical analysis of their limitations. Once these practices were established, support was sought for them in several biometric models specifically developed to attempt a generalized quantification of human risk (7, 44. 45). These statistical exercises would be justified if the animal data used in their elaboration reflected generalized human risk conditions, but they do not: nor is there a basis for de ciding in which direction their results should be adjusted. The situation was recognized by an expert panel of advi sers to FDA (23, p. 433) a few years ago: ". . . it would be imprudent to place ex cessive reliance on mathematical sleight of hand, particularly when the dose-re sponse curves used are largely empirical descriptions, lacking any theoretical, physical or chemical basis." Apparently statisticians have prudently competed with each other to produce methods that would give the most conservative esti mates, as the same FDA panel of experts noted (23, p. 435): "Although it is pos sible in principle to estimate 'safe' levels of carcinogens, uncertainties involved in the downward extrapolation from test re suits will usually result in permissible levels that are the practical equivalent of zero." In a general appraisal of current procedures for human risk determina tion, a compelling statement has been re cently advanced by Kraybill (29): ", . the [carcinogenic] response ... is mediated and limited by certain bio chemical, metabolic, and pharma cokinetic relationships. Such boundaries must not be exceeded in biologicah test ing and assessment of carcinogens, lest irreconcilable implications are left with the scientific community and the public, which result, in the long run, in a waste of national resources in the interest of public health." The conclusion is that past and current testing practices do not unequivocally identify human carcino gens and do not yield quantitative infor mation about conditions of human risk, and that biometric sophistication does not overcome the limitations of these data. An impasse is being felt in debates about regulations (46), also reflected in confusion and contradiction at the inter national level (3) as regulatory guidelines are elaborated by several agencies em powered by recent statutory mandates (15. 47). Most of these attempts are based on the traditional assumption that animal tests allow reliable and genralized quantitation of real-life carcino genic risk for man. Over the last decades it has been fash ionable to contrast the forthright sim plicity of science with the apparent looseness of political debate. Sciehtific solutions are implicitly expected for many social difficulties: but science can draw valid conclusions only on the basis of proven theories, controlled methods, and consistent results, none of which are yet available in carcinogen testing. Ought scientists countenance the use of inconsistent data even for such worthy causes as human health and a whole some environment (48, 49)1 This ques tion becomes yet more embarrassing if one considers that better safety might be achieved, with greater fairness, by an ap proach that explicitly recognized the sociopolitical nature of regulation and re sisted the temptation to force arguments under scientific disguise. Future Regulatory Directions Discouraging as it may seem, it is not plausible that animal carcinogenesis ex periments can be improved to the point where quantitative generalizations about human risk can be drawn from them. A multitude of disturbing variables is in volved. There are difficulties from a lo gistic and a design point of view. Even the expedient of large experiments is now regarded as an improbable solution, because background noise and sources of disturbance will increase with the number of animals. Nor will current pro posals of more complicated testing pro cedures provide a solution (50). because the real issue is the fundamental biologic difficulty of resolving the inconsistency of chronic response in different species. In vitro tests remain a possibility that is probably several years from practical application (J/). With current procedures, the assay ca pacity in the United States is limited to a few hundred compounds a year at best: the backlog of compounds that need to be tested and the new compounds that industry would like to have tested amount every year to several tens of thousands of individual items. The new Toxic Substance Control Act alone (32) is likely to create a crisis that could only be resolved by adopting new regulatory policies, expanding resources, and sim plifying testing requirements. The crisis would be exacerbated by the continuing pressures of a consumer society that is also environment-con scious. These could swell the outcry over what appears as an exorbitant or impossible regulatory burden, force the mitigation of current requirements for testing (34, 53. 54), and weaken enforce ment of statutes (35). However, societal concern on environmental issues during the last decade indicates that a rollback to nonenforcement is not very probable (56), short of a profound economic crisis and depression. It would seem desirable to think of an alternative scenario, one calling for offi cial recognition that risk is an unavoid able element of life and the common wel fare, that all human lives cannot be pre served at all costs, and that carcinoge nicity tests in animals cannot be reliable quantitative models of human risk. Es sential elements of such an approach have been identified and debated in a re cent report of the National Academy of Sciences (57). Today certainty in regulation is elu sive, and it is likely to remain so until adequate science develops. At the same time, judgment in the face of uncertainty does not call for an apology. Indeed, the current regulatory process may be in dis favor because it is not honestly judg mental and, by insisting on inadequate science and intransigent ideals, produces results that are perceived at times as ar bitrary, inconsistent, or unacceptable to the public at large. The central point of IS APRIL 1910 259 AP00024158 procedural reform is that resolution of uncertainties must be attempted in an open sociopolitical context, because use fulness, benefit, tolerable hazard, and safety cannot be defined on the inde pendent authority of scientific facts or statutory prerogative. The diversity of real-life situations would seem to make this task impos sible, but similar problems have been reconciled, traditionally, by flexible stat utes that offer standards of reference, to be used in the fair and consistent resolu tion of individual situations. For the regulatory process of our in terest, two sets of references need to be defined: a standard of usefulness or ben efit and a standard of safety or tolerable hazard. After this initial work, individual cases would be heard in open proceed ings, much as in a judiciary process. Initially, emphasis would be on identi fying functional classes of products and uses considered necessary to sustain a modem society. Analysis and definition of a standard of need would have to be extensive only for each class of use. For a particular agent it would have only to be proved that it belongs to the class, and its standing would improve if it of fered corollary benefits, such as addi tional therapeutic or nutritive properties. In other words, the analysis of benefits for individual agents could be largely set tled by precedent. The initial effort would eventually de fine standard categories of use, each being assigned a relative rank of useful ness. Primary items of need, such as bas ic foods, comforts, drugs, and fuels, and perhaps basic raw materials and chem ical intermediates, would be ranked at the top, less-needed items receiving less er ranking, depending on a sociopolitical judgment. This task need not have prohibitive di mensions. It has ample precedents in the legislative process, and would appear to be a natural function for Congress, per haps assisted by a systematic polling of public and expert opinion during the ex tensive activities necessary at the begin ning, and for revisions thereafter. Definitions and ranking would have to consider logistic, economic, hedonistic, esthetic, ethical, and other cultural is sues. In principle these criteria would have at least equal weight in a final judg ment. That man does not live by bread alone has never been so clear as in our time. The cultural mosaic of values that define happiness ought to be an impor tant element in the definition of useful ness and benefits, even while we take in to account the necessity of making choices among our desires. 260 Nevertheless, safety must remain an important objective, and in the process of ranking relative usefulness one could also identify a safety-standard agent rep resentative of each class, and prescribe appropriate use restrictions or tolerable conditions of exposure. This could be based on the minimum human intake, or environmental load, compatible with the fulfillment of that use, and would also de pend on the rank of usefulness of the par ticular category. The standard-of-need agent for each class would naturally be come its standard of safety or tolerable hazard, and new agents aspiring to be classified for the same use would be compared for safety with that standard. For example, sucrose could be selected as the reference for sweeteners, because it is the most widely used substance for sweetening, and because a long record of chronic exposure in mankind suggests side effects that are either ignored or ac cepted as tolerable by the vast majority of users. The safety of another sweet ener would then be compared with that of sucrose, at doses also including maxi mum conditions of exposure under in tended human use. Much research would still be neces sary to improve our ability to predict the relative toxic potency of two different compounds in man, because the quan titative difficulties in translating results of chronic tests across species would persist. In fact, even the direct human evidence of epidemiologic studies is not always sufficient for a regulatory deci sion, because of apparent or suspected confounders. Undoubtedly, complex assay protocols would be suggested: different routes of administration in different species, ex tensive dose-response kinetic studies, metabolic fate determinations, structureactivity inferences, chronic and acute toxicity tests, in vitro assays with human and animal tissues, and other approach es. The redundancy of these suggestions underlines their relative impotence, be sides being incompatible with the limited testing resources now available. All things considered, it would seem reasonable that until better methods for the definition of relative toxicity can be found, the role of science in regulation should be limited to those instances where nearly certain assessment of hu man risk is feasible and legitimate; at the same time more emphasis should be giv en to methodological and basic research for future application. In this light, while carcinogenicity may not be measured reliably today, relative safety could be defined by a formula that would assign nearly equal weights to oth er forms of acute and chronic toxicity, the tests being selected when their gener alization to real life is reasonable. The burden of proof would be left with the applicant, who could present the case to ajury of experts and users acting in a set ting similar to a judicial proceeding to ar rive at an opinion about the toxic po tency of the substance in question rela tive to that of a reference agent. For the sweetener of our example, this judiciary proceeding might succeed in defining its rank relative to sucrose, based on its rel ative toxicity and the estimated dose from exposure^under the intended condi tions of use; and it could achieve the same rank as the reference if, for ex ample, it were twice as toxic but its in tended use resulted in only half the ex posure. Indeed, a safety judgment that is influ enced by criteria of need would have to consider exposure as a prime determi nant of hazard, and it may become nec essary to develop more sophisticated ap proaches for determining human intake by various routes, under real-life condi tions of an agent's proposed use (14). The sum of regulatory restrictions would finally depend on the rank of need for the class of use, more necessary ones commanding fewer restrictions, and on the safety ranking of the agent consid ered, relative to the reference compound and the use restrictions applied to it; there might be a range of restrictions for special situations of exposure, such as pregnancy, young and old age, allergies, workplaces. Because of the uncertain ties, precise numerical structures for reaching regulatory pronouncements are unlikely, even though the formulation of decision frameworks has been discussed and appears feasible (.57). But who shall make regulatory deci sions? This question becomes important because the new scenario implies a shift from normative bureaucracy to an exer cise in sociopolitical judgment. Society has repeatedly faced the challenge of regulators preoccupied with their own survival; and traditional normative man dates have come to be questioned as remnants of an autocratic past, particu larly when situations are not clear-cut but defined by a range of judgment. The present regulatory system itself cannot avoid this situation, and in fact most of the important regulatory decisions are fi nally resolved by litigation. It has been suggested (58) that it may become expedient to provide for an im partial and fair resolution of the uncer tainties involved by instituting special courts independent of the regulatory agencies, the latter being left witl] the SCIENCE, VOL. 208 AP00024159 task, of proposing regulation and enforc Spring Harbor Laboratory. Cold Spring Harbor, 35. A. C. Kolbye, Oncology 33, 90 (1976), I, ing the courts' decisions by devising N.Y., 1977), pp. 1813-1820. 36. The other often-advanced proposition that 13. It is in this context that precursors and proxi known human carcinogens, with notable ex clear categories of restriction, easily un t derstood and accepted by all consumers r and special users alike (57). In this con text, the ethical and operational in mal* or ultimate carcinogens are defined as well as initiators and promotors, acting genetically or epigenetically. It appears that cellular damage leading to higher cancer frequency docs not nec essarily have to be of direct mutagenic nature (16,19). As such, the effectiveness of functional ceptions such as arsenic, are also carcinogenic in animals, and therefore the reverse must also be true, is a generalization based on summarily matching less than two dozen recognized human carcinogens against some 2000 animal ones. 37. E. I. Goldenthal, Current Views on Safety Eval competence of intransigent statutes carcinogens could be modulated at any stage by uation ofDrugs, FDA Pap. (May 1968), pp. 1-8. other modifying factors which, although not car 38. "Pesticide Residues in Food," WHO Tech. might come to be viewed as an embar cinogenic per se, would so appear to an experi Rep. Ser. No. 545 (1974). rassment to be rectified, and as incon menter who notices only the resulting cancer in 39. Federal Food, Drug and Cosmetic Act, as cidence. amended October 1976, Section 409(c)3A (Gov sistent with the safeguards of due pro cess that are at the philosophical core of 14. G. B. Gori, in Air Pollution and Cancer in Man, U. Mohr, D. Schmal, L. Tomatis, Eds; (Inter national Agency for Research and Cancer. ernment Printing Office. Washington, D.C., 1976). 40. Saccharin Study and Labelling Act, Public Law a free society. Lyon, 1977), pp. 99-111. 15. Guidelines for Carcinogen Bioassay in Small 95-203. 23 November 1967 (Government Print ing Office. Washington, D.C.. 1967). The proposed approach might also Rodents (Carcinogenesis Program, National 41. G. B. Gori, Science 194, 1243 (1976), have important economic consequences, Cancer Institute, Bethesda, Md., 1976); "Scien 42. Q. E. Hutchinson, Proc. Natl. Acad. Sci. tific Bases for Identification of Potential Carcin U.S.A. 51, 930 (1964); B. D. Dinman. Science because of the attribution of rank to each compound within a class, and of use re ogens and Estimation of Risk." Fed. Regist. 44, 39858 (1979). 16. Committee for the Revision of Publication No. 175, 495 (1972); G. 1. Claus. K. Bolander, K. Bolander.'Foorf Cosmet. Toxicol. 12.737 (1973). 43. Arsenic, for instance, otherwise a recognized strictions depending on rank. Clearly, a regulatory process that ranks efficacy 1138, NAS Committee on Toxicology, Prin ciples and Procedures for Evaluating the Tox icity of Household Substances (National Acad human carcinogen, must have a nontoxic thresh old because it is essential to the hematopoietic recess and as a catalyst in phosphorylation, and relative safety risks new dangers of emy of Sciences-National Research Council, Washington. D C.. 1977). imilar considerations are valid for mckel, chro mium, selenium, and other agents (26); and the intransigent interpretation. But when 17. "Evaluation of the Carcinogenic Hazards of presence of practical no-effect thresholds is such a policy is exercised as social judg Food Additives/' FAO Nutr. Meetings Rep. Ser. No. 29 (1961). clearly documented in smokers, not all of whom develop lung cancer or other smoking-depen ment, it could add a new incentive to de velop increasingly better products (57). 18. J. Cairns, Cancer: Science and Society (Free man, San Francisco. 1978); R. Peto, Proc. R. Soc. London. Ser. B 205, 111 (1979). dent diseases (41). Organotropism ofcertain car cinogens also implies cellular and tissue thresh olds; and the resistance of certain species to The future of manufacturing may well be characterized by restraints and solu 19. R. Peto, in Origins of Human Cancer, H. H. Hiatt, J. D. Watson, J. A. Winsten, Eds. (Cold Spring Harbor Laboratory, Cold Spring Harbor, known carcinogens, even st maximum tolerated doses, ought to be taken as evidence that noeffect thresholds are a most common class of tions unthinkable 10 years ago and only N.Y., 1977). 20. "Procedures for Investigating Intentional and real-life phenomena. Moreover, because the carcinogenesis outcome is time- and dose-dp- barely felt today, chiefly reflecting the in Unintentional Food Additives," WHO Tech. pendent, the finite lifespans of man and animals evitable depletion of raw commodities. Rep. Ser. No. 346 (1967). 21. Drag Research Board NAS-NRC, Clin. Phar are bound to impose no-effect thresholds at some level of exposure (see H. Druckrey, in Po Jn that context, a new regulatory posture of the general nature suggested becomes macol. Ther. 5, 607 (1969). 22. "Principles for the Testing and Evaluation of Drags for Carcinogenicity/' WHO Tech. Rep. tential Carcinogenic Hazardsfrom Drugs: Eval uation of Risk, R. Truhaut, Ed. (Springer, Ber lin, 1967), vol. 7, p. 60J. even more plausible, as it would provide incentives for a more farsighted utiliza Ser. No. 426 (1969). 44. M. A. Schneiderman, N. Mantel, C. C. Brown, 23. Panel on Carcinogenesis Report on Cancer Test Ann. N.Y. Acad. Scf. 246, 237 (1975). ing in the Safety Evaluation of Food Additives 45. D. Hoe) and N. Chand, in Reliability and Bio tion of diminishing resources while help and Pesticides. Food and Drag Administration Advisory Committee on Protocols for Safety metry, F. Prosehan and R. J. Serfling, Eds. (Siam, Philadelphia, 1974), p. 382; J. Cornfield, ing to preserve the values of enterprise. Evaluation, Toxicol. Appl. Pharmacol. 20, 419 Science 202, 1107(1978). Reference* and Nate* (1971). 46. L. i. Carter. Science 204, 811 (1979). 24. The Testing of Chemicals for Carcinogenicity. 47. J. Walsh, ibid. 202, 598 (1978). Mutagenicity and Teratogenicity (Health and 48. C. Comar, ibid. 200, 1225 (1978). 1. P. Pott, Cancer Scroti: The Chirurgical Works of Percival Pott (Clark & Collins, London, 1775), p. 734. 2. Statistical Abstracts of the United Stoles (Gov' emment Printing Office, Washington, D.C.,` 1959), table 3;/M (1960). table 11.ibid. (1977), table 104. 3. R. Montesano and L. Tomatis, Cancer Res. 37, 310(1977). 4. R. Doll, Proc. R. Soc. London, Ser. B 205 . 47 H979). 5. S. Epstein, in Origins of Human Cancer, H. H. Hiatt et at., Eds. (Cold Spring Harbor Laborato ry. Cold Spring Harbor, N.Y., 1977). 1727-1728. 6. "Procedures lor the Testing of Intentional Food Additives to Establish Their Safety for Use/* WHO Tech. Rep. Ser. No. 9 (1955). 7. "Drinking Water and Health, Recommenda tions of the National Academy of Sciences," Fed. Regist. 42, 35764 (1977). 8. Report of the Subcommittee on Environmental Carcinogenesis, National Cancer Advisoty Board, Criteria for Assessing the Evidence for Carcinogenicity of Chemical Substances (Na tional Cancer Institute, Bethesda, Md., 1976). 9. Even current advances in mechanistic hypothe ses, such as the experimentally justified multi stage carcinogenesis theory, although they give useful insight into this matter, do not clarify the final molecular complexities of transformation (12, 19). 10. V. Riley, Science 189. 465 (1975);_____ and D. Spademan, in The Pigment Ceil, vol. 2, Mela nomas: Basic Properties and Clinical Behavior, V. Riley. Ed. (Karger, Basel. 1976), pp. 163173;Lob. Anim. 4, 16(1977). 11. "Assessment of the Carcinogenicity and Muta genicity of Chemicals," WHO Tech. Rep. Ser. No. 546 (1974). 12. J. Cairns, in Origins of Human Cancer, H. H. Hiatt, J. D. Watson, J. A. Winsten, Eds. (Cold Welfare Department of Ottawa, Canada. 1973). 25. C. Maltoni and O. Lefemine, Environ. Res. 7, 387(1974). 26. H. G. Smyth. Food Cosmet. Toxicol. 5, 51 . (1967). 27. Appraisal of the Safety of Chemicals in Foods, Drugs and Cosmetics (Association of Food and Drug Oflkials of the United States, Baltimore, Md., 1975), pp. 79-82. 28. A. Tannenbaum. in The Physiopathology of Cancer, F. Homburger, Ed' (Hoeber, New York, 1959), pp. 517-562. 29. H. E. Kraybill, in Human Epidemiology and Laboratory Correlations in Chemical Carcino genesis, F. Coulston and P. E. Shubik. Eds. (Ablex, Norwood. N.J., 1980). 30. E. J. C. Roe and M. J. Tucker, Proc. Eur. Soc. Study Drug Toxic. 15, 171 (1974). 31. H. F. Kraybill. Clin, Pharmacol. Ther. 4, 73 (1963). 32. Some official guidelines for the care of experi mental animals actually prescribe lighting condi tions that are known to produce rapid blindness in rats commonly used in carcinogenesis experi ments, with likely profound effects on the com plex neurochenucal functions of the pineal gland. Department of Health, Education, and Welfare, Guide for the Care and Use of Labora tory Animals (Government Printing Office, Publ. No. 7423, Washington, D.C., 1974); W. K. NocU, V. S. Walker, B. S. Kang, S. Berman.in vest. Ophthalmol. 5, 450 (1966); W. K. Noetl and R. Albrecht, Science 172, 76(1971); W. K. O'Steen, K. V. Anderson. C. R. Shear, Invest. Ophthalmol. 13, 334 (1974); K. V. Anderson, F. P. Coyle, W. K. O'Steen, Exp. Neurol. 35, 233 (1972); J. Axelrod, Science 184. 1341 (1974); R. J. Wurtman and M. A. Moskowitz, N. Engl. J. Med. 296, 1329(1977). 33. V. Riley et ai.. Science 200, 124 (1978). 34. Q. E. Moore, ibid. 199, 1157 (1978). 49. M. G. Morgan, ibid. 201,971 (1978). 50. R. J. Smith, ibid. 204, 1287 (1979). 51. These tests are not yet fully developed; it is con ceivable that they could yield useful data, partic ularly if they made use of normal human cells, (issues, and organs. Coupled with metabolic studies in man or human tissues, they might pro vide quantitative data more reliable for certain human extrapolations than those supplied by current animal testing (16). 52. Toxic Substances Control Act, 11 October 1976, P.L. 94-469 (Government Printing Office, Washington, D.C., 1976); "General Provisions and Inventory Reporting Requirements, Supple mental Notice," Fed. Regist. 42, 19298 (1977); "General Provisions ana Inventory Require ments," ibid., p. 39182; "Notification of Sub stantial Risk under Section 8(e)/' ibid., p. 45363; "Supplemental Notice to Proposed In ventory Reporting Requirement; Drift Report ing Forms," ibid., p. 53884; "Inventory Report ing Requirements/* ibid,, p. 64572; Office of Management and Budget, Standard Industrial Classification Manual (Government Printing Of fice, Washington, D.C., 1972); Office of Toxic Substances, Environmental Protection Agency, Candidate List of Chemical Substances (1977). vols. 1-3; Addenda I and 2 (1978); Activities of Federal Agencies Concerning Selected High Volume Chemicals (1975). 53. L. J. Carter, Science 202, 30 (1978). 54. P. H. Abelson, ibid. 200. 487 (1978); ibid. 202, 473 (1978). 55. B. J. Culliton, ibid. 201, 1198 (1978); R. J. Smith, ibid. 203, 28 (1979). 56. Committee for a Study on Saccharin and Food Safety Policy, Food Safety Policy: Scientific andSocietal Considerations (National Academy of Sciences, Washington, D.C., 1979), pt. 2. 57. E. Marshall, Science 202, 949 (1978). 58. D. L. Bazelon, ibid. 205, 277 (1979). 18 APRIL 1980 261 AP00024160 system's own energy production. The l factors that increase the magnitude and duration of this drain are rapid growth rates of the energy systems. long times for single production units to achieve an energy payback, and long construction The Energy Impacts of Solar Heating times. The interrelation between these system characteristics and their influ Chris Whipple ence on net energy system outputs has been independently analyzed a number of times (J. 4). Further applications, usu ally to nuclear power systems, can be found for specific integration paths (5). The attractions of solar energy are ob industrial demand is largely met with in For solar heating technologies the net vious.: it is nondepletable. provides im digenous fuels--coal in the case of the energy dynamics can be important be munity to embargoes and fuel price in steel industry, and coal, nuclear, and hy- cause. relative to conventional energy creases. and is generally less damaging droelectricaily produced electricity in sources, solar energy collectors and con to the environment than conventional the case of the aluminum industry--the verters substitute an initial capital in sources (/). Further, specific solar tech effect of solar heating is both to sub vestment (frequently of energy-intensive nologies (for example, water and space stitute for depletable fuels and to cause a materials) for a lifetime of fuel consump heating) have achieved or are approach shift to domestic fuels. During the rest of tion. As such, the energy costs of solar ing costs that are competitive with con the century, this shift may be more im facilities are front-end costs. As a con ventional sources in many regions of the portant than the net energy produced in trast. the energy needed to provide coal country (2). helping the United States meet energy and equipment for a coal-fired system is The purpose of this article is to exam demand and reduce imports. more evenly distributed over the oper ine the impact of a rapid implementation In discussing the impacts of solar heat ating life of the plant (6). An additional of active solar space heating and water ing I first consider the general case of net reason for examining the net energy dy namics of solar energy is the rapid rate of growth of solar energy use proposed for Summary. The energy required to build and install solar space- and water-heating the remainder of the century. equipment is compared to the energy it saves under two solar growth paths corre As an illustration of the relative impor sponding to high and low rates of implementation projected by the Domestic Policy tance of net energy in a dynamic sense, Review of Solar Energy. For the rapid growth case, the cumulative energy invested to the payback time (at which cumulative the year 2000 is calculated to be Vj to 11h times the amount saved. An impact of rapid solar heating implementation is to shift energy demand from premium heating fuels (natural gas and oil) to coal and nuclear power use in the industries that provide energy investments and outputs are equal) for a system expanding linearly in time is exactly twice the payback time materials for solar equipment. for a single unit. For exponential growth paths, the system payback time is an in creasing function of both the growth rate heating on the U.S. energy supply. The energy analysis for an expanding supply and single unit payback time. findings are considerably different from system. Second, 1 review several esti For the analysis in this case, a number those of the recent Domestic Policy Re mates of the energy requirements of so of simplifying assumptions have been view of Solar Energy (DPR) (/), a multi lar heating equipment and adapt them to made: agency analysis of the possible range of the needs of this study. Third. I select 1) The time between solar heating implementation of solar energy during two solar heating integration paths to equipment manufacture and operation is the balance of the century. The dif represent low and high growth rate sce assumed to be zero. If this is not so, for ference is due to the fact that in calcu narios. The paths chosen were selected example, because of slow inventory lating solar heating contributions, I in to represent, as far as possible, an ex turnover in the industry, then this analy clude the energy required for construc ponential fit to high and low cases in the sis overestimates the net energy output. tion of the solar facilities. When these DPR. I then combine these results to 2) The energy requirements of the fa energy inputs are considered, the energy estimate the net contribution of solar cilities required for solar equipment saved by a rapidly expanding solar heat space and water heating (as contrasted manufacture, or for associated indus ing system becomes significantly less with the more prevalent gross energy tries, are not considered in either a static <*r. than if calculated without considering projections), and discuss the projected or dynamic sense, except to the extent these factors. Offsetting this result is the impact of solar energy implementation that this energy has been amortized in I very favorable effect of solar heating on on fuel mix. the specific mix of fuels. the estimates of materials energy con tent. For example, if rapid expansion of If solar heating is rapidly adopted, it glass manufacturing capacity accom will displace premium heating fuels (oil The Dynamics of Net Energy panies the growth of solar collectors, the gS and gas). This conversion to solar heat use of historically "average" energy in ing will shift load from the residential It is possible for an expanding energy tensity for glass will neglect the dynam and commercial sectors, in which the so system to act as a substantial energy ics of its expanding system. Similarly, lar equipment will be used, to the indus trial sector, and particularly to the steel and aluminum industries. Because this drain, because the consumption of ener gy for the' creation of new production fa cilities for the system can exceed the The author is Technical Manager at the Energy Study Center, Electric Power Research Institute, Palo Alto, California 94304. 262 0036-8075/8010418-0262501.00/0 Copyright C 1980 AAAS SCIENCE, VOL. 208, 18 APRIL 1980 1 < < AP00024161