DOI: 10.1126/science.831254 , 17 (1977); 195Science et al.A Blum,

hazardsFlame-retardant additives as possible cancer

www.sciencemag.org (this information is current as of November 28, 2006 ):The following resources related to this article are available online at

http://www.sciencemag.orgversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org#otherarticles, 5 of which can be accessed for free: cites 18 articlesThis article

http://www.sciencemag.org#otherarticles 10 articles hosted by HighWire Press; see: cited byThis article has been

http://www.sciencemag.org/help/about/permissions.dtl in whole or in part can be found at: this article

permission to reproduce of this article or about obtaining reprintsInformation about obtaining

registered trademark of AAAS. c 2005 by the American Association for the Advancement of Science; all rights reserved. The title SCIENCE is a

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Thousands of chemicals to which hu-mans have been exposed have been in-troduced into the environment withoutadequate toxicological testing. The tox-icological and biological properties offood additives and drugs have been moni-tored by the U.S. Food and Drug Admin-istration and now pesticides are moni-tored by the U.S. Environmental Protec-tion Agency, but most other new sub-stances are tested only superficially.Some chemical flame retardants pro-

vide a good example of a technologicalinnovation where adverse environmentaleffects may outweigh some of the bene-fits. Recent federal regulations, requiringthat children's sleepwear, mattresses,mattress pads, and carpets meet flam-mability standards, are said to have re-sulted in a decrease in the number ofburn injuries and deaths (1). As a result,flammability standards to cover all chil-dren's and adults' clothing, tents, sleep-ing bags, curtains, and upholstered furni-ture are being considered. Currentlyabout 300 million pounds of flame-retard-ant chemicals are being produced mainlyfor use in fabrics, plastics, and carpets(2, 3). Those added directly to textilesare often present in amounts as high as10 to 20 percent of the weight of thefabric. Further extension of the scope ofthe standards may increase their produc-tion and usage even more.

Inevitably, some fraction of the manymillions of pounds of flame retardantsthat are being produced will find theirway into people. The chemicals are rub-bing off on children's skins, may be in-haled from furniture, rugs, and tents,and, after "disposal" into the environ-ment, may enter the food chain. Thedecision to further extend flame-retard-ant standards should not be based onlyon the benefit of a reduction in firedeaths and injuries. The possible risk tothe population and environment of thewidespread production, use, and dis-7 JANUARY 1977

persal of these potentially hazardousflame-retardant compounds should alsobe taken into account.

Until recently, little attention was paidto the long-term biological effects ofthese flame-retardant compounds. Themain organic chemicals used in flameretardants contain bromine or chlorineor they are phosphate esters. Some havechemical structures (discussed below)that are closely related to compoundsknown to cause cancer or to be toxic toanimals. Several compounds previouslyused as flame retardants have beenshown to be teratogenic, carcinogenic,mutagenic, or highly toxic (4). In thisarticle, we discuss the implications ofthe finding that tris-(2,3-dibromopropyl)phosphate (tris-BP) the main flame retard-ant currently used in children's pajamas,is a mutagen (see Fig. 1).

History of Flammable Fabric Standards

The history (5) of the use of chemicalflame retardants goes back more than300 years to a treatment for canvas usedin Parisian theaters in 1638 and a reportfrom Oxford on a piece of unburnablecloth in 1684 (6). The French king LouisXVIII commissioned Gay-Lussac to finda way of protecting fabrics used in thetheater. In 1820 Gay-Lussac found thatammonium salts of sulfuric, hydro-chloric, or phosphoric acid were effec-tive in reducing fabric flammability (6).This work remains valid and applicabletoday.The Flammable Fabrics Act in the

United States was passed on 14 Decem-ber 1953 to regulate the manufacture ofhighly flammable clothing such asbrushed rayon sweaters, which werefirst sold during the 1940's. The act wasintended to protect the public from the"unreasonable risk" of fires leading todeath, personal injury, or significant

property damage. A general wearing-ap-parel standard, effective 1 July 1954, es-tablished minimum flammability stan-dards to keep highly flammable apparelout of the marketplace. The act wasamended in 1967 to allow flammabilitystandards to be set for many additionalconsumer products. Standards for car-pets and rugs became effective in 1971,and for mattresses and mattress pads in1973. The first children's sleepwearstandard (DOCFF3-71) for sizes 0 to 6Xbecame effective on 28 July 1972. Chil-dren's sleepwear fabric exposed to a gasflame along its bottom edge for 3 secondsis required to exhibit a char length nogreater than 7 inches (1 inch = 2.54 cm),even after the fabric has been laundered50 times.

In 1972, the Consumer Product SafetyCommission was established and as-sumed jurisdiction over the regulation offlammable fabrics. A children's sleep-wear standard for sizes 7 to 14 becameeffective on 1 May 1975. The require-ments of this standard are similar to, butslightly less stringent than those for thesleepwear sizes 0 to 6X.The Consumer Product Safety Com-

mission is in the process of establishinguniform federal standards for uphol-stered furniture and tents. The NationalBureau of Standards has carried out fea-sibility studies for a standard (based onboth garment design and flame-retardantfabrics and treatments) to regulate allarticles of adults' and children's cloth-ing. Various state laws are being insti-tuted to regulate upholstered furniture,curtains, tents, and sleeping bags al-though these may be preempted by feder-al regulations.

Consequences of Flammable FabricStandards

The ever-increasing scope of govern-ment regulations is leading to a vastlyexpanded market for chemical flame re-tardants. In 1971, a total of approximate-ly 175 million pounds (1 kilogram = 2.2pounds) of flame-retardant compoundswere produced. In 1975, the amount haddoubled to over 300 million pounds, andit is expected to reach 500 million poundsby 1980 (2) although about two-thirds ofthis is inorganic material, such as alu-mina trihydrate and antimony oxide usedin the carpet industry (2, 3). Large num-bers of different organic chemicals,many of which are brominated and

Arlene Blum is a postdoctoral fellow and Bruce N.Ames is a professor in the biochemistry department,University of California at Berkeley, Berkeley94720.

17

Flame-Retardant Additivesas Possible Cancer Hazards

The main flame retardant in children's pajamas is amutagen and should not be used.

Arlene Blum and Bruce N. Ames

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

chlorinated, are being introduced forvarious applications (3, 6, 7).The Flammable Fabrics Act has had a

major influence on the type of fabricsused in children's sleepwear (8). For ex-ample, in 1971, 56 percent of children'ssleepwear was constructed of cotton and27 percent consisted of polyester-cottonblends. Four years later, in 1975, 87 per-cent of children's sleepwear was con-structed of various synthetics. No sleep-wear was made of polyester-cotton andonly 13 percent was cotton. Untreatedcotton and polyester-cotton do not meetthe flammability standards, and treat-ments have not been commercially com-petitive. As a result, flame-retarded poly-esters, acetates, triacetates, and inher-ently flame-resistant fabrics such as mod-acrylics and those containing polyvinylchloride have replaced cotton.

Selection of Flame-Retardant Chemicals

In 1913, the chemist William HenryPerkin defined the requirements for aflameproofing process in terms that re-main applicable (6, p. 166):

A process, to be successful, must, in thefirst place, not damage the feel or durability ofthe cloth or cause it to go damp . .. or dusty.It must not affect the colors or the designwoven into the cloth or dyed or printed uponit. Nothing (such as arsenic, antimony, orlead) of a poisonous nature or in any waydeleterious to the skin may be used and thefireproofing must be permanent. That is to sayit must not be removed even in the case of agarment which may be possibly washed 50times or more. Furthermore in order that itmay have a wide application, the processmust be cheap.

One or more of six elements-bro-mine, chlorine, phosphorus, nitrogen,boron, and antimony-are currentlyused in compounds to reduce fabric flam-mability because of their effectiveness(6). There may be particular tox-icological problems with organic bro-

mine and chloride compounds (in addi-tion to antimony). Organic bromides andchlorides are used as flame retardants insynthetic fibers and are thought to act (6)as free radical traps and thus to suppresscombustion. Burning is oxidation in thevapor phase, involving H-, OH-, and O0free radicals. The halogen may workby a mechanism: RBr + H- -- HBr+ R-.For man-made fibers, tris-BP is by far

the most important flame-retardant com-pound in use and perhaps 10 millionpounds a year are used in fabrics andplastics (9). It is almost exclusively usedin polyester, acetate, and triacetate fi-bers, as well as being the basis for asuccessful finish to acrylic carpets. Theuse of tris-BP is currently the most eco-nomical, convenient way to meet thechildren's sleepwear standards (9).

Textiles vary greatly in their flamma-bility (Table 1), and each type presents adifferent problem in reducing flammabili-ty (IQ). Cotton and other cellulose-basedfibers can be flame-retarded by impreg-nated cellulose phosphate esters formedby direct esterification of the cellulosemolecule with a phosphate of the flame-retardant compound. Most treatments ofcotton are based on tetrakis(hydroxy-methyl)phosphonium (THP) compoundsor phosphonates, which are polymerizedin the fabric. These finishes can resultin a loss of tear strength in the fiberof up to 30 percent (9). For cotton tex-tiles, about 20 percent, by weight offlame-retardant compounds is added onin order to meet the standard (9). Cottonand synthetic blends present more of aproblem. The synthetic part of the fabricmelts, and the cotton serves as a supportthat keeps the synthetic burning. Flame-retardant strategies for such blends arebeing developed, but no economicallysuccessful flame-retardant treatment forthem is yet available.

Ironically, wool, which is inherentlyfire resistant (Table 1) (11, 12), does not

Table 1. Burning characteristics of textile fibers (10).

Fiber Characteristics

Cotton Supports combustion, burns rapidly, afterglowsRayon Supports combustion, burns very rapidly, no afterglowAcetate or triacetate Supports combustion, melts ahead of flameNylon Supports combustion with difficulty, melts and carries flame away in

falling dropletsNylon 66 Does not readily support combustion, melts and carries flame away in

falling dropletsAcrylic Burns readily with sputteringModacrylic Melts, shrinks away from flame and sometimes burns very slowlyPolyester Supports combustion with difficulty, melts and carries flame away in

falling dropletsPolyolefin Melts, burns slowlyWool Supports combustion with difficultyVinyon Does not readily support combustion

18

meet the children's sleepwear standardset up by the Consumer Product SafetyCommission. The regulations require thefabric to be bone dry (desiccated in anoven at 105°C for 1 hour, followed by acooling period) prior to the flammabilitytest (5). Objections have been raised tothis bone-dry regulation as discriminat-ing against wool and cotton, which havehigh moisture contents (11).

Flame-retardant treatments requirecompromises in economy, esthetics, andwear properties. Consequently, exten-sive research, costing tens of millionsof dollars, is still being done by govern-ment and industry to try to find betterflame-retardant treatments, particularlyfor polyester-cotton blends.

Biological Properties of "Tris," the

Main Flame Retardant in Pajamas

There is a growing realization thatchemicals can be absorbed through theskin and that long-term toxicological ef-fects of chemical additives to clothingshould be characterized more thorough-ly. A manufacturer typically carries outonly short-term tests for toxicity, anduntil recently little attention was given tolong-term effects, such as carcinoge-nicity, mutagenicity, and teratogenicity.The need for studying these long-term

effects is illustrated by the case of tris-BP, the flame retardant used in abouthalf of children's sleepwear (9). Tris-BPis padded on to the surface of polyesterfabrics in amounts up to 10 percent ofthefabric's weight. As much as half of this iscalled "surface tris" and is susceptibleto extraction and possible absorptionthrough the skin (13). A pair of children'spajamas, weighing 200 g, could easilycontain 6000 mg of surface tris-BP. Threelaunderings would reduce surface tris-BPappreciably (13, 14), while not alteringthe flame-retardant qualities of the gar-ment. This is not done prior to sale re-putedly because of the high cost andbecause consumers are said to value new-looking garments.The absorption of tris-BP through skin

is suggested by several studies. As partof short-term toxicity studies (15) byMichigan Chemical, tris-BP applied der-mally to rabbits increased blood bromineconcentrations. In another study, theurine from a rat wearing a gauze padimpregnated with tris-BP was found tocontain high concentrations of a tris-BPmetabolite, 2,3-dibromopropanol (16). Inhumans (13) tris-BP is a low-level aller-gen, an indication of some absorptionthrough human skin (17). Tris-BP causeddelayed hypersensitization in human

SCIENCE, VOL. 195

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

beings as did fabrics containing largeamounts of this compound on the sur-face. The degree of sensitization fromvarious fabrics was related to the amountof surface tris-BP available. Its per-cutaneous absorption is not surprisingas, in general, chemicals in contact withskin can be absorbed into the body (18,19).The most important question is wheth-

er tris-BP is likely to cause cancer orgenetic defects. The "high purity-lowvolatile" tris-BP made by MichiganChemical and used for textiles contains0.05 percent of the impurity 1,2-dibro-mo-3-chloropropane (15). Dibromochlo-ropropane caused a high incidence ofsquamous carcinoma of the stomach inboth rats and mice as early as 10 weeks-after initiation of feeding (oral in-tubation) (20). In addition, 50 percent ofthe female rats developed mammaryadenocarcinomas. This study, by the Na-tional Cancer Institute (NCI), was pub-lished in 1973, before the recent wide-spread use of tris-BP.Three impurities in commercial tris-

dibromopropanol (also a metabolite oftris-BP), the carcinogen 1,2-dibromo-3-chloropropane, and 1,2,3-tribromopro-pane-as well as tris-BP itself are allrelated in structure to the known carcino-gen 1,2-dibromoethane (ethylene dibro-mide). Ethylene dibromide is used (morethan 200 million pounds in 1970) as agasoline additive and grain fumigant (20).We reported that it is a mutagen in theSalmonella test (21) in 1971 (22) (it hadpreviously been shown to be a mutagenin a variety of other microorganisms).Because of its widespread use and muta-genicity, the NCI tested it for carcinogen-icity. Ethylene dibromide was found tobe a potent carcinogen on feeding, pro-ducing squamous cell carcinomas earlyand in practically all the surviving malerats treated (20).Tris-BP and the other brominated al-

kyl compounds discussed above aremutagens in our Salmonellalmicrosometest (21). This test shows an extremelyhigh correlation (on the order of 90 per-cent) between carcinogenicity and muta-genicity (21), and has been used to pre-dict a number of carcinogens. Liver mi-crosomal enzymes that convert carcino-gens to their active (and mutagenic)forms are combined with Salmonella bac-teria that are used for detecting mutagen-ic compounds. Prival et al. and Rosen-kranz first carried out such tests (23),showing the mutagenicity of tris-BP andsome of its impurities.Dose response curves for the muta-

genecity of these five compounds areshown in Fig. 2. The mutagenic potency7 JANUARY 1977

CH2Br CH2Br CH2BrII

CHBr CHBr CHBrICH2 CH2 CH2Ns I

p110

Tris(2,3-dibromopropyl)phosphate

Brn Brn

Polybrominated biphenyl

CH2BrCHBr

CH2OH

2,3-Dibromoproponol

CH2BrCHBr

CH2CI

1,2-Dibromo-3-chloropropane

JH20H +

HOCH2-P-CH20H Cl

[ CH2OH

Tetrokis(hydroxymethyl)-phosphonium chloride

Fig. 1. Structures of flame retardants and related compounds.

of tris-BP is considerably higher thanthat of the other chemicals shown. Tris-BP is more mutagenic in the test thanseveral known human carcinogens (21).However, the use of the Salmonella testfor predicting the rough potency of car-cinogens remains to be validated, al-though preliminary results look promis-ing (21, 24). Tris-BP and dibromopropa-nol, although more mutagenic than thecarcinogens ethylene dibromide and di-bromochloropropane, do differ in requir-

Tris-(2,3 - dibromopropyl)-phosphate

° 2000a.

a-

00

C

/c*/> 1000 _ / 1,2-Dibromo-1 Q000 3-chloropropone

1,2,3-Tribromo-

2,3-Dibromo- proprne.4prpaop r

,_EthyIene dibromide50 100

9g of Compound Per Plate

Fig. 2. All compounds were tested on Salmo-nella strain TAIOO as described (21). Theamount of ethylene dibromide added was 10times that indicated on the scale. The datapresented for tris-BP, 2,3-dibromopropanol,and dibromochloropropane were obtained inthe presence of rat liver homogenate (20,ul S-9/plate, Aroclor induced) (21): human livergave similar results. The potency (revertantsper nanomole) (21) of the various chemicalsis: tris-BP (0.1; 25 with S-9), 2,3-dibromo-propanol (0.15; 1.9 with S-9), 1,2,3-tribromo-propane (1.4; 1.4 with S-9), 1,2-dibromo-3-chloropropane (0.5; 0.9 with S-9), ethylenedibromide (0.02; 0.02 with S-9).

ing activation by liver microsomal en-zymes for efficient mutagenesis. Many ofthese microsomal activating enzymes areknown to be present in human skin (18,19).

Bacterial tests showing that tris-BP isa mutagen suggest that it is likely to be acarcinogen, but animal studies are neces-sary for more conclusive evidence. Feed-ing studies (in which tris-BP was addedto the animal chow) with rats and mice,at two dose levels, are being carried outat the NCI. The results should be knownin 1977.

Recently tris-BP has been found todamage human DNA in vitro, to be apotent mutagen in Drosophila, and tocause unscheduled DNA synthesis in hu-man cells in tissue culture (the latter testis quite effective in detecting carcinogensand is an indicator of a chemical's abilityto damage DNA) (25).The possible consequences of the

widespread use of tris-BP are serious. Itdoes come off fabric, is at least topicallyabsorbed, is known to be a strong muta-gen, and may contain a potent carcino-gen as an impurity. Infants' and youngchildren's habit of sucking their clothingcould lead to its ingestion. Therefore,tris-BP poses a potential hazard as ahuman carcinogen and mutagen.

In addition to the hazard posed by tris-BP and its impurities to those who make,work with, and wear fabrics treated withit, an environmental hazard may, or maynot (15), be posed by its disposal in largequantities into water and soil. The simu-lated washing of six treated sheets in atotal volume of 30 gallons of water yield-ed about 6 parts per million (ppm) of tris-BP in the wash water. A concentration of1 ppm in water is sufficient to kill goldfishwithin 5 days (26).

19

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Biological Properties of Other

Flame Retardants

The safety of a treatment for flame-retarding cotton fabrics has also beenquestioned although it is not clear thatthere is a significant problem with thecurrent technology. Tetrakis(hydroxy-methyl)phosphonium chloride (THPC)is polymerized and oxidized in cotton as

a flame retardant, and this fabric hasbeen reported to release formaldehydeand chloride when it is wet (27, 28). Thisis causing some concern (9, 27, 28) as

there exists the theoretical possibilitythat these ingredients could form bis-chloromethyl ether, an extremely potentcarcinogen in rats that has also causedcancer in factory workers. Bis-chlo-romethyl ether can be formed under cer-

tain acidic conditions from formaldehydeand chloride (29). To avoid any possi-bility of exposure of workers during thecuring process (and to minimize freechloride in the garment) the Americanindustry has switched to a salt that doesnot contain chloride, tetrakis(hydroxy-methyl)phosphonium sulfate (30). It hasbeen suggested that this could still be a

problem in the wearing of treated cottonbecause appreciable amounts of formal-dehyde can be extracted by syntheticsweat from cloth flame retarded with a

THP salt, and sweat contains appreciablechloride ion (28). No bis-chlorornethylether was found in this extract, however(28); as its formation requires acidic con-

ditions, it probably is not a serious prob-lem in nightwear.A more obvious example of the haz-

ards of flame-retardant chemicals is thepolybrominated biphenyl (PBB) tragedyoccurring in Michigan (31). Inadvert-ently 500 to 1000 pounds ofa polybromin-ated biphenyl-based retardant were

packed into bags similar to those inwhich Michigan Chemical Companypacks the magnesium oxide feed additiveknown as Nutrimaster. The contents ofthese bags-the flame-retardant Fire-master-were mixed into a cattle feedmixture at the Farms' Services Bureau inBattle Creek, Michigan, and distributedthroughout the -state. Soon, Michigandairy farmers noticed that their animalswere beginning to suffer from loss ofappetite, lowered milk production, ex-

cessive spontaneous abortion, birth de-fects among offspring, and eventualdeath of affected stock. After the cause

was discovered to be chemical con-

tamination, about 30,000 cattle, 6000pigs, 1500 sheep, and 1,500,000 chickenswere destroyed because their tissues con-tained PBB at concentrations greater

20

than 1 ppm. At least 365 tons of feed,18,000 pounds of cheese, 2600 pounds ofbutter, 34,000 pounds of dry milk prod-ucts, and nearly 5,000,000 eggs were alsodestroyed.Farmers and their families from all

over the state who have eaten largeamounts of dairy products and eggs fromthe affected animals have PBB's in theirblood and fat, and some have begun toreport a variety of illnesses, although theconnection to the PBB ingestion has yetto be proved (31).PBB has been reported to concentrate

in fat (32). Caged fish kept on the PineRiver below the Michigan ChemicalPlant were reported to accumulate PBB(1 mg/kg) in their fat tissue. This repre-sents a concentration factor greater than10,000-fold within 2 weeks of exposureto very low concentrations of PBB (32).The long-term carcinogenic and muta-

genic effects of PBB's remain to be deter-mined. Cancer tests in animals are inprogress and teratogenicity in mice hasbeen reported (33). The PBB's are closerelatives of the carcinogenic and ter-atogenic polychlorinated biphenyls(PCB's), which are a worldwide healthproblem, in that they have been spreadthroughout the biosphere and have be-come concentrated in the food chain.The fact that less than a thousandpounds of the brominated PBB flameretardant has caused such widespreadand persistent damage raises the ques-tion of the eventual consequences of themillions of pounds of PBB's that wereproduced in the past. It also shows thetype of hazards that may be incurred inproducing hundreds of millions ofpounds of flame retardants (much of itcontaining organic bromine and chlorine)that will eventually end up somewhere inthe environment.The environmental bioaccumulation of

these and other flame-retardant com-pounds may be a problem. Simple'leach-ing of flame retardants from fabrics dur-ing manufacturing, laundering, and dis-posal could lead to their presence inwater supplies and sewage. For ex-ample, the flame-retardant pentabro-motoluene was found in a sewage plantin Sweden (34). Mirex (Dechlorane), aclose relative of Kepone (it also oftencontains some Kepone), has been usedas a flame retardant in plastics for manyyears, and is a carcinogen and teratogen(4), and bioaccumulates in fish andpeople. The environmental effects of theflame-retardant chemicals must be con-sidered as well as the effects of the prod-ucts of these chemicals after they haveundergone oxidation and photochemical

breakdown by sunlight or micro-organisms in sludge, soil, water, or com-post. Furthermore, studies should be car-ried out on the uptake, storage, accumu-lation, biochemical change, and elimina-tion of the major flame-retardantcompounds in fish and other aquatic or-ganisms, in birds, and in mammals.

Alternatives to Preventing Burns

Without Chemical Additives

1) Self-extinguishing cigarettes. Themajor single cause, accounting for aboutone-third to one-half (35), of the approxi-mately 12,000 fire deaths and $11 billionin losses in the United States each year(36) is tobacco-smoking materials (35).The most common fire death scenariowas found to be the residential furnishingfire started by tobacco-smoking materi-als; alone it accounts for 27 percent offire deaths. The next largest single causewas residential furnishing fires started byopen flames, which accounted for 5 per-cent of the United States fire deaths. Allother single causes were 4 percent orless.Most American cigarettes when put

down will continue to smolder (someeven more than 20 minutes), apparentlybecause of additives and cigarette de-sign. This long smoldering is the key fac-tor in starting fires (37). Cigarettes areavailable that self-extinguish in a fewminutes and are much less likely to startfires. We have tested a large variety ofcigarettes and have determined that afew brands will go out in less than 5 min-utes while the majority of brands burnfrom 15 to 22 minutes (38). One possi-bility for this is stated on the package ofone brand. "Light an ordinary cigarette... there's a chemical in it to keep itburning." In contrast, this brand claimsthat its cigarettes have "NO flavorings,saltpeter, or tars added."The major cause of fire deaths and loss-

es could thus be attacked at its source bythe introduction of self-extinguishingcigarettes. As an added benefit a signifi-cant number of forest fires might also beeliminated if cigarettes were to self-ex-tinguish in a few minutes instead of smol-dering for 15 or 20 minutes. Unfortunate-ly one piece of legislation to give the Con-sumer Product Safety Commissionjurisdiction over the flammable proper-ties of cigarettes was defeated in theHouse of Representatives recently afterits passage in the Senate (39).The Consumer Product Safety Com-

mission does have jurisdiction overmatches and is developing standards for

SCIENCE, VOL. 195

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

self-extinguishing matches (30). The ma-jor single cause of fires of children'sclothing is due to children playing withcigarette lighters and matches (35). Bookmatches have been designed that arepractical and yet child-proof (40), andtheir adoption might have a major ef-fect.

2) Fire prevention. Some of the effortand expense that is being put into exten-sive flame retarding should be put intofire prevention. More effort could be putinto additional consumer education onthe causes and prevention of fires. ThePublic Education Office of the Fire Ad-ministration is set up to do this, but has aminimal budget (41). Improvements indesign also could be encouraged fornightwear because loose, flowing gar-ments have been shown to be involved inmany more fires than more tightly wo-ven, close fitting garments (42). Stovedesign could be modified so childrencould not easily turn on the burners.Space heaters could be changed so thatit would be more difficult for people toget close enough to them to ignite theirclothing. Gas heaters should not be putin garage workrooms where solventsare used.

3) Inherently fire-resistant fabrics.These provide possible safer alternativesto the addition of chemical flame retard-ants. Modacrylics and matrix fabricsbased on polyvinyl chloride and polyvi-nyl alcohol are inherently flame retard-ant without the addition of chemicals.Multimillion-pound-capacity vinyl bro-mide plants are being built to providemonomer for flame-retardant fabric pro-duction (43). It is to be hoped the type ofproblems caused by the carcinogen vinylchloride, such as monomer residues inthe polymer and worker exposure, willnot reoccur in the textile industry with vi-nyl chloride and vinyl bromide. Flame-retardant additives that are covalentlybonded to the fabric or those that are po-lymerized and entrapped within the fi-bers may also be safer than those that arepadded on, such as tris-BP.

4) Standards for fabrics. Theseshould be examined to see whether tech-nicalities could be changed to minimizethe need for additives. For example, a"melt-drip" provision is in force forsleepwear sizes 0 to 6X, but not for sizes7 to 14 (44). Because of this technicality,tris-BP addition is necessary for polyes-ter in the sizes 0 to 6X despite the factthat polyester is relatively flame resis-tant. There is evidence that the "melt-drip" phenomenon does not constitute asignificant burn hazard (44). In addition,several examples are discussed (below)7 JANUARY 1977

where it appears that standards could beredrafted so as to not require additives incertain fabric types (nylon tents) or initems (sleepwear of infants, who are lessthan 6 months of age) in which the bene-fit seems to be relatively small.

Benefit and Risk

A federal report published in 1972states that about 200,000 burn injuriesand 4000 deaths are associated annuallywith flammable fabrics (42). This reportis often cited by proponents of stricterclothing flammability standards. How-ever, a recent study (45) for the NationalAdvisory Committee for the FlammableFabric Act suggests that these numbersmay be about ten times too high, and esti-mates that there are about 16,000 annualtextile-related burns and about 500deaths in the United States (45). About20 percent of these burn injuries anddeaths might be associated with night-wear in children; that is, the standardsmight possibly prevent 3000 burn injuriesand 100 deaths per year among 50,000,000children. These numbers are only veryapproximate because of uncertainties inthe burn statistics and the recent esti-mate might be severalfold too low. Inany case, burns are a serious problem. Italso is relevant that healed burn tissue isat greater risk for developing an epider-moid carcinoma several decades after-ward (46).Adding flame-retardant chemicals to

almost all children's pajamas, as a con-sequence of the Consumer Product Safe-ty Commissions's standards, most prob-ably is reducing the number of burns anddeaths due to children's nightwear catch-ing fire, although statistics are unavail-able. As we have indicated, there are al-so other ways of reducing fire injuries.The risk of the exposure of tens of mil-

lions of children to a large amount of achemical must be balanced against therisk of fire. A calculation (47) suggeststhat the risk from cancer might bevery much higher than the risk frombeing burned. Flame retardants (andmost other large volume industrial chem-icals) either have not been tested or havenot been adequately tested for carcino-genicity. The use of an untested chem-ical as an additive to pajamas is unaccept-able in view of the enormous possiblerisks.Even if tris-BP is found not to show

any statistically significant increase in tu-mors in the current NCI feeding study,possible absorption of a highly mutagen-ic chemical by millions of children still

poses a considerable risk. Would tris-BPmutate human germ cells? Would tris-BPbe a potent carcinogen if it were paintedon the skin (the actual mode of exposureto people) rather than fed, as in the NCIstudy?Even if a chemical were tested, and

were found to be negative in a thoroughanimal cancer test in two species, thisdoes not guarantee safety. A thoroughanimal cancer test usually involves a fewhundred test animals at most (comparedwith millions of children in the case oftris-BP). This is an inherent statisticallimitation in animal cancer tests, andhigh doses in the animal may only partial-ly compensate. Thus, a chemical thatwould cause a tumor increase of lessthan 5 percent may easily go undetected.That sort of increase in a population ofmillions would result in tens of thou-sands of additional cases of cancer.The National Commission on Fire Pre-

vention and Control (36) has suggestedthat consumers be given a choice wheth-er to buy flame-retarded fabrics or not.

The Commission does not favor unbridled ex-tension of flammability standards to all cate-gories of fabrics. Only grossly hazardous fab-rics and fabrics implicated in a very large num-ber of fire accidents should be banned fromthe marketplace. A preferable direction ofem-phasis is toward labeling requirements as tocombustion hazards. This would honor thecherished principle of free choice, while at thesame time informing consumers of potentialrisks and reminding them of the importance offire. If reinforced by consumer education onfire safety, labeling requirements would havethe effect of spurring manufacturers to im-prove the flame-resistance of fabrics.

However, as flame retarding adds an ad-ditional 10 to 30 percent to the cost of thegarment and often adversely affects thefeel and ease of its care, many con-sumers, particularly those with lower in-comes, would not choose to buy theflame-retarded garment (48). One of themain reasons that has been given for thedecision of the Consumer Product SafetyCommission to implement compulsorystandards is to protect the poor.The strictness of these standards, their

compulsory nature, and their further ex-tension should be critically reexamined.The benefits of flame retarding all chil-dren's clothing, adult sleepwear andclothing, and upholstered furniture (49)are arguable, as is the benefit of addingflame retardants to wool and other lessflammable fabrics. Also unclear is thebenefit of chemical additives in hospitalgarments for newborn babies, and in in-fants' clothing and sleepwear (48).Another striking example of unneces-

sary flame-retarding treatment is in the

21

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

area of light-weight nylon tents used forbackpacking and mountaineering. Of 119documented injuries and deaths in theUnited States in tent fires during theyears 1971 to 1974 inclusive, none oc-

curred in nylon tents (50). Only 2 of 75fires reported in this study involved ny-

lon tents. In addition to the possible bio-logical hazards of the flame-retardantcompounds, their addition markedly in-creases the weight and cost and de-creases the fabric tear strength. Even un-

treated tents have been known to tear un-

der severe conditions leading to injuryand death of the occupants (51). Thus,treated nylon tents are both heavier andpotentially less safe. Nevertheless, legis-lation to require flame-retarding treat-ment of all tents is in force in California,as well as several other states.Flame retardants added to plastics are

obviously of less concern as environmen-tal hazards than those added to clothing,yet any that are going to be eventually re-

leased into the environment in largeamounts (such as the PBB's) should begiven thorough toxicological testing. Afew do appear to have been tested fairlythoroughly (for example, decabromobi-phenyl oxide).

Responsibility

It is not clear who has the responsibili-ty and authority for the establishment offlame-retardant standards that are safe,both from a fire and a biological point ofview. The Consumer Produ,ct SafetyCommission says that it has the responsi-bility to set performance standards, butnot the authority to require that flame re-

tardants be pretested for carcinogenicityor mutagenicity (52). The responsibilityfor safely meeting these standards is leftto the chemical industry. Many indus-tries do not accept this responsibility forcarrying out cancer tests on large-vol-ume chemicals. Thus, there is a conflictbetween government and industry as towho should be responsible for meeting bi-ological and environmental safety stand-ards. At present, no government agency

has the authority for ensuring long-termsafety of textile additives such as flameretardants, although the toxic substanceslaw might eventually solve this problem.The strict flammability standards vitallyaffect many industries that are caught inthe middle.Few cancer tests in animals have been

carried out with the large number ofchlorinated and brominated chemicals(7) that make up a good part of about100,000,000 pounds of organic flame re-

22

lardants used annually in the UnitedStates. A similar situation existed 20years ago when billions of pounds ofchlorinated and brominated chemicalswere introduced as pesticides and indus-trial chemicals even though animal can-

cer tests had not been performed. Bothsituations are disturbing for several rea-

sons. Organic chemicals containing chlo-rine and bromine (and fluorine) are notused in natural biochemical processes

and have not been normally present inthe diet. A large number of these haloge-nated and industrial chemicals to whichhumans have unwittingly been exposedare proving to be carcinogens in animalsnow that the cancer tests are being done.Many more compounds remain to be test-ed. As the 20- to 30-year lag time forchemical carcinogenesis in humans is al-most over, a steep increase in the humancancer rate from these suspect chemicalsmay soon occur. While waiting for the ef-fects of the large-scale human exposure

to the halogenated carcinogens-poly-chlorinated biphenyls (PCB's), vinylchloride, Strobane-toxaphene, aldrin-dieldrin, DDT, trichloroethylene, dibro-mochloropropane, chloroform, ethylenedibromide, Kepone-mirex, heptachlor-chlordane, pentachloronitrobenzene;and so forth-we might think about theavoidance of a similar situation withflame retardants (53).

References and Notes

1. A. McGuire, director, Northern California BurnCouncil, personal communication; G. F. Crike-lair, Proceedings of the Information Council onFabric Flammability, New York, 1975.

2. E. V. Anderson, Chem. Eng. News, 22 March1976, p. 12.

3. E. M. Pearce and R. Liepins, Environ. HealthPerspect. 11, 59 (1975).

4. Mirex (dechlorane) is a teratogen and carcino-gen and tris-aziridinyl phosphine oxide is a car-cinogen and mutagen (NIOSH) [National Insti-tute of Occupational Safety and Health, Wash-ington, D.C. Registry of Toxic Effects of Chem-ical Substances]; pentachlorobenzene is ateratogen [K. S. Khera and D. C. Villeneuve,Toxicology 5, 117 (1975)]. The highly toxicPBB's are suspected of being all of these and arediscussed in this article.

5. Background information has been obtainedfrom: Flammable Fabrics Act, Fed. Reg. 3, (30December 1975, pp. 59884-59957; Fact Sheet 25,U.S. Consumer Products Safety Commission,Washington, D.C., 1974).

6. J. W. Lyons, The Chemistry and Uses of FireRetardants (Wiley-Interscience, New York,1970).

7. 0. Hutzinger, G. Sundstrom, S. Safe, Chem-osphere 1, 3 (1976); R. B. LeBlanc, TextileIndustries (February 1976), p. 28; close to ahundred different flame retardants, from variousmanufacturers, are listed and available fromRFR Corp. (EnChem Environmental Division),1 Main Street, Hope, Rhode Island 02831.

8. B. A. Worley, (Spring Mills, P.O. Box 70, FortMill, S.C.); Techniforum 2, Arlington, Va.,April 1976.

9. T. J. McGeehan and J. T. Maddock, A Study ofFlame Retardants for Textiles Office of ToxicSubstances, EPA-560/1-76-004 (U.S. Environ-mental Protection Agency, Washington, D.C.,1976).

10. B. J. Eisenberg and E. D. Weil, J. Am. Assoc.Textile Color 6, 257 (1974).

11. M. Friedman, in Flame-Retardant PolymericMaterials, M. Lewin, S. M. Atlas, E. M.

Pearce, Eds. (Plenum, New York, in press). "Ithink that the oven-dry requirements should beeliminated for natural fibers, such as wool andcotton, that ordinarily have a high moisture con-tent. All fibers should be tested at their normalmoisture contents except for special uses whichrequire low humidity. The present standard un-reasonably favors certain synthetics, such aspolyester, that have low normal moisture con-tents." The Consumer Product Safety Commis-sion, however, has set these low humidity levelsbecause of a concern for low humidity environ-ments, such as homes without humidifiers andthe drier regions of the country (R. Adler, per-sonal communication).

12. One of us (B.N.A.) was not successful in tryingto set fire to an old woolen bathrobe.

13. R. W. Morrow, C. S. Hornberger, A. M. Klig-man, H. I. Maibach, Am. Ind. Hyg. Assoc. J.37, 192 (1976).

14. R. H. Harris, of the Environmental DefenseFund, Petition to the Consumer Product SafetyCommission, 24 March 1976.

15. F. A. Daniher, "Toxicology of tris-(2,3-dibromopropyl) phosphate," Proceedings of the 1975and 1976 Symposia on Textile Flammability (LeBlanc Research Corp., 1975); A. F. Kerst, J.Fire Flammability, Fire' Retardant Chem.(Suppl.) 1, 205 (1974). Tris-BP was found to beneither a skin nor an eye irritant. Rats fed dosesin amounts between 0.01 and 0.1 percent of theirfeed for 28 days showed some reversible weightloss and also reversible accumulation of bro-mine in liver, fat, and muscle tissue. In a 30-dayfeeding study with dogs, at doses of 50 mg/kgand 100 mg/kg, an increase in blood brominewas noted. Dermal application, in amounts up to2 g/kg, to rabbits resulted in increased brominelevels.

16. L. E. St. John, Jr., M. E. Eldefrawi, D. J. Lisk,Bull. Environ. Contam. Toxicol. 15, 192 (1976).

17. H. I. Maibach, personal communication.18. M. J. Bartek, J. A. LaBudde, H. I. Maibach, J.

Invest. Dermatol. 58, 114 (1972); D. R. Bickers,A. Kappas, A. P. Alvarez, J. Pharmacol. Exp.Ther. 188, 300 (1974).

19. R. J. Feldmann and H. I. Maibach, J. Invest.Derm. 54, 399 (1970).

20. W. A. Olson, R. T. Habermann, E. K. Weisbur-ger, J. M. Ward, J. H. Weisburger, J. Natl.Cancer Inst. 51, 1993 (1973); M. B. Powers, R.W. Voelker, N. P. Page, E. K. Weisburger, H.F. Kraybill, Toxicol. Appl. Pharmacol. 33, 171(1975).

21. J. McCann, E. Choi, E. Yamasaki, B. N. Ames,Proc. NatI. Acad. Sci. U.S.A. 72, 5135 (1975); J.McCann and B. N. Ames, ibid. 73, 950 (1976);B. N. Ames, J. McCann, E. Yamasaki, Mutat.Res. 31, 347 (1975). The test assays the muta-genicity (ability to damage DNA, the geneticmaterial) of chemicals. Salmonella are used formeasuring DNA damage, combined with tissuehomogenates from rodents (or humans) to pro-vide mammalian metabolism of chemicals. Thetest measures back mutations in a special set ofhistidine-requiring tester strains. A number ofgenetic modifications have been introduced intothe strains to further increase sensitivity. About1 billion bacteria are treated on a single petriplate. Those bacteria that are mutated back tohistidine independence are measured by count-ing the colonies appearing on the plate after anincubation period of 48 hours. Many carcino-gens cause the bacteria to mutate because theyare directly active. We have been able to detectthe large class of carcinogens activated to muta-gens by mammalian metabolism by adding rat(or human) liver homogenates to the test sys-tem. One of the main results to come out of thiswork, in addition to the test system itself, is that90 percent of the carcinogens tested in the sys-tem can be detected as mutagens, whereas veryfew noncarcinogens are mutagenic. Con-sequently we have supported the theory thatalmost all chemical carcinogens cause cancer bysomatic mutation. Mutations damaging normalregulatory mechanisms in a cell could result inuncontrolled growth in that cell and in its descend-ants, thus giving rise to a tumor.

22. B. N. Ames, in Chemical Mutagens: Principlesand Methods for Their Detection, A. Hollaen-der, Ed. (Plenum, New York, 1971), p. 267.

23. M. J. Prival, E. C. McCoy, B. Gutter, H. S.Rosenkranz, Science 195, 76 (1977); H. S. Ro-senkranz, Bull. Environ. Contam. Toxicol. 14, 8(1975).

24. M. Meselson and K. Russell in Origins of Hu-man Cancer, J. D. Watson and H. Hiatt, Eds.(Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y., in press); , personal com-munication; C. Sawyer and B. N. Ames, un-published results.

SCIENCE, VOL. 195

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

25. B. Gutter and H. S. Rosenkranz, in preparation;R. A. Valencia, R. C. Serman, Y. Buchholz, K.Houtchens, in preparation; H. F. Stich, personalcommunication.

26. W. H. Gutemann and D. J. Lisk, Bull. Environ.Contam. Toxicol. 14, 61 (1975). We have beeninformed of several experiments where the im-mersion of pajamas in goldfish tanks caused thedeath of the goldfish. This acute toxicity couldpossibly reflect an anticholinesterase activity oftris-BP.

27. L. V. Afanas'eva and N. S. Evseenko, Hyg.Sanit. 36, 450 (1971).

28. G. Loewengart and B. L. Van Duuren, ThirdInternational Symposium on Detection and Pre-vention of Cancer, New York City, 26 April1976; J. Toxical. Environ. Health, in press.These studies also report a small-scale carcino-genicity study of THPC on mice showing that,when 2 mg were applied topically three times aweek for more than a year to 20 mice, onesquamous carcinoma was observed in thetreated group. Since this sort of cancer has notbeen observed in control mice over the years (B.L. Van Duuren, personal communication), fur-ther work was suggested.

29. G. J. Kallos and R. A. Solomon, Am. Indust.Hyg. Assoc. J. 34, 469 (1973).

30. Textile Flammability Digest (December 1975),(Address: 5454 Post Road, East Greenwich,R.I.).

31. L. J. Carter, Science 192, 240 (1976); J. E.Brody, New York Times, 12 August 1976, p. 20.

32. J. L. Hesse, paper presented at the SecondNational Conference on Complete WateReuse,Palmer House, Chicago, 4 May 1975.

33. T. H. Corbett, A. R. Beaudoin, R. G. Cornell,M. R. Anver, R. Schumacher, J. Endres, M.Szwabowska, Environ. Res. 10, 390 (1975).

34. P. E. Mattsson, A. Norstrom, C. Rappe, J.Chromatogr. 111, 209 (1975).

35. F. B. Clarke III, and J. Ottoson, Fire J. 70, 20(1976); B. M. Halpin, E. P. Radford, R. Fisher,Y. Caplan, ibid. 69, 11 (1975); Bureau of Epide-miology, U. S. Consumer Product Safety Com-mission, "Analysis of injuries associated withspecified wearing apparel (Washington, D.C.,1975).

36. America Burning, Report of the National Com-mission on Fire Prevention and Control (U. S.Government Printing Office, 1973), pp. xi and68.

37. C. Huggett, National Bureau of Standards, per-sonal communication.

38. B. N. Ames and A. Blum, unpublished results.39. Congressional Record, Defeat of Hart Amend-

ment for Consumer Product Safety Commissionauthority over flammable properties of ciga-rettes. An amendment similar to the Hart

Amendment had been proposed earlier in March1975 by R. D. Pittle, Commissioner of the U.S.Consumer Product Safety Commission.

40. D. Byrd and J. Burgess, Fire J. 70, 11 (1976).41. A. McGuire, Director, Northern California Burn

Council, personal communication.42. Bureau of Product Safety, Food and Drug Ad-

ministration, Flammable Fabrics, 4th AnnualReport (U.S. Government Printing Office, Wash-ington, D.C., 1972).

43. Ethyl Corporation, advertisement, Chem. Eng.News (20 January 1975) p. 28.

44. W. I. Langstaff, Proceedings 1976 Symposiumon Textile Flammability (LeBlanc ResearchCorp., East Greenwich, R.I.).

45. L. M. Kushner (U.S. Consumer Safety Commis-sioner) in letter to D. S. Duke, 11 August 1975;see also (48); "Analysis of injuries, associationwith specified wearing apparel items," U.S.Consumer Product Safety Commission (30 May1975). These estimates only include cases whereclothing was the first item ignited.

46. D. Trunkey, personal communication.47. It is instructive to try to estimate the dose of tris-

BP that a child might absorb and to compare itwith the dose ofdibromochloropropane (the car-cinogenic impurity in tris-BP) that was used inthe carcinogenicity experiments (20) in rats. Acumulative dose of about 1600 mg of dibromochloropropane resulted in gastric squamous car-cinomas in 85 percent of the rats and mammaryadenocarcinomas in 50 percent of the femalerats. (Ethylene dibromide is a carcinogen ofroughly similar potency.) We think that a cu-mulative dose of about 1600 mg is likely to beabsorbed by a child during the course of a child-hood wearing polyester pajamas treated withtris-BP. [We estimate about 4000 mg of surfacetris-BP in a 150 g (new pair) of pajamas with asurface area of 8000 cm2: 500 ,ug/cm2 of surfacetris-BP. Absorption of nonpolar chemicalsthrough the skin at appreciable rates appears tobe the rule. In a study of 11 nonpolar chemicalsapplied to human skin (at 4 4Zg/cm2), Feldmanand Maibach (19) have shown the total absorp-tion was from 1.5 to 50 percent with a mean rateof about 0.25 percent per hour. If 1 ,g of tris-BPwas absorbed per night from each square cen-timeter of skin then 1600 mg could be absorbedin about half a year and even if the rate was only10 percent of this, that dose could be absorbedin 5 years. One would expect a decrease in thelevel of surface tris-BP as the pajamas wererepeatedly washed, until a new pair was substi-tuted. Thus, one does not know exactly whatlevel of tris-BP is available for absorptionthrough the skin and at what rate it is absorbed.To get the same effect of a chemical in a rat anda child, however, it is estimated that the child

should receive a 20-fold higher dose [D. G.Hoel, D. W. Gaylor, R. L. Kirschstein, U.Safflotti, M. A. Schneiderman, J. Toxicol. Envi-ron. Health 1, 133 (1975)]. However, tris-BP is aconsiderably stronger mutagen in our test thandibronlochloropropane (about tenfold). [Fromour previous work on mutagenicity it appearsextremely likely that tris-BP is a carcinogen.Although there is a rough correlation betweenmutagenic and carcinogenic potency (24), we donot have enough experience to attempt to pre-dict a carcinogenic potency for tris-BP.] Allthese calculations are based on an amount ofdibromochloropropane sufficient to produce can-cer in almost all of the rats. If our assumptionsare correct, even 1 percent of that dose of tris-BP could lead to an unacceptable incidence ofcancer, in view of the millions of children atrisk. A study (16) attempting to find the tris-BPmetabolite dibromopropanol (or its conjugates)in human urine from two volunteers wearingpajamas treated with tris-BP was negative. How-ever, in view of the risk we are discussing, theanalytical method was insensitive (less than 0.2ppm of the metabolite would not have beendetected), and we do not know what percentageof any absorbed tris-BP would appear as dibromopropanol or its conjugates.

48. TIPS (Textile Industry Product Safety) 4, 1(1976), available at 1750 Pennsylvania Avenue,NW, Washington, D.C. 20006.

49. A standard for all clothing is being formulated atthe Center for Fire Research at the NationalBureau of Standards for the Consumer ProductSafety Commission.

50. R. F. Johnson, Proceedings of the 8th AnnualMeeting of the ICFF, New York City (1974).

51. Accidents in North American Mountaineering(1948-1975), American Alpine Club, New York.

52. A. C. Shakin, of the U.S. Consumer ProductSafety Commission, in a memorandum to S.Butts, 26 March 1976; The National Commis-sion on Fire Prevention and Control has advised(36): "The impact of new materials, systems,and buildings on users and the communityshould be assessed during design stages, wellbefore use."

53. Supported by ERDA grant E(04-3)34 PA156 (toB.N.A.). We thank L. Haroun for assistance inthe mutagenicity assays and 0. Gold for otherhelp with the study; we also thank M. Prival andH. Rosenkranz for information concerning theirunpublished mutagenicity results with flame re-tardants and for help in other aspects of thiswork; B. L. Van Duuren for a sample of tris-BPand other help; and numerous colleagues in gov-ernment and industry for criticism of the manu-script. A.B. was supported in part by NIHgrant 1-F32-CA-05731-0.

Nuclear Waste Disposal:Two Social Criteria

Technical irreversibility and site multiplicity are

suggested as criteria for safe nuclear waste disposal.

Gene I. Rochlin

Del carratere degli abitanti d'Andria meritano di essere ricordate du virtu: lasicurezza in se stessi e la prudenza. Convinti che ogni innovazione nella citth influiscasul disegno del cielo, prima de'ogni decisione calcolano i rischi e i vantaggi per loro e

per l'insieme della cittd e die mondi.-ITALO CALVINO (1).

7 JANUARY 1977

There is a consensus that radio-logically hazardous wastes from the nu-clear fuel cycle should be separated fromthe biosphere to a sufficient degree andfor a long enough time so that they pre-sent no significant risk to life. But thisconsensus does not extend to the defini-tions of "sufficient," "long enough," or"significant risk." Our ability to predictmaterial or geological stability over thecontainment times required for long-lived components has been questioned(2-4). Moreover, the impossibility of pre-dicting socially relevant factors oversuch relatively short periods as a fewhundred years precludes accurate estima-tion of either the probability of an acci-dental or deliberate breach of con-tainment or the effects of such a breachon society (5).

The author is a Research Specialist at the Instituteof Governmental Studies, University of California,Berkeley 94720.

23

on

Nov

embe

r 28

, 200

6 w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from