offered for the first human-powered machine that can hover at a height of 10 feet for a full minute. -].H.

MEDICINE

Fallout New radiation-risk estimates prompt calls for tighter controls �omprehensive reassessment of

the dangers of low-level ionizing radiation roughly trebles esti­

mates of the risk of cancer per unit dose and finds no evidence for a safe dose-one below which the risk of carcinogenesis is insignificant.

The study, known as Biological Effects of Exposure to Low Levels of Ionizing Radiation ( BEIR) V, comes from an impeccable source : the Na­tional Research Council. Based primar­ily on updated information from sur­vivors of the Japanese atomic bomb blasts and from people exposed to ra­diation for medical purposes, it re­places a previous assessment the re­search council made in 1 980. The new study-a fifth in a series-concerns only X-ray and gamma-ray radiation; it is already being cited as an argument for lower limits on the exposure of nuclear-industry workers.

The 1 980 assessment was highly controversial : at the time, two of its authors issued statements disputing its estimates of cancer risk. The reas­sessment, developed with different computer models, supports the view that the 1 980 study underestimated risks. The new study concludes, for example, that if 1 00,000 people re­ceive a single exposure of 10 rems, the radiation would cause about 790 ex­tra cancer deaths (in addition to the 1 8, 3 3 0 otherwise expected). The risk of cancer-at least for most solid tu­mors-appears to be proportional to the dose of radiation. The study also increases the estimated risk of mental retardation for children exposed be­tween eight and 15 weeks after con­ception. The risk of inducing a herita­ble disease in subsequent generations, however, appears to be lower than was previously thought.

Arthur C. Upton of New York Uni­versity Medical Center, chairman of the committee that made the new as­sessment, says it "is not a revolution­ary change " and should not greatly concern most members of the pub­lic. ( The average annual exposure for Americans is less than half a rem, most of it attributable to radon break-

The 4-volume

Handbook of Artificial Intelligence . . . only $4.95 as your introduction to the Library of Computer and Information Sciences You simply agree to buy 3 more books-

at significant savings-within the next 12 months.

Edited by Paul R. Cohen, Edward A. Feigenbaum, and Avron Barr

The most complete reference on this subj e ct, the Handbook of Artificial Inte l l igence offers extensive coverage on topics from neural networks to expert systems and much more. Its convenient article format, bibliography of references, and readable style make it an invaluable resource.

The newly published 4th volume includes exciting articles on the very latest research in AI . Written by 14 leading authorities including Steven Tanimato, Edmund Durfee, Alfred Round, James Allen, and Alan Macworth, these articles reflect AI's expanding app'l ications and cover topics such as:

• Blackboard Systems

• Cooperative Distributed Problem Solving (COPS)

• Knowledge Based Symbolic Simulation

THE LIBRARY OF COMPUTER AND INFORMATION SCIENCES is the oldest, largest book club designed especially for the computer professional. We make it easy to stay on top with the most up-to-date information on systems and database design, software development, artificial intelligence, and more.

MEMBERSHIP BENEFITS. In addition to getting the Handbook of Artificial Inte l ligence for only $4.95 when you join, you keep saving substantially on the books you buy . • Also, you will immediately become eligible to participate in our Bonus Book Plan, with savings of 60% off the publishers' prices . •

At 3-4 week intervals ( 15 times per year) , you wil l receive the Library of Computer and Informatin Sciences News, describing the coming Main Selection and Alternate Selections, together with a dated reply card . • In addition, up to two times a year, you may receive offers of Special Selections which will be made available to a group of select members . • If you want the Main Selection, do nothing, and it will be sent to you automatically . • If you prefer another selection, or no book at all, simply indicate your choice on the card and return it by the date specified . • You will have at least 10 days to decide. If, because of late mail del ivery of the News, you should receive a book you do not want, we guarantee return postage.

r - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - , Library o f Computer and Information Sciences Delran, NJ 08075-9889

Please accept my application for trial membership and send me the four-volume Handbook of Artificial

Intelligence (52130) ' bil l ing me only $4.95, plus shipping and handling. I agree to purchase at least three

additional Selections or Alternates over the next 12 months. Savings range up to 30% and occasionally even

more. My membership is cancelable any time after I buy these three additional books. A shipping and

handling charge is added to all shipments. No-Risk Guarantee: If I am not satisfied-for any reason- I may return the four-volume Handbook of

Artificial Intelligence within 10 days. My membership will be canceled, and I will owe nothing.

Name __________________________________________________________ __

Address, __________________________________________________ ,Apt. __ _

City State Zip __ _

(Books purchased for professional purposes may be a tax-deductible expense. Offer good in Continental U.S.

and Canada only. Prices sl ightly higher in Canada.)

Scientific American 3/90 7 -FX5 L _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

SCIENTIFIC AMERICAN March 1 990 3 5 © 1990 SCIENTIFIC AMERICAN, INC

down products ; moreover, the new estimates of cancer risk should be at least halved when exposure occurs over a long period.} Upton does think, though, that workers who accumulate relatively high doses perhaps " ought to be watched more carefully." He also stresses that the findings are provi­sional , because the youngest victims of the atomic bomb blasts have only now reached middle age ; risk esti­mates may be revised up or down until the last victims die.

Energy Secretary Admiral James D. Watkins has instructed his depart­ment to assess the BEIR V study to determine whether changes in the de­partment 's operations are necessary. The National Council on Radiation Protection and Measurements and the Nuclear Regulatory Commission are also reviewing the new findings. And activist groups, such as the Nuclear Information and Resource Service in Washington, D.C. , intend to use the new estimates in their campaigns. The main target is a proposal by the Nucle­ar Regulatory Commission to deregu­late control of very-low-level radioac­tive waste. - Tim Beardsley

Safe Passage? Study fuels debate over safety of birth centers

The idea has its appeal : having a baby in a homey, caring birth center, away from an impersonal

hospital . But are free-standing birth centers as safe as hospitals ?

The authors o f a large nationwide study published last December in the New England Journal of Medicine say "yes," at least for women who have a low risk of a complicated delivery, as is the case for women normally admit­ted to such centers. (Women who, say, have diabetes or hypertension or are expecting twins are normally referred elsewhere.)

The answer to the safety question hinges largely on whether the centers can cope with unexpected, potential­ly life-threatening problems, such as prolonged fetal distress or profuse postdelivery bleeding by the mother. The centers, where births are generally attended by midwives rather than ob­stetricians, are not equipped for such emergency procedures as cesarean de­liveries and resuscitation of certain newborns, and so the rare individual who gets in trouble may have to be rushed to a hospital for special care.

To assess the safety of birth centers, Judith P. Rooks of the School of Public

Health at Columbia University, Eunice K. M. Ernst of the National Association of Childbearing Centers in Perkiomen­ville, Pa. , and their colleagues followed nearly 1 2 ,000 women admitted in la­bor to roughly half of the known out­of-hospital birth centers in the nation (by some indications, probably the bet­ter-run half ). As might be expected in any population of women in labor, se­rious complications that normally re­quire immediate hospital care arose in a number of the deliveries, almost 8 percent of them. Only about half of the affected women and newborns were moved to hospitals, however ; the others were not transferred, either be­cause delivery was imminent or be­cause the emergency was resolved by the time the baby was born.

In spite of the need for emergency transfer in some instances and the failure to accomplish the move in many of them, all of the mothers in the study survived, and the number of babies who died before or within a few weeks after delivery was low: 1 . 3

in 1 ,000. That rate is similar to rates reported in five studies of women who gave birth in hospitals and were deter­mined retrospectively (by examining their records) to have had a low risk of complications. This similarity in death rates accounts for the investigators ' assertion that free-standing birth cen­ters are as safe as hospitals for low­risk populations.

Some physicians dissent from that verdict. In an editorial accompanying the December report, Ellice Lieberman and Kenneth ]. Ryan of Brigham and Women's Hospital in Boston say they worry that the birth-center study did not include a prospectively followed control group of women (carefully matched by such characteristics as age and health status) who signed into hospitals or hospital-based birth cen­ters. The two acknowledge that fulfill­ing their prescription might be diffi­cult, but they contend that without a directly comparable control group, " it is impossible to make reliable infer­ences about the relative safety of hos­pitals and birth centers." The low in­fant death rate in the birth-center study, they say, might derive from having studied a particularly low-risk population.

" It is impossible to do the perfect study that would convince everyone," Rooks says. "This is a very large, well­designed study; the data are strong."

She also points out that no defin­itive data support the widely held assumption that hospitals are safer. Indeed, she notes, many hospitals can­not perform immediate cesarean de-

3 6 SCIENTIFIC AMERICAN March 1 990

liveries and lack neonatal intensive­care facilities, and many interventions in hospitals can themselves compli­cate deliveries. "There will be babies that die because the mother goes to a birth center ; there will be babies that die because the mother goes to a hos­pital ," she says.

Rooks notes, too, that cesarean de­liveries, which are not risk-free for either mother or baby, are much more commonly performed on women who undergo labor in hospitals. In the birth-center study 4.4 percent of the women had cesarean deliveries, half the rate cited for low-risk populations in the retrospective hospital studies that reported on the procedure.

Warren H. Pearse of the American College of Obstetricians and Gyne­cologists agrees that obstetric prac­tices in certain hospitals need im­provement. Yet he is disturbed by the need to transport women or newborns from birth centers to other facilities in emergencies. The number of prevent­able deaths-or serious health prob­lems-that actually occur because of transport-related delays is undoubt­edly extremely small . Nevertheless, because the dangers are real and the possible consequences catastrophic, many physicians, including Pearse, Lieberman and Ryan, prefer hospital­based birth centers to free-standing ones. -Ricki Rusting

OVERVIEW

Punctuated Equilibrium Darwin survives as the debate evolves

In 1 9 72 two paleontologists, Niles

. Eldredge of the American Museum of Natural History and Stephen

Jay Gould of Harvard University, star­tled-and in some cases dismayed­biologists by suggesting that the view of evolution most of them held was an " insufficient picture." Eldredge and Gould maintained that life was no stately unfolding of gradually chang­ing forms that slowly divided to cre­ate new species. Rather, they believed, species formed relatively quickly as a result of rapid bursts of evolutionary change. Eldredge and Gould thought their colleagues were wrong to blame the rarity of intermediate fossil se­quences on gaps in the fossil record. Instead they proposed that such se­quences were rare because evolution did not happen that way. Eldredge and Gould considered it more likely that

© 1990 SCIENTIFIC AMERICAN, INC