16 S C I E N T I F I C A M E R I C A N N O V E M B E R 2 0 0 1

newsSCAN

NAS

A/C

XC/S

AO

Several methods have sought todetermine the stability of alpha, a fundamental constant:

� The abundance of light elementssuch as helium and lithium in theuniverse suggests alpha wasunchanged to within 2 percent afew minutes after the big bang,when such elements formed.

� Atomic clocks in 1994 showedthat alpha was constant to 1.4parts in 100 trillion over 140days, which extrapolates to fourparts in 100,000 over a billionyears. An “atomic fountain”experiment has improved theprecision by a factor of five.

� In Oklo, Gabon, 1.8 billion yearsago, a natural nuclear reactorformed in a deposit of uranium.The isotopes remaining implythat alpha was the same then as it is today to within afew parts in 10 million—about100 times more precise thancurrent astrophysicalmeasurements.

CONSTANT STRUGGLE

I f the result holds up, it will be one ofthe biggest discoveries in decades: bil-lions of years ago the fundamental con-

stant of nature that governs electromagne-tism was slightly weakerthan it is today. That wouldseem to fly in the face ofone of the most cherishedprinciples in all of science,namely that the laws gov-erning the universe are thesame everywhere and at alltimes. The evidence comesfrom studies of light fromdistant quasars carried outby an international groupled by John K. Webb of theUniversity of New South

Wales in Australia beginning four yearsago. The results have remained consistenteven as the group has gathered more dataand refined its methods of analysis.

Still, most astrophysicists remain skepti-cal. “My gut feeling is that some other ex-planation will be discovered for this obser-vation,” says Robert J. Scherrer of Ohio StateUniversity. “Of course, I’d love to be provedwrong; that would be very exciting.”

Webb and his co-workers are also cau-tious. “Three independent samples of data,including 140 quasar absorption systems,give the same [amount of] variation” in theconstant, explains theorist Victor V. Flam-baum of the New South Wales group. “How-ever, as with any first observation, there isroom for doubts. Serious conclusions shouldbe made later, after independent checks ofour current results.”

The constant in question is the fine struc-ture constant, or alpha, for the Greek letterused by physicists to represent it in equations.The data indicate that between eight billionand 11 billion years ago, alpha was weaker byabout one part in 100,000. Among other ef-fects, electrons in atoms would have beenslightly more loosely bound to nuclei thanthey are today, increasing the characteristicwavelengths of light emitted and absorbed byatoms. Astronomers can study such ancientlight by looking at distant quasars. In partic-

ular, they focus on secondary effects that shiftindividual wavelengths of an atom by slight-ly different amounts; very precise measure-ments of the separation between wavelengthsprovide a measure of alpha’s change.

Astronomers have been conducting suchstudies since the mid-1960s and have seen noevidence of a change in alpha to the precisionachieved. Webb and his co-workers, howev-er, developed a new technique of looking atwavelengths from many chemical elements atonce to improve the accuracy. Extracting thetiny change in alpha from that data is a com-plicated process, combining informationfrom laboratory studies and intricate com-puter modeling of atomic quantum states.Many spurious phenomena and measure-ment errors could mimic the wavelengthshifts. Webb and his colleagues believe theyhave verified that none of these effects couldbe producing their results, but other re-searchers are unconvinced.

The question can best be resolved by fur-ther experimental work using different meth-ods, but few alternatives are known. Christo-pher L. Carilli of the National Radio Astron-omy Observatory in Socorro, N.M., and hisco-workers have studied microwave absorp-tion by hydrogen, but they have done so onlyfor redshifts corresponding to times more re-cent than six billion years ago. Their data andWebb’s agree that no detectable change in al-pha has occurred over that interval. Carillihopes to find suitable hydrogen clouds atlarge redshifts for a direct comparison at ear-lier times. “A major technical advance,” hesays, “is the new Green Bank Telescope inWest Virginia,” which is the largest steerableradio telescope in the world. It began opera-tions in August.

Studies of irregularities in the cosmic mi-crowave background correspond to the timea mere 300,000 years after the big bang, pro-viding a measure of alpha almost 14 billionyears ago. Using the most recent data, PedroP. Avelino of the University of Porto in Por-tugal and his colleagues have found no evi-dence of a change in alpha, to an accuracy ofabout 10 percent. Data in the next few yearsfrom the recently launched MAP satellite may

Plus Ça ChangeHAS A FUNDAMENTAL CONSTANT VARIED OVER THE AEONS? BY GRAHAM P. COLLINS

PH

YSIC

S

ANCIENT LIGHT from quasars mayharbor clues of altered physics.

Copyright 2001 Scientific American, Inc.

18 S C I E N T I F I C A M E R I C A N N O V E M B E R 2 0 0 1

newsSCAN

EII

CH

IRO

KO

KU

BO

Na

tion

al

Astr

onom

ica

l O

bse

rva

tory

Of

Jap

an

AN

D H

ITO

SHI

MIU

RA

Mu

sash

ino

Art

Un

iver

sity

� Why didn’t debris from theimpact just fall back to Earth?To reach orbit, a rocket has tofire its engines at least twice:first to lift off, then to circularizeits trajectory. Rockets thatforget the second burn areballistic missiles. Researchersthink that the lopsided gravity ofthe mutilated Earth andpressure gradients in thevaporized debris did the trick.

� Why is the moon’s orbit tilted?The impact debris should havesettled into a Saturn-like diskaligned with Earth’s equator.Last year researchers arguedthat gravitational interactionswith residual debris quicklywrenched the nascent moon outof that plane; much later on, thesun’s gravity reoriented theorbit yet again.

� Why is there only one moon? Asufficiently large debris diskcould have given birth to a familyof moons, rather like Jupiter’s.But recent work found that thesiblings would have merged orbeen ejected. Jupiter’s moonsescaped that fate because thetidal torques that cause orbitsto move around are weaker inthe Jovian system.

SOLVING MYSTERIES:MOON FORMATION

I f you ever find yourself at a cocktail partyof astrophysicists and don’t know whatto say, try this: “But what about the an-

gular momentum?” No matter what thetopic of conversation, you’ll be guaranteedto sound erudite. Nearly every field of as-tronomy, from galaxy formation to star for-mation, has an “angular momentum prob-lem.” Nothing in the cosmos ever seems tospin or orbit at the rate it should.

The moon is no exception. It is the fly-wheel to end all flywheels; if its orbital angu-lar momentum were transferred to Earth’s ax-ial rotation, our planet would come close tospinning apart. No other planetary sidekickwields such power, except for Pluto’s crypto-moon, Charon. The moon’s prodigious an-gular momentum is one reason that planetaryscientists believe that it formed when anoth-er planet—no piddling asteroid but an entireMars-size world—struck the proto-Earth.

Unfortunately, researchers have had trou-ble getting the giant-impact model to workwithout the contrivances that scuttled earliertheories. “Putting enough material into orbitto form the moon seemed to require a rathernarrow set of impact conditions,” says RobinM. Canup of the Southwest Research Institutein Boulder, Colo. But a new study by her andErik Asphaug of the University of Californiaat Santa Cruz may have broken the logjam.

Although the giant-impact model becamedominant in the mid-1980s, fleshing it out hasbeen a gradual process. Simulations have at-tempted to reconcile the angular momentumwith three other basic facts: Earth’s mass, themoon’s mass and the moon’s iron content.These four quantities depend on three basic at-tributes of the collision: the impactor’s mass,the proto-Earth’s mass and the impact angle.

Four facts and three parameters is a recipefor contradiction. To explain the moon’s low

iron content, you need to avoid a grazing col-lision (corresponding to a large impact angle),lest too much of the impactor’s iron spill intoorbit. Then, to explain the angular momen-tum, you need to compensate for the small-ish angle with a hefty impactor. Then, to ex-plain the moon’s mass, you need to adjust theproto-Earth’s mass. In the end, you might findthat the total mass is incorrect.

In 1997 Alastair G. W. Cameron, one ofthe fathers of the giant-impact theory, now atthe University of Arizona, arrived at a totalmass that was a third too low. He suggestedthat subsequent asteroid impacts made up thedifference. But few liked the idea, as the as-teroids would have added extra iron.

Canup and Asphaug argue that the faultlies not in the stars but in our simulations. Thecalculations rely on a technique known assmoothed-particle hydrodynamics, whichsubdivides the bodies and applies the laws ofphysics to each piece. Early runs tracked3,000 pieces—leaving the iron core of themoon to be represented by just a single piece.Even the slightest computational imprecisioncould vastly overstate the iron content, inwhich case the computer compensated by re-ducing the impact angle. The result was a biastoward heavy impactors and light proto-Earths. Because Canup and Asphaug use30,000 particles, they get by with a muchsmaller impactor. Everything—mass, iron,momentum—clicks into place.

Considering all the twists and turns in lu-nar science, nobody claims that the models arecomplete just yet. Cameron says Canup andAsphaug’s model doesn’t track events for along enough time, and moon modeler ShigeruIda of the Tokyo Institute of Technology saysthat further increases in resolution could causemore upheaval. Still, it may not be long beforeyou’ll need a different cocktail-party question.

Earth-Shattering TheoryFINALLY, THE DETAILS FOR FORMING THE MOON WORK OUT BY GEORGE MUSSER

ASTR

ON

OM

Y

tighten the limit to as little as 0.1 percent.One is left with a puzzle of no discernible

variation in the most recent epoch, none inthe earliest (when the largest change might beexpected), but the tiny variation of one in

100,000 between eight billion and 11 billionyears ago. “Even if their result doesn’t holdup,” Carilli says, “they certainly have spurredinterest in this field and have motivated manyexperimentalists to expand their efforts.”

WITHIN THE DEBRIS DISK thrownup by a giant impact, the moonbegan to coalesce after a few days.

Copyright 2001 Scientific American, Inc.