the macho camera system: searching for dark matter

NE of the most pressingastronomy and cosmology

issues in the 1990s is to identify andaccount for the “missing” cosmicmatter. This unsolved problemarises from overwhelming evidencesuggesting that some kind ofubiquitous and invisible mattersurrounds and permeates the bright,observable disks of our own andother spiral galaxies. To put thisextraordinary problem of cosmicbookkeeping into its most basic terms,we simply have no information aboutwhat makes up most of the universe.

The most compelling evidence forthe missing matter comes from twodifferent methods for studying ourgalaxy, the Milky Way. Using onemethod, astronomers can determine

Newton’s laws. In other words, morethan 90% of the mass of the MilkyWay, which must exist because ofits gravitational pull, cannot be seenby the techniques that have beenavailable. The unseen substance isreferred to as “dark matter.” Theconclusion that there is dark matterin the Milky Way is among the mostsecure in modern astronomy andastrophysics.

If the dark matter were made ofnormal stars, dust, or gas clouds, itwould be luminous and readilydetected. Not only is the substancenonluminous, but astronomers alsobelieve that most of the invisiblecomponents do not even lie in theplane of the Milky Way, where moststars are found. The following

the mass of an object (for example,the Sun) by analyzing the planetaryorbits around it. Similarly, theyestimate the mass of the Milky Wayby studying the nearly circularmotion of the visible stars and gasaround the galaxy’s center and bydirectly applying Newton’s laws.The value astronomers obtain in thismanner is the total mass of theMilky Way.

A second way to estimate galacticmass is to measure the total amountof starlight emitted by the galaxy.Astronomers know approximately theratio of starlight to stellar mass; thus,they can estimate the mass of starsthat emit light. The total from thissecond method, called the luminousmass, is only about 10% of the totalmass determined by motion and

The MACHO Camera System:Searching for Dark Matter

The award-winning MACHO camera system—an integrated, two-color, digital imaging system with 32 million pixels—is the leader in

the experimental search for cosmic dark matter. During its first threemonths of operation, this camera system recorded more photometric

measurements than were previously made in the history of astronomy.

7

O

argument leads astronomers to thatconclusion. Theoretical studies showthat flat systems are dynamicallyunstable. The age of our galaxy tellsus that the Milky Way is now stable,and stable systems are nearlyspherical. Therefore, most of thematter in our galaxy is spread outthrough an encompassing sphere,known as the galactic halo. But what,exactly, makes up the dark halo?

In the past, astronomers have comeup with several ideas to explain thenature of dark matter and the fact thatit does not emit or absorb detectableamounts of electromagnetic radiation.A fashionable and widely held notionis that it could consist of hypotheticalelementary particles not yet detected.This idea is called the exotic mattertheory. Axions, massive neutrinos,and weakly interacting massiveparticles (known as WIMPs) havebeen suggested as candidates. Themass of an axion, if it really exists, isproposed to lie between about 10–12

to 10–9 the mass of an electron.WIMPs would weigh more than10,000 times the mass of an electron.

An alternative and much lessexotic idea in many respects is thatthe dark matter could be made ofmaterial in the form of bodies withmasses ranging from that of a largeplanet to a few solar masses. Onecandidate is massive objects likebrown dwarf stars (also calleddegenerate dwarfs), ranging from 10to 80 times the mass of Jupiter; theseobjects are too small to heat up tonuclear-burning temperatures, whichwould make them luminous. Thecompact remnants of burned-outmassive stars, such as neutron stars orblack holes, have also been suggested.

Indeed, the dark matter could evenconsist of macroscopic objects similarto the planet Jupiter itself. WhereasJupiter is visible to us because it isrelatively close to Earth, the distant

Jupiter-like objects making up thedark matter would remain invisibleat the astronomical distances underconsideration (about 30,000 lightyears). The term MACHO, whichstands for massive compact haloobjects, has been adopted as a genericterm for all the proposed dark, massiveobjects in the Milky Way’s galactichalo whether they are like Jupiter ornot. Because temperatures at theircenters are not hot enough to ignitenuclear fusion, MACHOs remaindark and difficult to detect.

Even though MACHOs cannot bedirectly observed through a telescopeor otherwise, it is nevertheless possibleto infer their presence indirectly.This article describes a breakthroughscientific system we built that couplesthe power of large-format digitalimagers with low-cost, high-speedminicomputers. This new system willdefinitively answer the question ofwhether MACHOs make up some orpossibly all of the dark matter, or, asone astronomer puts it, reveal thehiding place of the “halo grail.”

Gravitational Microlensing

In 1986, an astrophysicist atPrinceton University, BohdanPaczynski, suggested a way that it might be possible to identifyMACHOs.1 Through the gravitationalmicrolensing effect, Paczynski noted,objects with masses ranging from onemillionth to one hundred times that ofthe Sun might be detected even if theobjects themselves are nonluminous.

In essence, the gravitational fieldof a MACHO acts as an amplifying(converging) lens. Gravitationalmicrolensing occurs when a MACHOmoves close to the line of sight froman observer on Earth to a backgroundstar outside our galaxy, as shown inFigure 1. The paths of light rays fromthe star (essentially a point lightsource) are bent by the MACHO

(essentially a point mass), and the starappears to brighten as the dark objectmoves across the field. The term“microlensing” is used because thelensing angle is too small to beobserved; in other words, the bendingangle is so small that no imagedistortion is seen. Even though wecannot see a change in the shape ofa star subject to microlensing, weshould be able to see the star gettingbrighter.

The amplification of light by agravitational lens can be significant,but microlensing events are extremelyrare. At any given time, only aboutone star in two million is microlensed.Those that are amplified can have theirbrightness increased by a factor up toseveral times the unamplified flux.

The amplification of apparentbrightness of a background star isalso transient—a most importantfeature. As the MACHO moves outof the line of sight from an observerto the distant star, the star returns toits original intensity. Because allobjects in our galaxy are in motion, ahighly characteristic pulse occurs inthe brightness of the star, providingthe MACHO signature. The durationof a microlensing event depends, inpart, on the mass of the lens. However,duration is also affected by otherparameters, such as the MACHOvelocity transverse to the line of sightand the distance from Earth to thedark object. A rough estimate of thetransverse velocity would be in therange of 200 km/s. For typical modelsof the galactic halo, the time of amicrolensing event

t is given as:

t (days) ≈ 100 √Mmacho/Msun ,

where Mmacho is the mass of theMACHO and Msun is the mass of theSun. For example, if the dark object’smass equaled that of Jupiter, the eventwould be about two days.

MACHO Camera E&TR April 1994

8

Measuring event duration gives usa mass estimate that we could notobtain by any other technique. Beforeany experiments began, gravitationalmicrolensing durations were predictedto range from a few days (for objectsas massive as Jupiter) to weeks (formore massive objects).

Optimal background sources tostudy MACHO microlensing eventsare the stars of the Large and SmallMagellanic Clouds—small galaxiesat the outer edge of the halo. Thephotograph in Figure 2 shows whatthe Large Magellanic Cloud lookslike from the point of view of anobserver on Earth. A perspective ofthe relative location of Earth in theMilky Way, the galactic halo aroundand beyond the Milky Way, and theLarge Magellanic Cloud can begained from this figure. This galaxyis distant enough to exploit thegravitational microlensing effect, andthe line of sight is favorable becauseit traverses much of the halo. Inaddition, the Large Magellanic Cloudis close enough that millions of

individual stars can be seen usingground-based telescopes.

The MACHO Signature vsAstronomical Background

Some classes of stars spontaneouslyvary in brightness. An importantchallenge in the search for MACHOsis to differentiate the microlensingevents of interest from anyastronomical background variations.Fortunately, we can distinguishmicrolensing light curves from thebackground of variable stars becausemicrolensing events that constitutethe MACHO signature are achromatic(i.e., do not change color over time),symmetric in time, and nonrepeating.This kind of brightening is unlike anyknown variable star phenomena. Inaddition, a microlensing light curve isdescribed by only three parameters:• Maximum amplification.• Time of maximum amplification.• Event duration.

Much like understanding MACHOsthemselves, exploring the astronomical

background also opens a scientificgold mine. By doing so, we canuncover important data on thevariability of all celestial objects,including regular and variable stars,explosive outbursts in stars, andquasars.

Scope of the Problem

The search for MACHO events isdaunting. To understand the extent ofthe problem and the many scientificcomponents required to investigateMACHOs, refer once again toFigure 2, which is an artist’s attemptto place the extraterrestrial elementsof our MACHO system in theirgalactic context. Back on Earth, thescope of the scientific problem entails:• Obtaining an extraordinary numberof images each night throughdedicated use of a telescope in theSouthern Hemisphere. (TheMagellanic Clouds are visible onlyfrom the southern sky).• Creating an optical imaging systemwith an exceptionally wide field of

E&TR April 1994 MACHO Camera

9

Figure 1. The paths of light rays from astar outside our galaxy are bent by aMACHO located close to the line of sightbetween the star and an observer on Earth.In this diagram, the amount of bending isgreatly exaggerated to illustrate the concept.The actual angle between the two light raysfrom the star is a tiny fraction of a degree—so small that the star continues to look likea point but becomes temporarily brighteruntil the MACHO moves out of the line ofsight. Because the gravitational field of theMACHO temporarily acts as an amplifyinglens, the phenomenon is called gravitationallensing. The highly characteristic pulse inthe brightness of the star provides theMACHO signature.


10

MACHO

Halo of dark matter

Line of sight

Target star

Milky Way galaxy

(a)

Figure 2. (a) Relative positions of ourplanet in the Milky Way (left), a potentialgravitational lens (a MACHO) in the galactichalo (near center), and a target star in theLarge Magellanic Cloud (right). Theundetected matter of our galaxy, the darkmatter we are investigating, is distributed ina spherical halo surrounding the observabledisk of stars. This halo extends around andbeyond the Milky Way to some unknowndistance. The target star is located in theLarge Magellanic Cloud about 45,000 parsec(pc) from the solar system (1 pc = 3.258 lightyears or 3.086 ¥ 1013 kilometers, so thetarget star is about 150,000 light yearsaway.) Photograph (b) shows what theLarge Magellanic Cloud looks like from theperspective of an observer on Earth. Thisextragalactic cloud is a small galaxy at theouter edge of the Milky Way’s halo. It isdistant enough to exploit the gravitationalmicrolensing effect yet close enough toEarth that individual stars can be seenusing ground-based telescopes.

(b)

view and a large detector to yield animaged area about 100 times largerthan that at most telescopes.• Designing and fabricating veryhigh-quality, large-format digitalimaging cameras by fully exploitingthe newest technologies.• Obtaining, storing, and analyzingmassive amounts of data with newalgorithms.

The search for MACHOs requiresmaking an unprecedented number ofregular photometric measurementson stars for several years. But what,exactly, constitutes enoughmeasurements? Our principaltechnical challenge arose when wedemonstrated that “enough” in thecontext of our search meant exceedingthe total number of photometricmeasurements made in the history of astronomy by two orders ofmagnitude. Because we did not knowexactly what event durations weremost probable, we proposed to use avariety of sampling techniques,ranging from several times an hour toonce every several nights. Existingequipment and instruments could notdo this work.

Finding a Solution

In 1990, we began collaboratingwith groups at the Center for ParticleAstrophysics from the University ofCalifornia campuses at Santa Barbara,Berkeley, and San Diego. An essentialpart of this collaboration involvesresearchers at the Mount Stromloand Siding Spring Observatoriesfrom the Australian NationalUniversity in Canberra. We startedby designing an innovative opticalsystem for the 1.27-m reflectingtelescope at this university. Thefollowing year, we designed andfabricated two charge-coupled device(CCD) cameras, assembled a systemto acquire and process data, anddeveloped data-analysis software.

The result of our efforts is theMACHO camera system, a fullyintegrated, two-color digital cameraand image-processing system. ThisR&D 100 award winner for 1993 (seethe box on p. 13) is the only opticalimaging system that fully exploits thenew generation of large-format CCDimagers. Our search for MACHOs inthe dark matter, which involvesgathering terabytes of data and nearly10 billion photometric measurements,commenced in the autumn of 1992.On completion, distilled results fromthe experiment will definitivelyanswer the question of whetherMACHOs make up the enigmaticdark matter of the Milky Way.

Telescope and Optics

The Magellanic Clouds are onlyvisible from the southern sky. Tomake the number of photometricmeasurements our survey requires, weneeded extended use of a telescope inthe southern hemisphere and anexceptionally wide field of view. Wearranged for four years of dedicateduse of the telescope (Figure 3) at theMount Stromlo and Siding SpringObservatories.

The mirror of the Great MelbourneTelescope is a classic parabolicreflector. A parabolic mirror does nothave good off-axis performance, sowe needed to design a system of


11

Figure 3. The newlyrecommissioned1.27-m telescope at Mount Stromlo,Australia, formerlycalled the GreatMelbourne Telescope.

corrector lenses to reduce the off-axiscoma and astigmatism to acceptablelevels. Coma describes the tendencyof optical systems to make a pointlook like an asymmetrical, pear-shaped spot.

To modify the telescope, weinstalled new drives and encodersand moved the mirror cell back alongthe optical axis about 40 cm so itoperated at the prime focus. We alsofabricated a new optical correctorcell, shown in Figure 4. This cellgives us a useful field of view that is1 deg in diameter, and we imaged anarea about 100 times larger than thatimaged by most detectors. The imagequality throughout the image plane isexceptionally good.

A dichroic beam splitter, shownin Figure 4, allows us to take imagessimultaneously in two colors. The450- to 630-nm spectral channel isidentified as “blue” in the illustrationand the 630- to 760-nm channel as“red.” We use the two color channelsto verify whether candidatemicrolensing events are achromatic(the same in all optical wavelengths).In a genuine microlensing event, thetime variation in brightness is thesame at different wavelengths. Suchcolor-independent brightening wouldgenerally not be expected fromintrinsic stellar variations.

Digital Cameras

Digital imaging systems are nowwidely used for scientific, commercial,and recreational purposes in devicesranging from ordinary video camerasto sophisticated astronomicalcameras. The detector of choice formost of these applications is thecharge-coupled device (CCD). Indeed,CCD detectors have virtuallyeliminated photographic media inmany astrophysical applications,


12

Figure 4. Dual-color, wide-field imaging system for the Mount Stromlo Telescope. Incidentlight from a star is reflected from the telescope’s parabolic mirror (bottom) and then splitinto two beams—red (throughgoing) and blue (reflected)—to allow simultaneous imaging ofa star in two colors. Our new optical corrector cell, shown in the gray area, was designedby E. H. Richardson (formerly from the National Research Council of Canada) and I. Z. Lewis(from LLNL) and was fabricated by OCA Applied Optics of Garden Grove, California.

CCD camera

Doublet lens

Reflected light

Doublet lens

Dewar (cold vacuum chamber)Red

Blue

CCD camera

Doublet lens

Beam splitter

Parabolic mirror

Incident light

Shutter

alcock fig 4 april ETR

remote sensing, and the recording oftime-varying events. One reason forthis shift is that the quantum efficiencyof photographic plates is only a fewpercent, whereas that for CCDs is 40 to 60%. The CCD is also wellmatched to the optical systems incameras and to modern digitalelectronics.

The full power of this technologyhas not been realized because ofseveral limitations. For one, CCDshave not been previously applied tohigh-resolution imaging of largefields because of low yields inmanufacturing uniform, thin CCDs

with large areas. In addition, theelectronics, computer hardware, andsoftware required for CCDs with verylarge numbers of pixels have not keptpace with other advances. Weaddressed and solved both problems.

Large-Format CCD MosaicsWorking with our UC collaborators,

we built the two largest CCD camerasin the astronomy world. Each CCDcamera contains a 2 ¥ 2 mosaic offour 2048- ¥ 2048-pixel CCDimagers (Figure 5) so that each arraycontains a total of 16 million pixels.(For comparison, a conventional

television tube has only 512 ¥ 512or about 260,000 pixels.) The newCCDs were designed by John Gearyof the Smithsonian AstrophysicalObservatory and fabricated by theLoral Aerospace foundry at NewportBeach, California.

Each 15- ¥ 15-µm pixel in oursystem corresponds to 0.63 ¥ 0.63arcsec on the sky, and each completecamera image covers half a squaredegree on the sky. For perspective,the Magellanic Clouds cover tens of square degrees. The CCDs areoperated at cryogenic temperatures(165 K) to ensure low-noise


13

R&D 100 Award for the MACHO System

Each year, R&D Magazine recognizes the 100 mostsignificant new products and technological innovations.At ceremonies held in Chicago on September 9, 1993,the new astronomical camera system built by LLNLand University of California researchers won an R&D100 award.

As befits any project designed to follow more thanten million stars over several years, the number ofcontributors to the overall effort is large. LLNLscientists working on the project, led by physicistCharles Alcock (pictured at right with LaboratoryDirector John Nuckolls), are Robyn Allsman, TimothyAxelrod, David Bennett, Kem Cook, Rob Hills, andHye-Sook Park. University of California researchersare Kim Griest, Stuart Marshall, Saul Permutter, MarkPratt, Will Sutherland, and Christopher Stubbs. TheMount Stromlo researchers are Simon Chan, KennethFreeman, Bruce Peterson, Peter Quinn, and AlexRodgers.

The MACHO camera system was recognized, inpart, because of the considerable scientific importanceof its current application—a definitive search formassive dark objects. Equally important, the instrumentis a model for many future applications requiring rapidimage-taking and processing of digital data. Thesystem can be used for any other project in astronomy,astrophysics, and remote sensing that involves large-

area surveys and the immediate processing ofinformation. For example, Earth-crossing asteroidshave the potential for catastrophic impacts with ourplanet. The early detection of such bodies might allowenough time to deflect them into a safe orbit. We arepreparing a new camera system for this purpose.

operation. It is these arrays that arenow searching for the dark matter ofthe universe.

Most CCDs have control andsignal circuitry spaced around theirperiphery, which creates a good dealof dead space. Our 2 ¥ 2 arrays, onthe other hand, have no circuitry onthe interior edges so that they can bebutted against each other. Becauseour square mosaics have about 50%less dead space than other CCDarrays, they provide far more area forimaging. Our mounting scheme alsoallows individual CCDs within amosaic to be replaced, if needed.

In our packaging scheme, we endup with a small gap of about 600 µmbetween adjacent chips. This gapcorresponds to about 40 pixels. Thegap is not an issue in our work becausewe never attempt to analyze an objectthat spans across two CCDs.

ElectronicsThe electronics we constructed

for our new cameras must not only

control them but must also handlethe exceptionally large volume ofdata these cameras produce. Ourelectronics combine commercial and custom-made components.

We paid particular attention toisolating the instrumentation for thetwo imaging packages. Furthermore,since all outputs run synchronously,any output can contaminate otherchannels through a variety ofmechanisms. We have gone toconsiderable lengths to controlcrosstalk. All communicationbetween on-telescope cameraelectronics and the downstreamcomputer system takes place overoptically isolated data links. Theamplified signals travel about 10 mto a signal-processing crate.

Data Analysis

Each CCD has two readoutamplifiers. The 16 analog outputchannels (2 mosaics ¥ 2 outputs perchip ¥ 4 chips per mosaic) are

digitized at the telescope and read outsimultaneously. We use all theavailable amplifiers to minimize thesystem’s readout time. It takes justover one minute to read out all thedata for a frame. The digital data arefed via optical fiber cables to thecontrol room located in thetelescope’s dome.

The MACHO system routinelygenerates one 5-min exposure of theMagellanic Clouds approximatelyevery 6 min, reading out 32 millionpixels. Each image consists of 64 Mbytes of data. The addition of important camera diagnosticinformation adds 20% to this volumeof data. Our goal is to attain real-timeprocessing that reduces each imageto photometry while the next imageis being exposed.

A multiprocessor Solbournecomputer is the primary dataprocessing system. This mini-computer controls the entire system,issues commands to the telescope,controls the camera system, manages


14

Figure 5. (a) View of one MACHO camera with the faceplateremoved to show the set of four CCD imagers. The array of CCDsis 6 ¥ 6 cm. Our mounting scheme allows individual CCDs within a2 ¥ 2 mosaic to be replaced, if needed (b).

CCD

3 cm

Precision-machined Kovar

Printed circuit-board wire bonded to CCD

Alcock Figure 5b

(a) (b)

the data flow, and archivesinformation. The software wedesigned effectively manages theseongoing operations.

Images read out through the 16-channel system are written intodual-ported memory in the data-acquisition system using a customdescrambler board. Descramblingensures that physically adjacentpixels appear in contiguous memorylocations. The files are also writtento 8-mm tape for archiving.

The disk-resident data are reducedwith a code known as Sodophot (apoint-spread, function-fitting routinefor photometric analysis). First, oneimage of each field obtained in goodseeing conditions is used to produce atemplate catalog of star positions andmagnitudes. Bright stars in routineobservation are matched with thiscatalog, and the catalog is transformedto the coordinate system of theobservation. We then photometricallyfit each template star in descendingorder of brightness.

When we find a star that variessignificantly, it and its neighborsundergo a second iteration of fitting.The output consists of magnitudesand errors for the two color channels(red and blue) plus six additional

useful parameters, including the chi-square statistic (which tests forsignificant differences betweenseveral samples). We use thisinformation to flag questionablemeasurements that can arise fromcosmic rays hitting the imagers, badpixels, and other fake events. Thedual-camera arrangement gives us areassuring level of redundancy. Forexample, it is extremely unlikely thatcosmic rays would cause errors onthe same star at the same time in bothimagers. To minimize anomalousresults, we insist that a lensing signalappear with the same magnificationin both colors.

An automatic time-series analysisof photometric data applies a set offilters to search for microlensingcandidates and variable stars, whichare abundant. The final selectionprocess for each microlensingcandidate is automatic. It incorporatescriteria we established empiricallyusing Monte Carlo addition of fakeevents into real light curves. Thesecriteria include:• Signal-to-noise ratio.• Quality of fit.• Wavelength independence of thelight curve.• Color of the star.

Results to Date and Conclusions

The MACHO project has broughtto operation the most powerful surveysystem in the history of astronomy.On a clear night, photometricmeasurements on 20 to 30 millionstars are possible, and we canmeasure twice in one night most ofthe stars in the Large MagellanicCloud. More than 500,000 stars canbe recorded in a single image of thedense, central region of this nearestgalactic neighbor. In the first threemonths of operation, we recordedmore photometric measurementsthan had previously been made in thehistory of astronomy. The survey willoperate for a minimum of four years.

By late 1993, the American andAustralian team, led by LLNLphysicist Charles Alcock, hadmonitored about 3.3 million stars for a year. Each star was observedhundreds of times. In one remarkableevent, shown in Figure 6, a starappeared to increase to 7.2 times itsnormal brightness and then returnedto normal over a 34-day period.2

This event can be plotted in otherways. Figures 7a and 7b show twocurves, one for the amplification ofred light and one for the amplification


15

(a) Baseline signal, 1/24/93 (b) Peak amplification, 3/10/93 (c) Baseline signal, 4/24/93

Figure 6. Photos of our first candidate microlensing event. The peak amplification on March 10, 1993, is more than sevenfold that of thebaseline amplification recorded about a month earlier and later.

of blue light over time. In fact, sincethese two curves are essentially thesame, we can speak of “the lightcurve for this event” as if it were asingle plot. Many features of thiscurve are consistent with gravitationalmicrolensing.

The light curve is achromaticwithin the error of measurement, aswould be expected from a genuineevent. That is, as shown in Figure 7c,the brightness does not differ at thered and blue wavelengths. The curvehas the expected symmetrical shape,again conforming to our expectationof a true microlensing event. Inaddition, the event was nonrepeatingin that this ordinary star had given noprior indication of pulsing or otheractivity that could account for theincreased brightness. Recall that thesethree features are what define theMACHO signature. The finding,announced simultaneously byastrophysicists from the Laboratoryand UC, could be the first evidence ofdark matter in the form of MACHOs.

Recall from our earlier discussionthat the duration of a microlensingevent depends, in part, on the massof the dark object serving as the lens.This means that the mass of ourcandidate object can be estimated(or bounded) if this is a genuinemicrolensing event. The estimate isnecessarily a rough one because wedo not know the relative velocitytransverse to the line of sight or thedistance to the lens. Nevertheless,by using a model of the mass andvelocity distributions of halo darkmatter, we find that the most likelymass for our candidate MACHO isabout 0.12 that of the Sun. (Massesof 0.03 and 0.5 that of the Sun areroughly half as likely.) This massrange includes brown dwarfs andmain-sequence stars. The mass rangeis too small to be consistent with thatof neutron stars or black holes and


16

Figure 7. Observed light curve for the candidate microlensing event recorded by theMACHO camera system during February and March 1993. Amplification plotted on theordinate is the star’s flux divided by the median observed flux in the blue (a) and red (b)wavelengths. The smooth curves are the best-fit theoretical microlensing models. The peakis 7.2 times the baseline flux. The ratio of red to blue flux (c) shows that this candidatemicrolensing event is achromatic. (The flatness of this curve means the brightness is thesame at different wavelengths.) We would not expect this kind of result from an intrinsicstellar variation, but we do expect it for a MACHO.

2

0

2/6/93 3/10/93 4/17/93

4

6

8

Am

plif

icat

ion

, blu

e

2

0

4

6

8

Am

plif

icat

ion

, red

0.5

0

1.0

1.5

2.0

Rat

io o

f re

d t

o b

lue

flu

x

(a)

(b)

(c)

Alcock Figure 7

too large to be consistent with anobject as small as the planet Jupiter.It is simply too soon to say, on thebasis of this single event, whetherother candidates will be in the Jupitermass range.

It is remotely possible that thestellar brightening we observed couldbe caused by some unknown sourceof intrinsic stellar variation. Onecrucial test for the hypothesis thatwe are seeing true gravitationalmicrolensing by MACHOs is thedetection of other candidates. In early1994, we found two more potentialmicrolensing events after an initialexamination of new data. These newcandidates for microlensing are notso dramatic as the first: one starappeared to increase to 1.5 times itsnormal brightness, and anotherincreased to about 2 times its normalbrightness. Further analysis of thedata will confirm whether or notthese two candidates are truemicrolensing events.

In this regard, it is important tonote that similar observations on twoother dim stars using a differenttechnique were made by the EROScollaboration of French researchers.3In addition, the Optical GravitationalLens Experiment group, which is ajoint venture between American and Polish researchers headed byPaczynski, recently reported a singlemicrolensing event using a telescopelocated in Chile.4

Whereas observing a fewmicrolensing events is excitingbecause it demonstrates that ourexperiment is working, far-rangingconclusions are not warranted atpresent. We have only analyzed about50% of our first year’s data, and ourobservations will continue until 1996.We cannot yet say how plentifulMACHOs may be or how much ofthe galaxy’s dark matter consists ofthese objects. Additional events(say, more than ten) will give us asufficient statistical sample to allowfurther tests. For example, theorypredicts that larger magnificationswill be rarer than smaller ones.Although our results to date do notrepresent final answers, our findingsare consistent with the idea thatmost—or all—of the dark matter inthe universe can be considerednormal rather than exotic. Moreover,our experiment is likely to be thefirst of the many dark-matterinvestigations now taking placearound the world to report definitiveresults. We were successful becauseour system gathers data at a muchgreater rate, and we can processthose data at a faster rate than otherinvestigators.

The implications associated withverifying the existence and nature ofdark matter are truly cosmic. Theconfirmation of compact objects willaffect theories of galaxy formationand evolution. If we can account forsome, but not all, of the dark matter

in terms of MACHOs, thenarguments for at least two differentkinds of dark matter will bestrengthened.

Key Words: Center for Particle Astrophysics;charge-coupled device (CCD) camera; computercode—Sodophot; dark matter; exotic matter theory;gravitational microlensing; Large MagellanicCloud; massive compact halo objects (MACHOs);Milky Way; Mount Stromlo and Siding SpringObservatories; Optical Gravitational LensExperiment; R&D 100 award.

References1. J. B. Paczynski, “Gravitational Microlensing in

the Galactic Halo,” Astrophys. J. 304, 1 (1986).2. C. Alcock, et al., “Possible Gravitational

Microlensing of a Star in the Large MagellanicCloud,” Nature 365, 621 (1993).

3. E. Aubourg, et al., “Evidence for GravitationalMicrolensing by Dark Objects in the GalacticHalo,” Nature 365, 623 (1993).

4. A. Udalski, M. Szymanski, J. Kaluzny, M.Kubiak, W. Krzeminski, M. Mateo, G. Preston,and B. Paczynski, “The Optical GravitationalLensing Experiment. Discovery of the FirstCandidate Microlensing Event in the Directionof the Galactic Bulge,” Acta Astonomica 43,289 (1993).


17

For furtherinformationcontact Charles Alcock(510) 423-0666.

the macho camera system: searching for dark matter

Documents