ice age lamps

The controlled use of Þre, Þrstachieved at least half a millionyears ago, is one of the great in-

novations in human culture. Althougharchaeologists and anthropologists gen-erally emphasize the importance of Þrefor cooking, warmth and protectionfrom predators, the light accompanyingÞre was also a precious resource, onethat made it possible to extend humanactivity to times and places that are nat-urally dark. The invention of stone, fat-burning lamps, which happened in IceAge Europe nearly 40,000 years ago, of-fered the Þrst eÝective, portable meansof exploiting this aspect of Þre. The ap-pearance of lamps broadly coincideswith a number of other extraordinarycultural changes, including the emer-gence of art, personal adornment andcomplex weapons systems.

Many scholars have hypothesizedabout how Ice Age lamps functionedand were used, but nobody had everundertaken a systematic study of them.

One of us (de Beaune) therefore set outto examine these lamps in detail and toclassify them by type. In conjunctionwith that project, we built working rep-licas of stone lamps in order to analyzetheir eÝectiveness as light sources andto learn about their design, fabricationand use. The results of this investigationprovide a provocative insight into thetechnology and behavior of some ofthe earliest modern humans in Europe.

The Þrst object explicitly identiÞed asan Ice Age lamp was discovered in 1902,the year researchers authenticated thewall art in the cave at La Mouthe,France. Archaeologists had presumedthat the creation of paintings and en-gravings hundreds of meters under-ground must have required an artiÞciallight source. In the course of exploringLa Mouthe, they uncovered compellingsupport of that notion: a carefully fab-ricated and heavily burned sandstonelamp bearing the engraved image of anibex on its underside.

Since then, hundreds of more orless hollowed-out objects havebeen excavated and rather indis-

criminately lumped into the category oflamps. The initial research goals were tosift through the potpourri, establish cri-teria for identifying lamps and examinevariation within this category of objects.A search of the literature and of muse-um collections turned up 547 artifactsthat had been listed as possible lamps.The Þrst hurdle was to distinguishlamps from other similarly shaped im-plements, such as grinding stones. It

quickly became obvious that the sizeand shape of an object are insuÛcientas deÞning criteria. For example, lampsneed not have a bowl-shaped depres-sion; many perfectly ßat slabs showclear traces of localized burning, whichin these and other instances provide

108 SCIENTIFIC AMERICAN March 1993

SOPHIE A. DE BEAUNE and RANDALLWHITE share a fascination with the cul-ture and technology of Ice Age humans.De Beaune is a member of the Labora-tory of Prehistoric Ethnology at the Na-tional Center for ScientiÞc Research(CNRS) in Paris. She also teaches biolog-ical and cultural anthropology to highschool students in France. Her currentresearch focuses on nonßint artifactsfrom the Paleolithic era; she has partic-ipated in and directed several excava-tions in southwest France. White is anassociate professor of anthropology atNew York University. He specializes inUpper Paleolithic art and technology andis currently preparing a monograph onthe earliest forms of personal adorn-ment among European cultures. Whiteco-edited French Upper Paleolithic Col-lections in the Logan Museum of Anthro-pology, which brought to light importantlate Ice Age art and artifacts in an im-portant U.S. repository. De Beaune con-tributed most of the new research forthis article, which White has helped placein a broader scientiÞc context.

Ice Age LampsThe invention of fat-burning lamps toward the end

of the Ice Age helped to transform European culture.It coincided with several other major technological advances

by Sophie A. de Beaune and Randall White

CARVED-HANDLE LAMP, 17,500 yearsold, represents one of the most elabo-rate designs used by Paleolithic humansin France. The lamp consists of abradedred sandstone. The bowl holds a fattyfuel and wick; the long handle keepsthe end of the lamp cool. Engravings onthe handle resemble patterns paintedon the walls in the cave at Lascaux,where the lamp was found.

Copyright 1993 Scientific American, Inc.

the only incontrovertible evidence thatan object served as a lamp.

We judged that 245 of the 547 puta-tive lamps clearly served other purpos-es (mortars, ocher receptacles and soon). The remaining 302 objects were of uncertain status as lamps. We thendivided that sample (285 of which havea well-known site of origin) into twocategories. We considered 169 of theitems to be certain, probable or possi-ble lamps. The other 133 we classiÞedas doubtful or unavailable for study.Markings left by the burning of fueland wick tend to disappear over time,so the oldest lamps were the most like-ly to fall into the dubious category. Thelamps that we consider here all datefrom the Upper Paleolithic era, between40,000 and 11,000 years ago.

The 285 lamps of known origin comefrom 105 diÝerent archaeological sites,mainly in southwest France. The Aqui-

taine basin has yielded 60 percent of thelamps, the Pyrenean region 15 percent.Considerably fewer lamps have been re-covered from other parts of France, andlamps found outside FranceÑin Spain,Germany and CzechoslovakiaÑare ex-ceedingly rare. Although this patternmay be explained in part by the histori-cally greater intensity of research andthe greater number of sites in south-west France, it seems that lamp-pro-ducing cultures were in fact restrictedto a particular European region.

The vast majority of the known stonelamps consist of limestone or sand-stone, both of which are fairly abun-dant. Limestone has the advantage of often occurring naturally in slab-like shapes that require little alteration.Moreover, limestone conducts heatpoorly, so lamps of this material do notget hot enough to burn the userÕs Þn-gers. Sandstone is a much better heat

conductor, so simple sandstone lampsquickly become too hot to hold afterthey are lit. Paleolithic people solvedthis problem by carving handles intomost sandstone lamps. Perhaps part ofthe appeal of sandstone lay in its at-tractive red color and smooth texture.

Our experiments suggest that the sizeand shape of the bowl are the primaryfactors that control how well a stonelamp functions. Setting bowl shape asour primary criterion, we divided the302 Upper Paleolithic lamps into threemain types: open-circuit lamps, closed-circuit bowl lamps and closed-circuitlamps with carved handles.

Open-circuit lamps are the simplestkind. They consist of either small, ßator slightly concave slabs or of largerslabs having natural cavities open toone side to allow excess fuel to drainaway as the fat melts; the largest onesare roughly 20 centimeters across. Be-

SCIENTIFIC AMERICAN March 1993 109Copyright 1993 Scientific American, Inc.


cause open-circuit lamps show no no-ticeable signs of carving or shaping,large numbers of them may have goneunrecognized in premodern excava-tions. As a result, open-circuit lampsprobably are underrepresented in thecurrent sample.

Any slab of rock will work as an open-

circuit lamp, so fashioning one requiresextremely little eÝort. The trade-oÝ isthat these kinds of lamps inevitablywaste a lot of fuel. Open-circuit lampsmay be best interpreted as makeshiftor expedient devices, easily made andfreely discarded. Studies of the modernInuit show that human groups, even

those capable of building large, elabo-rate lamps, occasionally burn a piece offat on a stone slab when no alternativelies readily at hand.

Closed-circuit bowl lamps are themost common variety. They arefound in all regions, in all peri-

ods and in all types of sites wherelamps have been recovered. Closed-cir-cuit bowl lamps have shallow, circularor oval depressions designed to retainthe melted fuel. The recovered lamps ofthis kind range from crude to elaborate.Some bowl lamps are entirely natural,some have a slightly retouched bowland others are completely fabricated.The exterior part of the lamp also maybe natural, partly retouched or entirelysculpted. These lamps consist of oval orcircular pieces of limestone that are usu-ally the size of a Þst or slightly larger.The bowl has sloping sides capable ofretaining liquid when the lamp is placedon a horizontal surface. A typical bowlmeasures a few centimeters across butonly 15 to 20 millimeters deep. Thelargest bowls can hold about 10 cubiccentimeters of liquid.

Ice Age closed-circuit lamps resemblethose employed by certain Inuit peo-plesÑsuch as the Caribou, Netsilik andAleutÑwho had access to wood forfuel and were therefore not dependenton lamps for heat. Inuit living north ofthe treeline, where wood was scarce, de-signed large lamps from slabs of soap-stone that were up to a meter across.Those giant lamps (perhaps more cor-rectly thought of as stoves) served manyof the same functions as hearths else-where, including drying clothes, cook-ing and heating. There may be directrelations between the quality and abun-dance of locally available wood for fuel,the presence of Þreplaces and the formof lamps at a site.

The most intricate lamps are thosewe classiÞed as closed-circuit lampswith carved handles. The 30 such lampsin our sample are shaped, smoothedand Þnely Þnished entirely by abrasion.Each has a carved handle; 11 of themare decorated with engravings. Theselamps appear in the archaeological rec-ord somewhat later than the others.The Þrst carved-handle lamps show up in either the Solutrean (22,000 to18,000 years ago) or Lower Magdaleni-an (18,000 to 15,000 years ago) cultures.They are particularly abundant in theMiddle and Upper Magdalenian (15,000to 11,000 years ago). Most carved-han-dle lamps are found in the Dordogneregion of France. They are most abun-dant in rock-shelter sites but are alsofound in caves and open-air camps.

The elegant design, rarity and limited

ENGRAVED DECORATIONS often appear on the sides or bottoms of closed-circuitlamps. This carved-handle lamp, which features the incised image of an ibex, wasfound at La Mouthe in 1902. It was the Þrst object explicitly identiÞed as a lamp.

LAMP DESIGNS fall into three main categories. Open-circuit lamps (top) consist oflargely unaltered slabs of rock. When the lamp is lit, melted fat runs oÝ throughnatural crevices in the rock. Closed-circuit lamps (middle) have carved depressionsto contain the runoÝ. Carved-handle, closed-circuit lamps (bottom) also have bowl-shaped fuel chambers but are more Þnely Þnished and have formed extensions foreasier handling. Burn marks indicate that the wick was placed away from the handle.

OPEN CIRCUIT

CLOSED CIRCUIT

CARVED HANDLE


distribution in time and space of carved-handle lamps may imply that theyserved primarily ceremonial purposes.A well-known example from Lascaux,which has been dated to 17,500 yearsago, was found on the cave ßoor at thebottom of a vertical shaft, below a draw-ing of a hunter confronting a woundedbison. This lamp was discovered by theAbb� Glory, a Catholic lay priest whosuggested that such lamps were usedto burn aromatic twigs and hence wereanalogous to incense burners. Too fewchemical analyses have been performed,however, to test this hypothesis ade-quately. The other kinds of stone lampsprobably served exclusively as sourcesof light.

To be eÝective, a fat-burning lampmust be reliable, easy to handle andbright enough to throw usable light adistance of a few meters in, for exam-ple, a darkened cave. The form of lampthat predominates in our sample of Pa-leolithic lamps is precisely that whichour experiments revealed to be optimal-ly eÛcient. It is a closed-circuit lamphaving an oval or circular depressionand gently sloping rather than verticalsides. Sloping the side of the bowl facil-itates emptying the lamp (so that thewick does not become swamped inmelted fat) without dislodging the wick.Carving a gap or notch in the rim ofthe lamp oÝers an alternative way toempty the bowl while keeping the wick

in place. Eighty percent of the Paleolith-ic lamps we studied use the sloped-side approach.

Anthropologists have long assumedthat animal fat was the fuelburned in Ice Age lamps. From

our experiments, we learned that thebest fats are those that melt quicklyand at a low temperature. Also, theymust not contain too much adipose tis-sue, the connective tissue in fat. Fatfrom seals, horses and bovids provedmost eÝective in experimental lamps.But were these in fact the fuels favoredby Paleolithic humans?

Guy L. Bourgeois of the University ofBordeaux and de Beaune analyzed resi-dues from several Paleolithic lamps toidentify the substances they contained.Using two sensitive chemical analysistechniques (vapor-phase chromatogra-phy and mass spectrometry), they mea-sured the carbon isotope ratios in fattyacids in the residues. The abundanceratios resemble those in animal fatsfrom modern herbivores, such as cat-tle, pigs and horses. Unfortunately, sci-entists have no samples of fat from theactual animals that lived during thelate Pleistocene. Nevertheless, the ob-served ratios of carbon isotopes arequite unlike those in vegetable fats, prov-ing that animals were indeed the sourceof fuel for Ice Age lamps.

Our investigations also provided new

information about the materials fromwhich wicks were made. A good wickmust be able to attract melted fat bycapillary action and convey it to the free,burning end without being too quick-ly consumed. Of the wicks we tested,lichen (known to be used by modernInuit), moss and then juniper workedbest. Fritz H. Schweingr�ber of theSwiss Federal Research Institute forForest, Snow and Landscape analyzedseveral lamp residues. He detectedremnants of conifers, juniper and grass,as well as nonwoody residues, possiblylichen or moss. In our experience, ju-niper wicks are never completely con-sumed by the ßame and so may be bet-ter preserved than wicks composed ofother plants.

The traces of use on our experimen-tal lamps make it possible to interpretwith conÞdence the markings observedon Paleolithic lamps. Those signs of us-age come in three broad forms: lightaccumulations of soot, deposits of char-coal and reddening of the rock itself, aprocess known as rubefaction. In 80percent of all the lamps observed, sootand charcoal deposits are situated with-in or on the rim of the fuel chamber,where one would expect the wick to lie.Occasional blackening of the side orunderside of the lamp can be producedby trickles of melted fat that carriedwith them small particles of soot. Char-coal deposits result from carbonization

SCIENTIFIC AMERICAN March 1993 111

EXPERIMENTAL CLOSED-CIRCUIT LAMP (left ) clariÞes howthese objects were used in Paleolithic times. A lump of fatserves as the fuel; the wick consists of bits of bark, lichen ormoss. Melted fat collects in a depression in the rock and

must occasionally be poured oÝ. Chemical analysis of IceAge lamps reveals the presence of residues whose composi-tion resembles that of fat from animals that were common inPaleolithic France (right ); vegetable fats clearly were not used.

CARBON ISOTOPE RATIO40 1 2 3 5 6 7 8 9

PALM OIL

WALNUT OIL

HOMO SAPIENS

OLIVE OIL

HORSE

DOG

PIG

LION

HIPPOPOTAMUS

OX

SHEEP

CAMEL

DEER

RANGE OFCOMPOSITIONOF 14MEASURED LAMPS


of the wick or from the heat alteration,or calcination, of adipose tissue in theburning fat.

Thermal reddening often appears onthe sides and undersides of lamps, butit, too, most frequently appears in oron the rim of the fuel chamber (in 67.5percent of the cases). Experience withmodern replicas indicates that suchreddening took place when hot, meltedfat ran onto the side or bottom of thelamp, either as the lamp was being emp-tied or when it overßowed on its own.Thermal reddening evidently can occurafter only a few uses and so provides ahelpful indicator of which artifactsserved as lamps.

Repeated reuse of a lamp leaves dis-tinct patterns. If a standard open- orclosed-circuit lamp is lit on several oc-casions, the placement of the fat andwick tends to change from one time tothe next. Because there is no preferredorientation for those simple lamps,they eventually become blackened andreddened over the entire bowl or sur-face. The carefully worked closed-cir-cuit lamps that have handles displaystrikingly diÝerent signs of usage. Theyare oriented the same way each timethey are lit, so soot deposits build up

on one part of the bowl only, generallythe area opposite the handle.

Open-circuit and simple closed-circuitlamps probably were lit only a few timesbefore being discarded. They are so easyto manufacture that there would haveexisted little incentive to carry themfrom site to site; we found that we couldmake a decent lamp in about half anhour. Decorated, carved-handle lamps,which represent a greater investmentof labor, were more likely to have beenused repeatedly.

To evaluate the eÝectiveness of Pa-leolithic fat-burning lamps, oneneeds to know how much light

those lamps could provide. De Beauneinvestigated this matter by measuringthe light output of modern replicas inthe metrology laboratories of Kodak-Path�, France. In quantity, intensity andluminescence, the experimental lampsprovided distinctly less light than astandard candle but nonetheless wouldhave been suÛcient to guide a personthrough a cave or to illuminate Þnework when placed nearbyÑassuming,of course, that the visual acuity of Pale-olithic people was the same as ours.

The limitations of Ice Age lamps sug-

gest that the creators of cave drawingsnever saw them as they appear in mod-ern photographs. Human color percep-tion is constrained and distorted at lev-els less than 150 lux (for comparison,1,000 lux is typical in a well-lit oÛce).It seems doubtful that the creators ofthe cave art worked under such brightconditions. Achieving full and accuratecolor perception of the cave imagesalong a Þve-meter-long panel would re-quire 150 lamps, each of them placed50 centimeters from the cave wall.Torches could have provided supple-mentary light, but few traces of torch-es have been found in deep caves. Onthe other hand, the absence or scarcityof lamps in vast cave galleries such asthose at RouÛgnac, Niaux and Les TroisFr�res implies that the creator of thepaintings had access to some alterna-tive light sources.

Today when one views the famouscave art in France and Spain, the arti-Þcial illumination creates an eÝect fun-damentally unlike that experienced byPaleolithic visitors. Electric lights in thecave of Font de Gaume yield a steadylight level of about 20 to 40 lux acrossa full panel of drawings. Ten to 15thoughtfully placed stone lamps wouldbe needed to attain 20 lux. A personcarrying a single lamp would get a verydiÝerent impression of the cave art andcould view only small portions of thewall at a time. The dim illumination pro-duced by ßickering lamps may well havebeen part of the desired eÝect of view-ing art deep within a cave. The illusion


ICE AGE LAMPS have been found primarily in southwestFrance (left ). Lamps appear in all eras of the Upper Paleolithic(40,000 to 11,000 years ago); more of them have been recov-

ered from the later periods. Surprisingly, most lamps havebeen retrieved not from deep caves but from open-air sitesand from under rock shelters (right ).

EARLY UPPER PALEOLITHIC (40,000–22,000 B.P.) SOLUTREAN OR MAGDALENIAN(22,000–11,000 B.P.)

PARIS

F R A NC

E

51%ROCK SHELTERS

29.5%DEEP CAVES

12%CAVE ENTRIES

7.5%OPEN AIR

LOWER MAGDALENIAN (18,000–15,000 B.P.)MIDDLE AND UPPER MAGDALENIAN(15,000–11,000 B.P.)


of animals suddenly materializing outof the darkness is a powerful one, andsome cave images are all the more con-vincing if one cannot see them too well.

Of course, fat-burning lamps wereemployed for many tasks other thancreating and viewing cave art. Lampsare found in such abundance at sitesthroughout southwest France that theymust have been a fairly ordinary itemof day-to-day existence. Only about 30percent of the known lamps were re-covered in deep caves. Open-air sites,rock shelters exposed to plentiful day-light and cave entries have provided therest. The number of lamps at each site(two to three, on average) does not dif-fer signiÞcantly from caves to rockshelters to open-air sites.

The location of lamps within sitesprovides clues to how people exploit-ed them. In deep caves, lamps are of-ten recovered from places where peo-ple had to pass, such as cave entrances,the intersections of diÝerent galleriesand along walls. It would seem thatlamps were placed at strategic or pre-dictable points where they could easilybe found and reused. The discovery ofmany lamps lying togetherÑmost no-tably at Lascaux, where 70 lamps havebeen recoveredÑimplies that lampswere stored in particular locations be-tween uses. Unfortunately, one cannotdeduce how many of the lamps were litat any one time.

Lamps are frequently discovered nearÞreplaces. Perhaps they were preheat-ed in the Þre in order to warm the fatand make it easier to ignite or wereabandoned and reused as hearthstones.More likely, Þreplaces served as centralpoints of heat and light from which peo-ple departed into and returned from thedarkness. Many lamps are found invert-ed in the soil, implying that on return-ing, people extinguished them simplyby turning them over.

In at least one location, a lamp seemsto have provided a permanent, Þxedsource of light within a campsite. Ar-chaeologists found two lamps in asmall, natural cavity in the wall of therock shelter of La Garenne. One lamphad been turned over as if to extin-guish the ßame. The other was placedupright in a natural hollow in the rockthat held it level. The cavity itself wouldhave served as a natural reßector thatmaximized the lampÕs light output.

Sorting through the sample of fat-burning lamps, we sought to learn howtheir abundance and design changedover time. That analysis is somewhatrestricted by the paucity of data. Ac-curate radioactive dates are availablefor only the most recently discoveredlamps. In most cases, ages are inferred

from the archaeological levels in whichthe lamps were found, and in manyearly excavations even that informationwas not recorded. Nevertheless, enoughinformation exists for us to make somegeneral observations.

Many more lamps appear in the lastcultural period of the Upper Paleolithic,the Magdalenian, than in preceding pe-riods. This may reßect the fact thatthere are simply more Magdalenian sitesknown than is the case for earlier peri-ods, as well as the fact that most deep-cave painting took place in the Magda-lenian. Older lamps are also harder toidentify with certainty.

The form of lamps seems to haveevolved surprisingly little throughthe ages. Some variation in form,

material and design occurred, but thereis no clear progression from crude toelaborate. Although carved-handle lampsare more common in the later eras, allthree primary types of lamp are foundthroughout the Magdalenian, and eventhe most elaborate lamp designs dateback to the earliest Upper Paleolithicperiods, which roughly corresponds tothe time when Cro-Magnon, anatomical-ly modern humans, appeared in Europe.The various forms of lamp most likelyrepresent functional responses to par-ticular contexts of use; the need for bothsimple, easy-to-make lamps and carved,aesthetically pleasing ones apparentlywas common to all Paleolithic culturesin France.

It is diÛcult to overstate the impor-

tance of artiÞcial light in freeing hu-mans from their evolutionary adap-tation to the daylight world. Cave artspecialist Denis Vialou of the Museumof Natural History in Paris lauds theMagdalenian cave artists as the peoplewho conquered the world of the under-ground. But perhaps it is more accurateto see them as the most daring of a longline of our Cro-Magnon ancestors, who,through intelligence and technologicalinnovation, changed the human experi-ence forever by domesticating the realmof darkness.

SCIENTIFIC AMERICAN March 1993 113

FURTHER READINGDARK CAVES, BRIGHT VISIONS: LIFE IN ICEAGE EUROPE. Randall White. AmericanMuseum of Natural History and W. W.Norton & Company, 1986.

PALEOLITHIC LAMPS AND THEIR SPECIAL-IZATION: A HYPOTHESIS. S. de Beaune inCurrent Anthropology, Vol. 28, No. 4,pages 569Ð577; August/October 1987.

TECHNOLOGICAL CHANGES ACROSS THEMIDDLE-UPPER PALEOLITHIC TRANSI-TION: ECONOMIC, SOCIAL AND COGNI-TIVE PERSPECTIVES. P. Mellars in TheHuman Revolution: Behavioural and Biological Perspectives on the Origins of Modern Humans. Edited by P. Mellarsand C. Stringer. Princeton UniversityPress, 1989.

NONFLINT STONE TOOLS OF THE EARLYUPPER PALEOLITHIC. Sophie de Beaunein Before Lascaux: The Complex Recordof the Early Upper Paleolithic. Edited byH. Knecht, A. Pike-Tay and R. White.CRC Press, 1993.

CAVE ART must have required artiÞcial illumination both to create it and to viewit. The dim, ßickering light provided by fat-burning lamps may have been integralto the intended appearance of these subterranean paintings.


ice age lamps

Documents