1

Rapa Nui JournalXIV, #3, 85-88September 2000

An Archeoastronomical Investigation: Does a ConstellationPattern Appear in Rapanui Rock Art?

Thomas Hockey, Professor of AstronomyUniversity of Northern Iowa

Alice Hoffman, PEHoffman Management Partners, LLC

Introduction

Krupp (1997) broadly defines the interdisciplinary field ofarcheoastronomy as embracing “calendrics; practicalobservation; sky lore and celestial myth; symbolicrepresentation of celestial objects, concepts, and events;astronomical orientation of tombs, temples, shrines, andurban centers;” and other, similar trappings of culture.While some of these aspects might better be classified as“ethnoastronomy,” it remains a useful definition.

The archeoastronomy of Rapa Nui largely has dealt with the“astronomical orientations” (= alignments) of ahu (i. e.,the seminal work summarized by Liller, 1989). This is inkeeping with the modern tradition of archeoastronomy thatbegan with analyses of megalithic stone circles in WesternEurope (e. g., Stonehenge). (See Ruggles, 1999 for adetailed history of archeoastronomy.) Here, we wish todiscuss the possible “symbolic representation of celestialobjects” in rock art, a task pioneered in the AmericanSouthwest by scholars such as Von del Chamberlain, SteveMcCluskey, and Ray Williamson.

There exists ethnographic evidence for Rapanui depictionsof the sky, the first recorded being that of Routledge(1919). Rapanui interest in Matariki, which refersspecifically to the Pleiades but might more generallyindicate constellations of stars, has been noted by dozensof investigators from the nineteenth century onward (e. g.,Van Tilberg, 1994). In addition, Lee (1992) and Liller(1989) have found petroglyphs that clearly look like acomet.

2

To our knowledge, no one has pointed out an unambiguousconstellation, or significant portion of a constellation,in extant Rapanui rock art. While there are allusions tothe Pleiades, no pictogram, petroglyph, or set of stoneshas been identified as being the Pleiades. Even so, Lee,in her comprehensive documentation of Rapanui rock art(1992), has encountered contemporary informants who attachastronomical significance to certain patterns within theubiquitous cupules (circular depressions pecked andabraided on rock throughout the island). It is one suchsite that we investigate in this paper.

Field Work

The “matariki stone” (from here used without quotationmarks) examined in this work is a multi-faceted, fine-grained basalt boulder, approximately 1 m X 1.5 m X 2 m insize. (Its weight is estimated to exceed 10,000 kg.) Theboulder has sustained a large crack.

At least six cupules have been placed in the boulder,averaging 6 cm in diameter and 5 cm in depth. The crackdestroys the spatial relationship between cupules on eitherside. Those on the east side were used in this analysis.

3

Figure 1. Portion of the Matariki Stone. The image hasbeen rotated 90O. (Courtesy of Georgia Lee.)

4

The boulder is now sitting in the yard of a privateresidence in the Mataveri section of Hanga Roa, west of theHotel Iorana. Information about its original setting,orientation, and context (provided by other nearbyartifacts) is lost. Likewise, it is not possible to useeven the inexact dating techniques sometimes applied topetroglyphs. (The current residents of the property canprovide no historical information about the boulder, otherthan that it has attracted the attention of modernvisitors.)

We are left, then, with but the pattern itself. Does itbear a likeness to any star patterns visible from Rapa Nui?A difficulty encountered with celestial-pattern recognitionis that any three stars form a triangle. That is, there islittle to distinguish stars from each other besidesposition in the sky. (Stars are featureless points on theCelestial Sphere; there is nothing to suggest that stellarmagnitude or color was meant to be portrayed inpetroglyphs.) Still, a data set of n=6 is great enough toprovide, at least, the hope of pattern identification, ifthere is indeed a naturalistic, representationalarrangement, and if it is indeed celestial.

We do not mean to suggest that the set of cupules might bea map of stars: Pre-contact Rapanui technology does notprovide the precise angular measurement required. Instead,our search is for an artistic interpretation, with all thepotential utilities (ceremony, aesthetics, commemoration,etc.) motivating human artistic expression.

Analysis

Photographs of the matariki stone, taken in situ, weredigitized. The present orientation of the stone leads towhat is presumably a distortion of the intended patternfrom any available viewing angle. Because the intendedviewing angle is unknown, we did not attempt any three-dimensional rotation of the digitized image. Doing sowould introduce several more free variables and increasethe likelihood of a false-positive pattern recognition.

The centers of the cupules were tagged, based on thelocations of cupule rims and shadows. The resultingdiagram was compared to a computer simulation of the nightsky as seen from the latitude of Rapa Nui (Lane, 1996).

5

Figure 2. Same as Figure 1, with Cupules Numbered.

6

The night-adapted naked eye, under dark, moonless skies atsea level, may discern stars as faint as magnitude 6. (Inthe astronomical apparent-magnitude scale, increasingbrightest corresponds to smaller numbers.) However, theseconditions rarely exist on Rapa Nui, even in pre-historictimes. It seems unlikely that magnitude 5 and 6 starswould have been rendered, without especially compellingreasons for doing so.

One documented exception to this is aboriginal peoplesdepicting the Pleiades, a group of faint-but-visible starscovering an usually small angular extent in the sky. Yetthe arrangement of cupules on the matariki stone inquestion does not resemble the Pleiades.

The simulated sky was surveyed using limiting magnitudes of1, 2, 3, and 4. Western constellation boundaries and“stick-figure” connecting lines were eliminated from thesimulation so as not to introduce a bias: The hypothesizedmatariki-stone stars might not conform to any modernconstellation. Rather, they might combine parts ofmultiple modern constellations.

Natural groupings of stars were searched for, under angularresolutions that portrayed (one at a time) five, minimallyoverlapping sections of the sky on a computer screen.These individual views may be referred to as the “North,”“South,” “East,” West,” and “Zenith” skies. We know of nopattern of stars in art, either ancient or modern, thatattempts to link stars spread over large areas of theentire sky.

The magnitude 1 simulations were not very interesting;there are few stars in the sky of that magnitude. Evenacknowledging the possible passage of a bright planet intoa star field along the ecliptic, nothing resembling thematariki-stone configuration was observed.

The magnitude 4 simulations were not useful for anotherreason: At that magnitude limit, there are simply too manystars. Identifying one particular pattern becomeshopeless.

Thus, we were left with magnitude 2 and 3 simulations. Twocandidate star patterns were identified. Both wererecognizable in magnitude 3 simulations. Note that there

7

are approximately 114 stars on the Celestial Sphere,magnitude 3 or greater, visible from Rapa Nui.

The first candidate includes stars from the modernconstellations Crux and Centaurus, which transit ataltitude 60O in the Rapanui sky. One star in Crux—ofcomparable brightness to the others and necessary to formthe familiar “cross” pattern—is missing from the artwork.Nevertheless, the present orientation of the boulder admitsthe possibility that the “missing” star is underneath.This candidate was exciting because of the stars’ well-known use in identifying the South Celestial Pole (forpurposes of celestial navigation).

After further consideration, the first candidate wasdiscarded. The second candidate is more robust: It moreclosely matches the matariki pattern. It is a smallergrouping (15O in angular extent upon the sky). And itconsists of one of the most distinctive patterns of brightstars in the South Celestial Hemisphere: the “teapot”asterism of Sagittarius.

8

Figure 3. Same as Figure 1, with a Map of SagittariusSuperimposed.

9

All but one of the remaining stars of Sagittarius arefainter than magnitude 3. The spatial relation of the oneexception (a magnitude 2.9 star, 7O away) is such that itconceivably could be represented by a cupule on the otherside of the crack. (The next nearest bright stars in thesky are in the constellation Scorpius.)

Sagittarius does not present any horizon-proximityproblems. From Rapa Nui, the constellation happens to passthrough the zenith—that is, it can appear directlyoverhead. There are long intervals of time when it is farfrom the horizon.

Post hoc reasoning yields the following, other uniqueattributes of Sagittarius: Sagittarius culminates on thezenith at dusk on the vernal equinox (and, therefore, dawnon the autumnal equinox). Similarly, Sagittarius sets asthe Sun rises on the winter solstice (and, therefore, risesas the Sun rises at the summer solstice). (Astronomicalprecession plays a small role over the time span of Rapanuicivilization; still, a precessional correction forapproximately 1500 CE improves the coincidental timing ofthese events over that of modern times.)

Furthermore, Sagittarius is a zodiac constellation; brightplanets may appear in it from time to time. Sagittariusalso is in the direction of a particularly bright part ofthe Milky Way.*

Even without the view-point distortion, the relativepositions of the matariki stone cupules are almostcertainly not a one-to-one match with the proper angularpositions for the Sagittarius stars. As stated before,there is no reason to assume that positional verisimilitudewas an objective in producing this rock art. The accuracyis, though, similar to that displayed in some published,western star charts of the Middle Ages (contemporary withthe Rapanui). (See North, 1995 for examples.) Moreover,old-world uranographers had much easier media, drawingtools, and “studio environment” with which to work.

* A mystic once asked an author [TH] to compute the location from whichhe could stare in the direction of the Galactic center (while facingthe zenith) on the instant of the summer solstice. The answer lay inthe ocean. However, the nearest inhabited land was Rapa Nui.

10

Figure 4. From an Eleventh-century Star Map by Al-Sufi.The relevant stars have been marked with light dots. (Fromthe Bibliotecque Nationale de France athttp://www.bnf.fr/web-bnf/expos/ciel/grand/2-044.htm)

11

Identification of the matariki pattern ultimately rests onthe authenticity of certain cupules and their concurrentmanufacture. Cupule #5 differs in morphology from theothers. While it is of the same diameter, it is not asdeep. It appears incomplete. If cupule #5 is natural tothe boulder, or an earlier/later addition, theidentification is compromised.

Is the matariki-stone pattern a coincidence? Consider acircle of radius R (part of the rock face incorporating thepattern) completely within which are six, identical, non-overlapping circles of radius r. (These smaller circlesrepresent the Sagittarius star positions to a given levelof precision.)

(The real boulder was not so much a tabla Rasa. Itssurface topography surely influenced the finished rock art.This is a sound first-order approximation, though.)

What is the probability, P, of six cupules being placedrandomly such that one (and only one) falls into each ofthe smaller circles? The area of a circle is proportionalto radius squared. If ρ is the ratio of the diameter of thelarge circle to that of the smaller circles, then

1/P = (ρ2)(ρ2-1)(ρ2-2)(ρ2-3)(ρ2-4)(ρ2-5)

The size of the cupules and their proximity to each othersuggest that ρ=10 is a reasonable assumption. Thus,

P = 1.165 X 10-12

Obviously, these statistics are for uncorrelated datapoints. We assume the cupules are correlated. Even if nopattern was intended, the fact that cupules are of finitesize yet not superimposed introduces a correlation. Still,this exercise demonstrates the significance of the numberof points involved.

Conclusion

The calendrical importance of zenith transits toinhabitants of low latitudes is well documented. (See, forinstance, Aveni, 1997.) Particular concern with the wintersolstice is worldwide. Liller (1989) makes a convincing

12

argument for architectural alignments, on Rapa Nui, markingthe solstices or equinoxes.

Attention paid to the regions of the Milky Way is alsouniversal, but it figures most prominently in the sky loreof South and Central America (Krupp, 1991). By suggestingthat the Rapanui watched Sagittarius and the Milky Way, wedo not mean to imply any cultural connection between nativeAmerican peoples and those of Polynesia. (The Milky Way isreally more spectacular the farther south one observes it.)We propose, alternately, similar (but independent)astronomical awareness, demonstrated by civilizations atthe same latitude.

Of course, the association of our matariki-stone patternwith astronomical events must remain a hypothesis, becauseof the standard vagaries involved in rock-artinterpretation, and because of the ad hoc nature of thehypothesis itself. Nonetheless, having entered into thisexercise with pessimism about making any linkage betweenthe matariki stone and any real star pattern, we find thishypothesized correlation heartening.

Future work should include 1) the discovery andidentification of other star patterns in Rapanui rock art.While a single example by a single artist cannot bedisproved, a statistical sample of hypothesized star-patterned rock art would help the analysis considerably.

2) The matariki stone under present consideration must berotated so that the hypothesized figure can be viewed “faceon” (as it was likely created and as the artist likelyintended). This might be simulated with multi-axis image-rotating software. Otherwise, it might be donephysically.**

3) Pattern-recognition software should be brought to bear,in order to eliminate human bias in identifying patterns,both celestial and in rock art. Such software exists, butit is tailored toward practical duplication (for example,

** An author [AH] attempted to rotate the boulder—with permission!—whileavoiding damage to its surface. Six adult Iowans, using a six-footsteel lever, changed the boulder’s placement by 1-2 centimeters. Thisaction was sufficient to allow insertion of two small wedges, therebystabilizing the new orientation. It provided an only slightly betterviewing angle.

13

fingerprints, handwriting, or counterfeit bills) and not tothe needs of archeologists or astronomers.

4) A computer model should be constructed to morerealistically ascertain the probability of random cupuleplacement resulting in “a Sagittarius,” within some levelof precision.

Acknowledgements

This project was undertaken at the suggestion of GeorgiaLee and benefited from her encouragement.

The authors would like to thank all of the participants inthe 2000 University of Northern Iowa Geological Field Trip,led by Kenneth DeNault (Department of Earth Science). Eachprovided support or advice to our project.

We also are grateful for the guidance of Ramón Edmunds,while on Rapa Nui. Upon our return, Siobahn Morgan helpedwith the computer analysis. The UNI Statistical ConsultingCenter also is acknowledged.

References

Aveni, A. Stairway to the Stars: Skywatching in ThreeGreat Ancient Cultures. New York: John Wiley and Sons.1997.

Krupp, E. Beyond the Blue Horizon: Myths and Legends ofthe Sun, Moon, Stars, and Planets. New York: HarperCollins. 1991.

Krupp, E. “Archeoastronomy” in History of Astronomy: AnEncyclopedia, edited by J. Lankford. New York: GarlandPress. 1997.

Lane, D. Earth Centered Universe V3.0A. Halifax: NovaAstronomics. 1996.

Lee, G. Easter Island Rock Art: Symbols of Power, Prayersto the Gods. Los Angeles: UCLA Institute of Archeology.1992.

14

Liller, W. “The Archeoastronomy of Easter Island.”Journal for the History of Astronomy (ArcheoastronomySupplement). XIII, S21, 1989.

North, J. Astronomy and Cosmology. New York: W. W.Norton and Company. 1995.

Routledge, K. The Mystery of Easter Island. London:Hazell, Watson, and Viney. 1919.

Ruggles, C. The Astronomy of Prehistoric Britain andIreland. New Haven: Yale University Press. 1999.

Van Tilburg, J. Easter Island: Archeology, Ecology, andCulture. Washington: Smithsonian Institution Press. 1994.